Version Control with Subversion

For Subversion 1.7

(Compiled from r4881)

Ben Collins-Sussman

Brian W. Fitzpatrick

C. Michael Pilato

This work is licensed under the Creative Commons Attribution License. To view a copy of this license, visit http://creativecommons.org/licenses/by/2.0/ or send a letter to Creative Commons, 559 Nathan Abbott Way, Stanford, California 94305, USA.


Table of Contents

Foreword
Preface
What Is Subversion?
Is Subversion the Right Tool?
Subversion's History
Subversion's Architecture
Subversion's Components
What's New in Subversion
Audience
How to Read This Book
Organization of This Book
This Book Is Free
Acknowledgments
1. Fundamental Concepts
Version Control Basics
The Repository
The Working Copy
Versioning Models
The problem of file sharing
The lock-modify-unlock solution
The copy-modify-merge solution
Version Control the Subversion Way
Subversion Repositories
Revisions
Addressing the Repository
Subversion Working Copies
How the working copy works
Fundamental working copy interactions
Mixed-revision working copies
Summary
2. Basic Usage
Help!
Getting Data into Your Repository
Importing Files and Directories
Recommended Repository Layout
What's In a Name?
Creating a Working Copy
Basic Work Cycle
Update Your Working Copy
Make Your Changes
Review Your Changes
See an overview of your changes
Examine the details of your local modifications
Fix Your Mistakes
Resolve Any Conflicts
Viewing conflict differences interactively
Resolving conflict differences interactively
Postponing conflict resolution
Merging conflicts by hand
Discarding your changes in favor of a newly fetched revision
Punting: using svn revert
Commit Your Changes
Examining History
Examining the Details of Historical Changes
Examining local changes
Comparing working copy to repository
Comparing repository revisions
Generating a List of Historical Changes
Browsing the Repository
Displaying file contents
Displaying line-by-line change attribution
Listing versioned directories
Fetching Older Repository Snapshots
Sometimes You Just Need to Clean Up
Disposing of a Working Copy
Recovering from an Interruption
Dealing with Structural Conflicts
An Example Tree Conflict
Summary
3. Advanced Topics
Revision Specifiers
Revision Keywords
Revision Dates
Peg and Operative Revisions
Properties
Why Properties?
Manipulating Properties
Properties and the Subversion Workflow
Automatic Property Setting
File Portability
File Content Type
File Executability
End-of-Line Character Sequences
Ignoring Unversioned Items
Keyword Substitution
Sparse Directories
Locking
Creating Locks
Discovering Locks
Breaking and Stealing Locks
Lock Communication
Externals Definitions
Changelists
Creating and Modifying Changelists
Changelists As Operation Filters
Changelist Limitations
Network Model
Requests and Responses
Client Credentials
Caching credentials
Disabling password caching
Removing cached credentials
Command-line authentication
Authentication wrap-up
Summary
4. Branching and Merging
What's a Branch?
Using Branches
Creating a Branch
Working with Your Branch
The Key Concepts Behind Branching
Basic Merging
Changesets
Keeping a Branch in Sync
Reintegrating a Branch
Mergeinfo and Previews
Undoing Changes
Resurrecting Deleted Items
Advanced Merging
Cherrypicking
Merge Syntax: Full Disclosure
Merges Without Mergeinfo
More on Merge Conflicts
Blocking Changes
Keeping a Reintegrated Branch Alive
Merge-Sensitive Logs and Annotations
Noticing or Ignoring Ancestry
Merges and Moves
Blocking Merge-Unaware Clients
The Final Word on Merge Tracking
Traversing Branches
Tags
Creating a Simple Tag
Creating a Complex Tag
Branch Maintenance
Repository Layout
Data Lifetimes
Common Branching Patterns
Release Branches
Feature Branches
Vendor Branches
General Vendor Branch Management Procedure
svn_load_dirs.pl
To Branch or Not to Branch?
Summary
5. Repository Administration
The Subversion Repository, Defined
Strategies for Repository Deployment
Planning Your Repository Organization
Deciding Where and How to Host Your Repository
Choosing a Data Store
Berkeley DB
FSFS
Creating and Configuring Your Repository
Creating the Repository
Implementing Repository Hooks
Berkeley DB Configuration
FSFS Configuration
Repository Maintenance
An Administrator's Toolkit
svnadmin
svnlook
svndumpfilter
svnrdump
svnsync
fsfs-reshard.py
Berkeley DB utilities
Commit Log Message Correction
Managing Disk Space
How Subversion saves disk space
Removing dead transactions
Purging unused Berkeley DB logfiles
Packing FSFS filesystems
Berkeley DB Recovery
Migrating Repository Data Elsewhere
Repository data migration using svnadmin
Repository data migration using svnrdump
Filtering Repository History
Repository Replication
Replication with svnsync
Partial replication with svnsync
A quick trick for mirror creation
Replication wrap-up
Repository Backup
Managing Repository UUIDs
Moving and Removing Repositories
Summary
6. Server Configuration
Overview
Choosing a Server Configuration
The svnserve Server
svnserve over SSH
The Apache HTTP Server
Recommendations
svnserve, a Custom Server
Invoking the Server
svnserve as daemon
svnserve via inetd
svnserve over a tunnel
svnserve as a Windows service
svnserve as a launchd job
Built-in Authentication and Authorization
Create a users file and realm
Set access controls
Using svnserve with SASL
Authenticating with SASL
SASL encryption
Tunneling over SSH
SSH Configuration Tricks
Initial setup
Controlling the invoked command
httpd, the Apache HTTP Server
Prerequisites
Basic Apache Configuration
Authentication Options
Basic authentication
Digest authentication
Authorization Options
Blanket access control
Per-directory access control
Disabling path-based checks
Protecting network traffic with SSL
Subversion server SSL certificate configuration
Subversion client SSL certificate management
Extra Goodies
Repository browsing
Apache logging
Write-through proxying
Other Apache features
Path-Based Authorization
High-level Logging
Server Optimization
Data Caching
Network Compression of Data
Supporting Multiple Repository Access Methods
7. Customizing Your Subversion Experience
Runtime Configuration Area
Configuration Area Layout
Configuration and the Windows Registry
Configuration Options
Servers
Config
Localization
Understanding Locales
Subversion's Use of Locales
Using External Editors
Using External Differencing and Merge Tools
External diff
External diff3
External merge
Summary
8. Embedding Subversion
Layered Library Design
Repository Layer
Repository Access Layer
Client Layer
Using the APIs
The Apache Portable Runtime Library
Functions and Batons
URL and Path Requirements
Using Languages Other Than C and C++
Code Samples
Summary
9. Subversion Complete Reference
svn—Subversion Command-Line Client
svn Options
svn Subcommands
svn add
svn blame (praise, annotate, ann)
svn cat
svn changelist (cl)
svn checkout (co)
svn cleanup
svn commit (ci)
svn copy (cp)
svn delete (del, remove, rm)
svn diff (di)
svn export
svn help (h, ?)
svn import
svn info
svn list (ls)
svn lock
svn log
svn merge
svn mergeinfo
svn mkdir
svn move (mv)
svn patch
svn propdel (pdel, pd)
svn propedit (pedit, pe)
svn propget (pget, pg)
svn proplist (plist, pl)
svn propset (pset, ps)
svn relocate
svn resolve
svn resolved
svn revert
svn status (stat, st)
svn switch (sw)
svn unlock
svn update (up)
svn upgrade
svnadmin—Subversion Repository Administration
svnadmin Options
svnadmin Subcommands
svnadmin crashtest
svnadmin create
svnadmin deltify
svnadmin dump
svnadmin help (h, ?)
svnadmin hotcopy
svnadmin list-dblogs
svnadmin list-unused-dblogs
svnadmin load
svnadmin lslocks
svnadmin lstxns
svnadmin pack
svnadmin recover
svnadmin rmlocks
svnadmin rmtxns
svnadmin setlog
svnadmin setrevprop
svnadmin setuuid
svnadmin upgrade
svnadmin verify
svnlook—Subversion Repository Examination
svnlook Options
svnlook Subcommands
svnlook author
svnlook cat
svnlook changed
svnlook date
svnlook diff
svnlook dirs-changed
svnlook filesize
svnlook help (h, ?)
svnlook history
svnlook info
svnlook lock
svnlook log
svnlook propget (pget, pg)
svnlook proplist (plist, pl)
svnlook tree
svnlook uuid
svnlook youngest
svnsync—Subversion Repository Mirroring
svnsync Options
svnsync Subcommands
svnsync copy-revprops
svnsync help
svnsync info
svnsync initialize (init)
svnsync synchronize (sync)
svnrdump—Remote Subversion Repository Data Migration
svnrdump Options
svnrdump Subcommands
svnrdump dump
svnrdump help
svnrdump load
svnserve—Custom Subversion Server
svnserve Options
svndumpfilter—Subversion History Filtering
svndumpfilter Options
svndumpfilter Subcommands
svndumpfilter exclude
svndumpfilter include
svndumpfilter help
svnversion—Subversion Working Copy Version Info
svnversion
mod_dav_svn—Subversion Apache HTTP Server Module
mod_dav_svn Configuration Directives
mod_authz_svn—Subversion Apache HTTP Authorization Module
mod_authz_svn Configuration Directives
Subversion Properties
Versioned Properties
Unversioned Properties
Repository Hooks
start-commit
pre-commit
post-commit
pre-revprop-change
post-revprop-change
pre-lock
post-lock
pre-unlock
post-unlock
A. Subversion Quick-Start Guide
Installing Subversion
High-Speed Tutorial
B. Subversion for CVS Users
Revision Numbers Are Different Now
Directory Versions
More Disconnected Operations
Distinction Between Status and Update
Status
Update
Branches and Tags
Metadata Properties
Conflict Resolution
Binary Files and Translation
Versioned Modules
Authentication
Converting a Repository from CVS to Subversion
C. WebDAV and Autoversioning
What Is WebDAV?
Autoversioning
Client Interoperability
Standalone WebDAV Applications
Microsoft Office, Dreamweaver, Photoshop
cadaver, DAV Explorer
File-Explorer WebDAV Extensions
Microsoft Web Folders
Nautilus, Konqueror
WebDAV Filesystem Implementation
WebDrive, NetDrive
Mac OS X
Linux davfs2
D. Copyright
Index

List of Figures

1. Subversion's architecture
1.1. A typical client/server system
1.2. The problem to avoid
1.3. The lock-modify-unlock solution
1.4. The copy-modify-merge solution
1.5. The copy-modify-merge solution (continued)
1.6. Tree changes over time
1.7. The repository's filesystem
4.1. Branches of development
4.2. Starting repository layout
4.3. Repository with new copy
4.4. The branching of one file's history
8.1. Files and directories in two dimensions
8.2. Versioning time—the third dimension!

List of Tables

1.1. Repository access URLs
2.1. Common log requests
4.1. Branching and merging commands
5.1. Repository data store comparison
6.1. Comparison of subversion server options
C.1. Common WebDAV clients

List of Examples

4.1. Merge-tracking gatekeeper start-commit hook script
5.1. txn-info.sh (reporting outstanding transactions)
5.2. Mirror repository's pre-revprop-change hook script
5.3. Mirror repository's start-commit hook script
6.1. A sample svnserve launchd job definition
6.2. A sample configuration for anonymous access
6.3. A sample configuration for authenticated access
6.4. A sample configuration for mixed authenticated/anonymous access
6.5. Disabling path checks altogether
7.1. Sample registration entries (.reg) file
7.2. diffwrap.py
7.3. diffwrap.bat
7.4. diff3wrap.py
7.5. diff3wrap.bat
7.6. mergewrap.py
7.7. mergewrap.bat
8.1. Using the repository layer
8.2. Using the repository layer with Python
8.3. A Python status crawler

Foreword

Karl Fogel

Chicago, March 14, 2004.

A bad Frequently Asked Questions (FAQ) sheet is one that is composed not of the questions people actually ask, but of the questions the FAQ's author wishes people would ask. Perhaps you've seen the type before:

Q: How can I use Glorbosoft XYZ to maximize team productivity?

A: Many of our customers want to know how they can maximize productivity through our patented office groupware innovations. The answer is simple. First, click on the File menu, scroll down to Increase Productivity, then…

The problem with such FAQs is that they are not, in a literal sense, FAQs at all. No one ever called the tech support line and asked, How can we maximize productivity? Rather, people asked highly specific questions, such as How can we change the calendaring system to send reminders two days in advance instead of one? and so on. But it's a lot easier to make up imaginary Frequently Asked Questions than it is to discover the real ones. Compiling a true FAQ sheet requires a sustained, organized effort: over the lifetime of the software, incoming questions must be tracked, responses monitored, and all gathered into a coherent, searchable whole that reflects the collective experience of users in the wild. It calls for the patient, observant attitude of a field naturalist. No grand hypothesizing, no visionary pronouncements here—open eyes and accurate note-taking are what's needed most.

What I love about this book is that it grew out of just such a process, and shows it on every page. It is the direct result of the authors' encounters with users. It began with Ben Collins-Sussman's observation that people were asking the same basic questions over and over on the Subversion mailing lists: what are the standard workflows to use with Subversion? Do branches and tags work the same way as in other version control systems? How can I find out who made a particular change?

Frustrated at seeing the same questions day after day, Ben worked intensely over a month in the summer of 2002 to write The Subversion Handbook, a 60-page manual that covered all the basics of using Subversion. The manual made no pretense of being complete, but it was distributed with Subversion and got users over that initial hump in the learning curve. When O'Reilly decided to publish a full-length Subversion book, the path of least resistance was obvious: just expand the Subversion handbook.

The three coauthors of the new book were thus presented with an unusual opportunity. Officially, their task was to write a book top-down, starting from a table of contents and an initial draft. But they also had access to a steady stream—indeed, an uncontrollable geyser—of bottom-up source material. Subversion was already in the hands of thousands of early adopters, and those users were giving tons of feedback, not only about Subversion, but also about its existing documentation.

During the entire time they wrote this book, Ben, Mike, and Brian haunted the Subversion mailing lists and chat rooms incessantly, carefully noting the problems users were having in real-life situations. Monitoring such feedback was part of their job descriptions at CollabNet anyway, and it gave them a huge advantage when they set out to document Subversion. The book they produced is grounded firmly in the bedrock of experience, not in the shifting sands of wishful thinking; it combines the best aspects of user manual and FAQ sheet. This duality might not be noticeable on a first reading. Taken in order, front to back, the book is simply a straightforward description of a piece of software. There's the overview, the obligatory guided tour, the chapter on administrative configuration, some advanced topics, and of course, a command reference and troubleshooting guide. Only when you come back to it later, seeking the solution to some specific problem, does its authenticity shine out: the telling details that can only result from encounters with the unexpected, the examples honed from genuine use cases, and most of all the sensitivity to the user's needs and the user's point of view.

Of course, no one can promise that this book will answer every question you have about Subversion. Sometimes the precision with which it anticipates your questions will seem eerily telepathic; yet occasionally, you will stumble into a hole in the community's knowledge and come away empty-handed. When this happens, the best thing you can do is email and present your problem. The authors are still there and still watching, and the authors include not just the three listed on the cover, but many others who contributed corrections and original material. From the community's point of view, solving your problem is merely a pleasant side effect of a much larger project—namely, slowly adjusting this book, and ultimately Subversion itself, to more closely match the way people actually use it. They are eager to hear from you, not only because they can help you, but because you can help them. With Subversion, as with all active free software projects, you are not alone.

Let this book be your first companion.

Preface

 

It is important not to let the perfect become the enemy of the good, even when you can agree on what perfect is. Doubly so when you can't. As unpleasant as it is to be trapped by past mistakes, you can't make any progress by being afraid of your own shadow during design.

 
 --Greg Hudson, Subversion developer

In the world of open source software, the Concurrent Versions System (CVS) was the tool of choice for version control for many years. And rightly so. CVS was open source software itself, and its nonrestrictive modus operandi and support for networked operation allowed dozens of geographically dispersed programmers to share their work. It fit the collaborative nature of the open source world very well. CVS and its semi-chaotic development model have since become cornerstones of open source culture.

But CVS was not without its flaws, and simply fixing those flaws promised to be an enormous effort. Enter Subversion. Subversion was designed to be a successor to CVS, and its originators set out to win the hearts of CVS users in two ways—by creating an open source system with a design (and look and feel) similar to CVS, and by attempting to avoid most of CVS's noticeable flaws. While the result wasn't—and isn't—the next great evolution in version control design, Subversion is very powerful, very usable, and very flexible.

This book is written to document the 1.7 series of the Apache Subversion™[1] version control system. We have made every attempt to be thorough in our coverage. However, Subversion has a thriving and energetic development community, so already a number of features and improvements are planned for future versions that may change some of the commands and specific notes in this book.

What Is Subversion?

Subversion is a free/open source version control system (VCS). That is, Subversion manages files and directories, and the changes made to them, over time. This allows you to recover older versions of your data or examine the history of how your data changed. In this regard, many people think of a version control system as a sort of time machine.

Subversion can operate across networks, which allows it to be used by people on different computers. At some level, the ability for various people to modify and manage the same set of data from their respective locations fosters collaboration. Progress can occur more quickly without a single conduit through which all modifications must occur. And because the work is versioned, you need not fear that quality is the trade-off for losing that conduit—if some incorrect change is made to the data, just undo that change.

Some version control systems are also software configuration management (SCM) systems. These systems are specifically tailored to manage trees of source code and have many features that are specific to software development—such as natively understanding programming languages, or supplying tools for building software. Subversion, however, is not one of these systems. It is a general system that can be used to manage any collection of files. For you, those files might be source code—for others, anything from grocery shopping lists to digital video mixdowns and beyond.

Is Subversion the Right Tool?

If you're a user or system administrator pondering the use of Subversion, the first question you should ask yourself is: "Is this the right tool for the job?" Subversion is a fantastic hammer, but be careful not to view every problem as a nail.

As a first step, you need to decide if version control in general is required for your purposes. If you need to archive old versions of files and directories, possibly resurrect them, and examine logs of how they've changed over time, then version control tools can do that. If you need to collaborate with people on documents (usually over a network) and keep track of who made which changes, a version control tool can do that, too. In fact, this is why version control tools such as Subversion are so often used in software development environments—working on a development team is an inherently social activity where changes to source code files are constantly being discussed, made, evaluated, and even sometimes unmade. Version control tools facilitate that sort of collaboration.

There is cost associated with using version control, too. Unless you can outsource the administration of your version control system to a third-party, you'll have the obvious costs of performing that administration yourself. When working with the data on a daily basis, you won't be able to copy, move, rename, or delete files the way you usually do. Instead, you'll have to do all of those things through the version control system.

Even assuming that you are okay with the cost/benefit tradeoff afforded by a version control system, you shouldn't choose to use one merely because it can do what you want. Consider whether your needs are better addressed by other tools. For example, because Subversion replicates data to all the collaborators involved, a common misuse is to treat it as a generic distribution system. People will sometimes use Subversion to distribute huge collections of photos, digital music, or software packages. The problem is that this sort of data usually isn't changing at all. The collection itself grows over time, but the individual files within the collection aren't being changed. In this case, using Subversion is overkill.[2] There are simpler tools that efficiently replicate data without the overhead of tracking changes, such as rsync or unison.

Once you've decided that you need a version control solution, you'll find no shortage of available options. When Subversion was first designed and released, the predominant methodology of version control was centralized version control—a single remote master storehouse of versioned data with individual users operating locally against shallow copies of that data's version history. Subversion quickly emerged after its initial introduction as the clear leader in this field of version control, earning widespread adoption and supplanting installations of many older version control systems. It continues to hold that prominent position today.

Much has changed since that time, though. In the years since the Subversion project began its life, a newer methodology of version control called distributed version control has likewise garnered widespread attention and adoption. Tools such as Git (http://git-scm.com/) and Mercurial (http://mercurial.selenic.com/) quickly rose to the tops of the distributed version control system (DVCS) ranks. Distributed version control harnesses the growing ubiquity of high-speed network connections and low storage costs to offer an approach which differs from the centralized model in key ways. First and most obvious is the fact that there is no remote, central storehouse of versioned data. Rather, each user keeps and operates against very deep—complete, in a sense—local version history data stores. Collaboration still occurs, but is accomplished by trading changesets (collections of changes made to versioned items) directly between users' local data stores, not via a centralized master data store. In fact, any semblance of a canonical master source of a project's versioned data is by convention only, a status attributed by the various collaborators on that project.

There are pros and cons to each version control approach. Perhaps the two biggest benefits delivered by the DVCS tools are incredible performance for day-to-day operations (because the primary data store is locally held) and vastly better support for merging between branches (because merge algorithms serve as the very core of how DVCSes work at all). The downside is that distributed version control is an inherently more complicated model, which can present a non-negligible challenge to comfortable collaboration. Also, DVCS tools do what they do well in part because of a certain degree of control withheld from the user which centalized systems freely offer—the ability to implement path-based access control, the flexibility to update or backdate individual versioned data items, etc. Fortunately, many wise organizations have discovered that this needn't be a religious debate, and that Subversion and a DVCS tool such as Git can be used together harmoniously within the organization, each serving the purposes best suited to the tool.

Alas, this book is about Subversion, so we'll not attempt a full comparison of Subversion and other tools. Readers who have the option of choosing their version control system are encouraged to research the available options and make the determination that works best for themselves and their fellow collaborators. And if, after doing so, Subversion is the chosen tool, there's plenty of detailed information about how to use it successfully in the chapters that follow!

Subversion's History

In early 2000, CollabNet, Inc. (http://www.collab.net) began seeking developers to write a replacement for CVS. CollabNet offered[3] a collaboration software suite called CollabNet Enterprise Edition (CEE), of which one component was version control. Although CEE used CVS as its initial version control system, CVS's limitations were obvious from the beginning, and CollabNet knew it would eventually have to find something better. Unfortunately, CVS had become the de facto standard in the open source world largely because there wasn't anything better, at least not under a free license. So CollabNet determined to write a new version control system from scratch, retaining the basic ideas of CVS, but without the bugs and misfeatures.

In February 2000, they contacted Karl Fogel, the author of Open Source Development with CVS (Coriolis, 1999), and asked if he'd like to work on this new project. Coincidentally, at the time Karl was already discussing a design for a new version control system with his friend Jim Blandy. In 1995, the two had started Cyclic Software, a company providing CVS support contracts, and although they later sold the business, they still used CVS every day at their jobs. Their frustration with CVS had led Jim to think carefully about better ways to manage versioned data, and he'd already come up with not only the Subversion name, but also the basic design of the Subversion data store. When CollabNet called, Karl immediately agreed to work on the project, and Jim got his employer, Red Hat Software, to essentially donate him to the project for an indefinite period of time. CollabNet hired Karl and Ben Collins-Sussman, and detailed design work began in May 2000. With the help of some well-placed prods from Brian Behlendorf and Jason Robbins of CollabNet, and from Greg Stein (at the time an independent developer active in the WebDAV/DeltaV specification process), Subversion quickly attracted a community of active developers. It turned out that many people had encountered the same frustrating experiences with CVS and welcomed the chance to finally do something about it.

The original design team settled on some simple goals. They didn't want to break new ground in version control methodology, they just wanted to fix CVS. They decided that Subversion would match CVS's features and preserve the same development model, but not duplicate CVS's most obvious flaws. And although it did not need to be a drop-in replacement for CVS, it should be similar enough that any CVS user could make the switch with little effort.

After 14 months of coding, Subversion became self-hosting on August 31, 2001. That is, Subversion developers stopped using CVS to manage Subversion's own source code and started using Subversion instead.

While CollabNet started the project, and still funds a large chunk of the work (it pays the salaries of a few full-time Subversion developers), Subversion is run like most open source projects, governed by a loose, transparent set of rules that encourage meritocracy. In 2009, CollabNet worked with the Subversion developers towards the goal of integrating the Subversion project into the Apache Software Foundation (ASF), one of the most well-known collectives of open source projects in the world. Subversion's technical roots, community priorities, and development practices were a perfect fit for the ASF, many of whose members were already active Subversion contributors. In early 2010, Subversion was fully adopted into the ASF's family of top-level projects, moved its project web presence to http://subversion.apache.org, and was rechristened Apache Subversion.

Subversion's Architecture

Figure 1, “Subversion's architecture” illustrates a mile-high view of Subversion's design.

Figure 1. Subversion's architecture

Subversion's architecture


On one end is a Subversion repository that holds all of your versioned data. On the other end is your Subversion client program, which manages local reflections of portions of that versioned data. Between these extremes are multiple routes through a Repository Access (RA) layer, some of which go across computer networks and through network servers which then access the repository, others of which bypass the network altogether and access the repository directly.

Subversion's Components

Subversion, once installed, has a number of different pieces. The following is a quick overview of what you get. Don't be alarmed if the brief descriptions leave you scratching your head—plenty more pages in this book are devoted to alleviating that confusion.

svn

The command-line client program

svnversion

A program for reporting the state (in terms of revisions of the items present) of a working copy

svnlook

A tool for directly inspecting a Subversion repository

svnadmin

A tool for creating, tweaking, or repairing a Subversion repository

mod_dav_svn

A plug-in module for the Apache HTTP Server, used to make your repository available to others over a network

svnserve

A custom standalone server program, runnable as a daemon process or invokable by SSH; another way to make your repository available to others over a network

svndumpfilter

A program for filtering Subversion repository dump streams

svnsync

A program for incrementally mirroring one repository to another over a network

svnrdump

A program for performing repository history dumps and loads over a network

What's New in Subversion

The first edition of this book was published by O'Reilly Media in 2004, shortly after Subversion had reached 1.0. Since that time, the Subversion project has continued to release new major releases of the software. Here's a quick summary of major new changes since Subversion 1.0. Note that this is not a complete list; for full details, please visit Subversion's web site at http://subversion.apache.org.

Subversion 1.1 (September 2004)

Release 1.1 introduced FSFS, a flat-file repository storage option for the repository. While the Berkeley DB backend is still widely used and supported, FSFS has since become the default choice for newly created repositories due to its low barrier to entry and minimal maintenance requirements. Also in this release came the ability to put symbolic links under version control, auto-escaping of URLs, and a localized user interface.

Subversion 1.2 (May 2005)

Release 1.2 introduced the ability to create server-side locks on files, thus serializing commit access to certain resources. While Subversion is still a fundamentally concurrent version control system, certain types of binary files (e.g. art assets) cannot be merged together. The locking feature fulfills the need to version and protect such resources. With locking also came a complete WebDAV auto-versioning implementation, allowing Subversion repositories to be mounted as network folders. Finally, Subversion 1.2 began using a new, faster binary-differencing algorithm to compress and retrieve old versions of files.

Subversion 1.3 (December 2005)

Release 1.3 brought path-based authorization controls to the svnserve server, matching a feature formerly found only in the Apache server. The Apache server, however, gained some new logging features of its own, and Subversion's API bindings to other languages also made great leaps forward.

Subversion 1.4 (September 2006)

Release 1.4 introduced a whole new tool—svnsync—for doing one-way repository replication over a network. Major parts of the working copy metadata were revamped to no longer use XML (resulting in client-side speed gains), while the Berkeley DB repository backend gained the ability to automatically recover itself after a server crash.

Subversion 1.5 (June 2008)

Release 1.5 took much longer to finish than prior releases, but the headliner feature was gigantic: semi-automated tracking of branching and merging. This was a huge boon for users, and pushed Subversion far beyond the abilities of CVS and into the ranks of commercial competitors such as Perforce and ClearCase. Subversion 1.5 also introduced a bevy of other user-focused features, such as interactive resolution of file conflicts, sparse checkouts, client-side management of changelists, powerful new syntax for externals definitions, and SASL authentication support for the svnserve server.

Subversion 1.6 (March 2009)

Release 1.6 continued to make branching and merging more robust by introducing tree conflicts, and offered improvements to several other existing features: more interactive conflict resolution options; de-telescoping and outright exclusion support for sparse checkouts; file-based externals definitions; and operational logging support for svnserve similar to what mod_dav_svn offered. Also, the command-line client introduced a new shortcut syntax for referring to Subversion repository URLs.

Subversion 1.7 (October 2011)

Release 1.7 was primarily a delivery vehicle for two big plumbing overhauls of existing Subversion components. The largest and most impactful of these was the so-called WC-NG—a complete rewrite of the libsvn_wc working copy management library. The second change was the introduction of a sleeker HTTP protocol for Subversion client/server interaction. Subversion 1.7 delivered a handful of additional features, many bug fixes, and some notable performance improvements, too.

Audience

This book is written for computer-literate folk who want to use Subversion to manage their data. While Subversion runs on a number of different operating systems, its primary user interface is command-line-based. That command-line tool (svn), and some additional auxiliary programs, are the focus of this book.

For consistency, the examples in this book assume that the reader is using a Unix-like operating system and is relatively comfortable with Unix and command-line interfaces. That said, the svn program also runs on non-Unix platforms such as Microsoft Windows. With a few minor exceptions, such as the use of backward slashes (\) instead of forward slashes (/) for path separators, the input to and output from this tool when run on Windows are identical to that of its Unix counterpart.

Most readers are probably programmers or system administrators who need to track changes to source code. This is the most common use for Subversion, and therefore it is the scenario underlying all of the book's examples. But Subversion can be used to manage changes to any sort of information—images, music, databases, documentation, and so on. To Subversion, all data is just data.

While this book is written with the assumption that the reader has never used a version control system, we've also tried to make it easy for users of CVS (and other systems) to make a painless leap into Subversion. Special sidebars may mention other version control systems from time to time, and Appendix B, Subversion for CVS Users summarizes many of the differences between CVS and Subversion.

Note also that the source code examples used throughout the book are only examples. While they will compile with the proper compiler incantations, they are intended to illustrate a particular scenario and not necessarily to serve as examples of good programming style or practices.

How to Read This Book

Technical books always face a certain dilemma: whether to cater to top-down or to bottom-up learners. A top-down learner prefers to read or skim documentation, getting a large overview of how the system works; only then does she actually start using the software. A bottom-up learner is a learn by doing person—someone who just wants to dive into the software and figure it out as she goes, referring to book sections when necessary. Most books tend to be written for one type of person or the other, and this book is undoubtedly biased toward top-down learners. (And if you're actually reading this section, you're probably already a top-down learner yourself!) However, if you're a bottom-up person, don't despair. While the book may be laid out as a broad survey of Subversion topics, the content of each section tends to be heavy with specific examples that you can try-by-doing. For the impatient folks who just want to get going, you can jump right to Appendix A, Subversion Quick-Start Guide.

Regardless of your learning style, this book aims to be useful to people of widely different backgrounds—from those with no previous experience in version control to experienced system administrators. Depending on your own background, certain chapters may be more or less important to you. The following can be considered a recommended reading list for various types of readers:

Experienced system administrators

The assumption here is that you've probably used version control before and are dying to get a Subversion server up and running ASAP. Chapter 5, Repository Administration and Chapter 6, Server Configuration will show you how to create your first repository and make it available over the network. After that's done, Chapter 2, Basic Usage and Appendix B, Subversion for CVS Users are the fastest routes to learning the Subversion client.

New users

Your administrator has probably set up Subversion already, and you need to learn how to use the client. If you've never used a version control system, then Chapter 1, Fundamental Concepts is a vital introduction to the ideas behind version control. Chapter 2, Basic Usage is a guided tour of the Subversion client.

Advanced users

Whether you're a user or administrator, eventually your project will grow larger. You're going to want to learn how to do more advanced things with Subversion, such as how to use Subversion's property support (Chapter 3, Advanced Topics), how to use branches and perform merges (Chapter 4, Branching and Merging), how to configure runtime options (Chapter 7, Customizing Your Subversion Experience), and other things. These chapters aren't critical at first, but be sure to read them once you're comfortable with the basics.

Developers

Presumably, you're already familiar with Subversion, and now want to either extend it or build new software on top of its many APIs. Chapter 8, Embedding Subversion is just for you.

The book ends with reference material—Chapter 9, Subversion Complete Reference is a reference guide for all Subversion commands, and the appendixes cover a number of useful topics. These are the chapters you're mostly likely to come back to after you've finished the book.

Organization of This Book

The chapters that follow and their contents are listed here:

Chapter 1, Fundamental Concepts

Explains the basics of version control and different versioning models, along with Subversion's repository, working copies, and revisions.

Chapter 2, Basic Usage

Walks you through a day in the life of a Subversion user. It demonstrates how to use a Subversion client to obtain, modify, and commit data.

Chapter 3, Advanced Topics

Covers more complex features that regular users will eventually come into contact with, such as versioned metadata, file locking, and peg revisions.

Chapter 4, Branching and Merging

Discusses branches, merges, and tagging, including best practices for branching and merging, common use cases, how to undo changes, and how to easily swing from one branch to the next.

Chapter 5, Repository Administration

Describes the basics of the Subversion repository, how to create, configure, and maintain a repository, and the tools you can use to do all of this.

Chapter 6, Server Configuration

Explains how to configure your Subversion server and offers different ways to access your repository: HTTP, the svn protocol, and local disk access. It also covers the details of authentication, authorization and anonymous access.

Chapter 7, Customizing Your Subversion Experience

Explores the Subversion client configuration files, the handling of internationalized text, and how to make external tools cooperate with Subversion.

Chapter 8, Embedding Subversion

Describes the internals of Subversion, the Subversion filesystem, and the working copy administrative areas from a programmer's point of view. It also demonstrates how to use the public APIs to write a program that uses Subversion.

Chapter 9, Subversion Complete Reference

Explains in great detail every subcommand of svn, svnadmin, and svnlook with plenty of examples for the whole family!

Appendix A, Subversion Quick-Start Guide

For the impatient, a whirlwind explanation of how to install Subversion and start using it immediately. You have been warned.

Appendix B, Subversion for CVS Users

Covers the similarities and differences between Subversion and CVS, with numerous suggestions on how to break all the bad habits you picked up from years of using CVS. Included are descriptions of Subversion revision numbers, versioned directories, offline operations, update versus status, branches, tags, metadata, conflict resolution, and authentication.

Appendix C, WebDAV and Autoversioning

Describes the details of WebDAV and DeltaV and how you can configure your Subversion repository to be mounted read/write as a DAV share.

Appendix D, Copyright

A copy of the Creative Commons Attribution License, under which this book is licensed.

This Book Is Free

This book started out as bits of documentation written by Subversion project developers, which were then coalesced into a single work and rewritten. As such, it has always been under a free license (see Appendix D, Copyright). In fact, the book was written in the public eye, originally as part of the Subversion project itself. This means two things:

  • You will always find the latest version of this book in the book's own Subversion repository.

  • You can make changes to this book and redistribute it however you wish—it's under a free license. Your only obligation is to maintain proper attribution to the original authors. Of course, we'd much rather you send feedback and patches to the Subversion developer community, instead of distributing your private version of this book.

The online home of this book's development and most of the volunteer-driven translation efforts regarding it is http://svnbook.red-bean.com. There you can find links to the latest releases and tagged versions of the book in various formats, as well as instructions for accessing the book's Subversion repository (where its DocBook XML source code lives). Feedback is welcomed—encouraged, even. Please submit all comments, complaints, and patches against the book sources to .

Acknowledgments

This book would not be possible (nor very useful) if Subversion did not exist. For that, the authors would like to thank Brian Behlendorf and CollabNet for the vision to fund such a risky and ambitious new open source project; Jim Blandy for the original Subversion name and design—we love you, Jim; and Karl Fogel for being such a good friend and a great community leader, in that order.[4]

Thanks to O'Reilly and the team of professional editors who have helped us polish this text at various stages of its evolution: Chuck Toporek, Linda Mui, Tatiana Apandi, Mary Brady, and Mary Treseler. Your patience and support has been tremendous.

Finally, we thank the countless people who contributed to this book with informal reviews, suggestions, and patches. An exhaustive listing of those folks' names would be impractical to print and maintain here, but may their names live on forever in this book's version control history!




[1] We'll refer to it simply as Subversion throughout this book. You'll thank us when you realize just how much space that saves!

[2] Or as a friend puts it, swatting a fly with a Buick.

[3] CollabNet Enterprise Edition has since been replaced by a new product line called CollabNet TeamForge.

[4] Oh, and thanks, Karl, for being too overworked to write this book yourself.

Chapter 1. Fundamental Concepts

This chapter is a short, casual introduction to Subversion and its approach to version control. We begin with a discussion of general version control concepts, work our way into the specific ideas behind Subversion, and show some simple examples of Subversion in use.

Even though the examples in this chapter show people sharing collections of program source code, keep in mind that Subversion can manage any sort of file collection—it's not limited to helping computer programmers.

Version Control Basics

A version control system (or revision control system) is a system that tracks incremental versions (or revisions) of files and, in some cases, directories over time. Of course, merely tracking the various versions of a user's (or group of users') files and directories isn't very interesting in itself. What makes a version control system useful is the fact that it allows you to explore the changes which resulted in each of those versions and facilitates the arbitrary recall of the same.

In this section, we'll introduce some fairly high-level version control system components and concepts. We'll limit our discussion to modern version control systems—in today's interconnected world, there is very little point in acknowledging version control systems which cannot operate across wide-area networks.

The Repository

At the core of the version control system is a repository, which is the central store of that system's data. The repository usually stores information in the form of a filesystem tree—a hierarchy of files and directories. Any number of clients connect to the repository, and then read or write to these files. By writing data, a client makes the information available to others; by reading data, the client receives information from others. Figure 1.1, “A typical client/server system” illustrates this.

Figure 1.1. A typical client/server system

A typical client/server system


Why is this interesting? So far, this sounds like the definition of a typical file server. And indeed, the repository is a kind of file server, but it's not your usual breed. What makes the repository special is that as the files in the repository are changed, the repository remembers each version of those files.

When a client reads data from the repository, it normally sees only the latest version of the filesystem tree. But what makes a version control client interesting is that it also has the ability to request previous states of the filesystem from the repository. A version control client can ask historical questions such as What did this directory contain last Wednesday? and Who was the last person to change this file, and what changes did he make? These are the sorts of questions that are at the heart of any version control system.

The Working Copy

A version control system's value comes from the fact that it tracks versions of files and directories, but the rest of the software universe doesn't operate on versions of files and directories. Most software programs understand how to operate only on a single version of a specific type of file. So how does a version control user interact with an abstract—and, often, remote—repository full of multiple versions of various files in a concrete fashion? How does his or her word processing software, presentation software, source code editor, web design software, or some other program—all of which trade in the currency of simple data files—get access to such files? The answer is found in the version control construct known as a working copy.

A working copy is, quite literally, a local copy of a particular version of a user's VCS-managed data upon which that user is free to work. Working copies[5] appear to other software just as any other local directory full of files, so those programs don't have to be version-control-aware in order to read from and write to that data. The task of managing the working copy and communicating changes made to its contents to and from the repository falls squarely to the version control system's client software.

Versioning Models

If the primary mission of a version control system is to track the various versions of digital information over time, a very close secondary mission in any modern version control system is to enable collaborative editing and sharing of that data. But different systems use different strategies to achieve this. It's important to understand these different strategies, for a couple of reasons. First, it will help you compare and contrast existing version control systems, in case you encounter other systems similar to Subversion. Beyond that, it will also help you make more effective use of Subversion, since Subversion itself supports a couple of different ways of working.

The problem of file sharing

All version control systems have to solve the same fundamental problem: how will the system allow users to share information, but prevent them from accidentally stepping on each other's feet? It's all too easy for users to accidentally overwrite each other's changes in the repository.

Consider the scenario shown in Figure 1.2, “The problem to avoid”. Suppose we have two coworkers, Harry and Sally. They each decide to edit the same repository file at the same time. If Harry saves his changes to the repository first, it's possible that (a few moments later) Sally could accidentally overwrite them with her own new version of the file. While Harry's version of the file won't be lost forever (because the system remembers every change), any changes Harry made won't be present in Sally's newer version of the file, because she never saw Harry's changes to begin with. Harry's work is still effectively lost—or at least missing from the latest version of the file—and probably by accident. This is definitely a situation we want to avoid!

Figure 1.2. The problem to avoid

The problem to avoid


The lock-modify-unlock solution

Many version control systems use a lock-modify-unlock model to address the problem of many authors clobbering each other's work. In this model, the repository allows only one person to change a file at a time. This exclusivity policy is managed using locks. Harry must lock a file before he can begin making changes to it. If Harry has locked a file, Sally cannot also lock it, and therefore cannot make any changes to that file. All she can do is read the file and wait for Harry to finish his changes and release his lock. After Harry unlocks the file, Sally can take her turn by locking and editing the file. Figure 1.3, “The lock-modify-unlock solution” demonstrates this simple solution.

Figure 1.3. The lock-modify-unlock solution

The lock-modify-unlock solution


The problem with the lock-modify-unlock model is that it's a bit restrictive and often becomes a roadblock for users:

  • Locking may cause administrative problems. Sometimes Harry will lock a file and then forget about it. Meanwhile, because Sally is still waiting to edit the file, her hands are tied. And then Harry goes on vacation. Now Sally has to get an administrator to release Harry's lock. The situation ends up causing a lot of unnecessary delay and wasted time.

  • Locking may cause unnecessary serialization. What if Harry is editing the beginning of a text file, and Sally simply wants to edit the end of the same file? These changes don't overlap at all. They could easily edit the file simultaneously, and no great harm would come, assuming the changes were properly merged together. There's no need for them to take turns in this situation.

  • Locking may create a false sense of security. Suppose Harry locks and edits file A, while Sally simultaneously locks and edits file B. But what if A and B depend on one another, and the changes made to each are semantically incompatible? Suddenly A and B don't work together anymore. The locking system was powerless to prevent the problem—yet it somehow provided a false sense of security. It's easy for Harry and Sally to imagine that by locking files, each is beginning a safe, insulated task, and thus they need not bother discussing their incompatible changes early on. Locking often becomes a substitute for real communication.

The copy-modify-merge solution

Subversion, CVS, and many other version control systems use a copy-modify-merge model as an alternative to locking. In this model, each user's client contacts the project repository and creates a personal working copy. Users then work simultaneously and independently, modifying their private copies. Finally, the private copies are merged together into a new, final version. The version control system often assists with the merging, but ultimately, a human being is responsible for making it happen correctly.

Here's an example. Say that Harry and Sally each create working copies of the same project, copied from the repository. They work concurrently and make changes to the same file A within their copies. Sally saves her changes to the repository first. When Harry attempts to save his changes later, the repository informs him that his file A is out of date. In other words, file A in the repository has somehow changed since he last copied it. So Harry asks his client to merge any new changes from the repository into his working copy of file A. Chances are that Sally's changes don't overlap with his own; once he has both sets of changes integrated, he saves his working copy back to the repository. Figure 1.4, “The copy-modify-merge solution” and Figure 1.5, “The copy-modify-merge solution (continued)” show this process.

Figure 1.4. The copy-modify-merge solution

The copy-modify-merge solution


Figure 1.5. The copy-modify-merge solution (continued)

The copy-modify-merge solution (continued)


But what if Sally's changes do overlap with Harry's changes? What then? This situation is called a conflict, and it's usually not much of a problem. When Harry asks his client to merge the latest repository changes into his working copy, his copy of file A is somehow flagged as being in a state of conflict: he'll be able to see both sets of conflicting changes and manually choose between them. Note that software can't automatically resolve conflicts; only humans are capable of understanding and making the necessary intelligent choices. Once Harry has manually resolved the overlapping changes—perhaps after a discussion with Sally—he can safely save the merged file back to the repository.

The copy-modify-merge model may sound a bit chaotic, but in practice, it runs extremely smoothly. Users can work in parallel, never waiting for one another. When they work on the same files, it turns out that most of their concurrent changes don't overlap at all; conflicts are infrequent. And the amount of time it takes to resolve conflicts is usually far less than the time lost by a locking system.

In the end, it all comes down to one critical factor: user communication. When users communicate poorly, both syntactic and semantic conflicts increase. No system can force users to communicate perfectly, and no system can detect semantic conflicts. So there's no point in being lulled into a false sense of security that a locking system will somehow prevent conflicts; in practice, locking seems to inhibit productivity more than anything else.

Version Control the Subversion Way

We've mentioned already that Subversion is a modern, network-aware version control system. As we described in the section called “Version Control Basics” (our high-level version control overview), a repository serves as the core storage mechanism for Subversion's versioned data, and it's via working copies that users and their software programs interact with that data. In this section, we'll begin to introduce the specific ways in which Subversion implements version control.

Subversion Repositories

Subversion implements the concept of a version control repository much as any other modern version control system would. Unlike a working copy, a Subversion repository is an abstract entity, able to be operated upon almost exclusively by Subversion's own libraries and tools. As most of a user's Subversion interactions involve the use of the Subversion client and occur in the context of a working copy, we spend the majority of this book discussing the Subversion working copy and how to manipulate it. For the finer details of the repository, though, check out Chapter 5, Repository Administration.

Revisions

A Subversion client commits (that is, communicates the changes made to) any number of files and directories as a single atomic transaction. By atomic transaction, we mean simply this: either all of the changes are accepted into the repository, or none of them is. Subversion tries to retain this atomicity in the face of program crashes, system crashes, network problems, and other users' actions.

Each time the repository accepts a commit, this creates a new state of the filesystem tree, called a revision. Each revision is assigned a unique natural number, one greater than the number assigned to the previous revision. The initial revision of a freshly created repository is numbered 0 and consists of nothing but an empty root directory.

Figure 1.6, “Tree changes over time” illustrates a nice way to visualize the repository. Imagine an array of revision numbers, starting at 0, stretching from left to right. Each revision number has a filesystem tree hanging below it, and each tree is a snapshot of the way the repository looked after a commit.

Figure 1.6. Tree changes over time

Tree changes over time


Addressing the Repository

Subversion client programs use URLs to identify versioned files and directories in Subversion repositories. For the most part, these URLs use the standard syntax, allowing for server names and port numbers to be specified as part of the URL.

  • http://svn.example.com/svn/project
  • http://svn.example.com:9834/repos

Subversion repository URLs aren't limited to only the http:// variety. Because Subversion offers several different ways for its clients to communicate with its servers, the URLs used to address the repository differ subtly depending on which repository access mechanism is employed. Table 1.1, “Repository access URLs” describes how different URL schemes map to the available repository access methods. For more details about Subversion's server options, see Chapter 6, Server Configuration.

Table 1.1. Repository access URLs

SchemaAccess method
file:///Direct repository access (on local disk)
http://Access via WebDAV protocol to Subversion-aware Apache server
https://Same as http://, but with SSL encryption
svn://Access via custom protocol to an svnserve server
svn+ssh://Same as svn://, but through an SSH tunnel


Subversion's handling of URLs has some notable nuances. For example, URLs containing the file:// access method (used for local repositories) must, in accordance with convention, have either a server name of localhost or no server name at all:

  • file:///var/svn/repos
  • file://localhost/var/svn/repos

Also, users of the file:// scheme on Windows platforms will need to use an unofficially standard syntax for accessing repositories that are on the same machine, but on a different drive than the client's current working drive. Either of the two following URL path syntaxes will work, where X is the drive on which the repository resides:

  • file:///X:/var/svn/repos
  • file:///X|/var/svn/repos

Note that a URL uses forward slashes even though the native (non-URL) form of a path on Windows uses backslashes. Also note that when using the file:///X|/ form at the command line, you need to quote the URL (wrap it in quotation marks) so that the vertical bar character is not interpreted as a pipe.

[Note]Note

You cannot use Subversion's file:// URLs in a regular web browser the way typical file:// URLs can. When you attempt to view a file:// URL in a regular web browser, it reads and displays the contents of the file at that location by examining the filesystem directly. However, Subversion's resources exist in a virtual filesystem (see the section called “Repository Layer”), and your browser will not understand how to interact with that filesystem.

The Subversion client will automatically encode URLs as necessary, just like a web browser does. For example, the URL http://host/path with space/project/españa — which contains both spaces and upper-ASCII characters — will be automatically interpreted by Subversion as if you'd provided http://host/path%20with%20space/project/espa%C3%B1a. If the URL contains spaces, be sure to place it within quotation marks at the command line so that your shell treats the whole thing as a single argument to the program.

There is one notable exception to Subversion's handling of URLs which also applies to its handling of local paths in many contexts, too. If the final path component of your URL or local path contains an at sign (@), you need to use a special syntax—described in the section called “Peg and Operative Revisions”—in order to make Subversion properly address that resource.

In Subversion 1.6, a new caret (^) notation was introduced as a shorthand for the URL of the repository's root directory. For example, you can use the ^/tags/bigsandwich/ to refer to the URL of the /tags/bigsandwich directory in the root of the repository. Note that this URL syntax works only when your current working directory is a working copy—the command-line client knows the repository's root URL by looking at the working copy's metadata. Also note that when you wish to refer precisely to the root directory of the repository, you must do so using ^/ (with the trailing slash character), not merely ^.

Subversion Working Copies

A Subversion working copy is an ordinary directory tree on your local system, containing a collection of files. You can edit these files however you wish, and if they're source code files, you can compile your program from them in the usual way. Your working copy is your own private work area: Subversion will never incorporate other people's changes, nor make your own changes available to others, until you explicitly tell it to do so. You can even have multiple working copies of the same project.

After you've made some changes to the files in your working copy and verified that they work properly, Subversion provides you with commands to publish your changes to the other people working with you on your project (by writing to the repository). If other people publish their own changes, Subversion provides you with commands to merge those changes into your working copy (by reading from the repository).

A working copy also contains some extra files, created and maintained by Subversion, to help it carry out these commands. In particular, each working copy contains a subdirectory named .svn, also known as the working copy's administrative directory. The files in the administrative directory help Subversion recognize which of your versioned files contain unpublished changes, and which files are out of date with respect to others' work.

[Note]Note

Prior to version 1.7, Subversion maintained .svn administrative subdirectories in every versioned directory of your working copy. Subversion 1.7 offers a completely new approach to how working copy metadata is stored and maintained, and chief among the visible changes to this approach is that each working copy now has only one .svn subdirectory which is an immediate child of the root of that working copy.

[Tip]Tip

While .svn is the de facto name of the Subversion administrative directory, Windows users may run into problems with the ASP.NET Web application framework disallowing access to directories whose names begin with a dot (.). As a special consideration to users in such situations, Subversion will instead use _svn as the administrative directory name if it finds a variable named SVN_ASP_DOT_NET_HACK in its operating environment. Throughout this book, any reference you find to .svn applies also to _svn when this ASP.NET hack is in use.

How the working copy works

For each file in a working directory, Subversion records (among other things) two essential pieces of information:

  • What revision your working file is based on (this is called the file's working revision)

  • A timestamp recording when the local copy was last updated by the repository

Given this information, by talking to the repository, Subversion can tell which of the following four states a working file is in:

Unchanged, and current

The file is unchanged in the working directory, and no changes to that file have been committed to the repository since its working revision. An svn commit of the file will do nothing, and an svn update of the file will do nothing.

Locally changed, and current

The file has been changed in the working directory, and no changes to that file have been committed to the repository since you last updated. There are local changes that have not been committed to the repository; thus an svn commit of the file will succeed in publishing your changes, and an svn update of the file will do nothing.

Unchanged, and out of date

The file has not been changed in the working directory, but it has been changed in the repository. The file should eventually be updated in order to make it current with the latest public revision. An svn commit of the file will do nothing, and an svn update of the file will fold the latest changes into your working copy.

Locally changed, and out of date

The file has been changed both in the working directory and in the repository. An svn commit of the file will fail with an out-of-date error. The file should be updated first; an svn update command will attempt to merge the public changes with the local changes. If Subversion can't complete the merge in a plausible way automatically, it leaves it to the user to resolve the conflict.

Fundamental working copy interactions

A typical Subversion repository often holds the files (or source code) for several projects; usually, each project is a subdirectory in the repository's filesystem tree. In this arrangement, a user's working copy will usually correspond to a particular subtree of the repository.

For example, suppose you have a repository that contains two software projects, paint and calc. Each project lives in its own top-level subdirectory, as shown in Figure 1.7, “The repository's filesystem”.

Figure 1.7. The repository's filesystem

The repository's filesystem


To get a working copy, you must check out some subtree of the repository. (The term check out may sound like it has something to do with locking or reserving resources, but it doesn't; it simply creates a working copy of the project for you.) For example, if you check out /calc, you will get a working copy like this:

$ svn checkout http://svn.example.com/repos/calc
A    calc/Makefile
A    calc/integer.c
A    calc/button.c
Checked out revision 56.
$ ls -A calc
Makefile  button.c integer.c .svn/
$

The list of letter As in the left margin indicates that Subversion is adding a number of items to your working copy. You now have a personal copy of the repository's /calc directory, with one additional entry—.svn—which holds the extra information needed by Subversion, as mentioned earlier.

Suppose you make changes to button.c. Since the .svn directory remembers the file's original modification date and contents, Subversion can tell that you've changed the file. However, Subversion does not make your changes public until you explicitly tell it to. The act of publishing your changes is more commonly known as committing (or checking in) changes to the repository.

To publish your changes to others, you can use Subversion's svn commit command:

$ svn commit button.c -m "Fixed a typo in button.c."
Sending        button.c
Transmitting file data .
Committed revision 57.
$

Now your changes to button.c have been committed to the repository, with a note describing your change (namely, that you fixed a typo). If another user checks out a working copy of /calc, she will see your changes in the latest version of the file.

Suppose you have a collaborator, Sally, who checked out a working copy of /calc at the same time you did. When you commit your change to button.c, Sally's working copy is left unchanged; Subversion modifies working copies only at the user's request.

To bring her project up to date, Sally can ask Subversion to update her working copy, by using the svn update command. This will incorporate your changes into her working copy, as well as any others that have been committed since she checked it out.

$ pwd
/home/sally/calc
$ ls -A
Makefile button.c integer.c .svn/
$ svn update
Updating '.':
U    button.c
Updated to revision 57.
$

The output from the svn update command indicates that Subversion updated the contents of button.c. Note that Sally didn't need to specify which files to update; Subversion uses the information in the .svn directory as well as further information in the repository, to decide which files need to be brought up to date.

Mixed-revision working copies

As a general principle, Subversion tries to be as flexible as possible. One special kind of flexibility is the ability to have a working copy containing files and directories with a mix of different working revision numbers. Subversion working copies do not always correspond to any single revision in the repository; they may contain files from several different revisions. For example, suppose you check out a working copy from a repository whose most recent revision is 4:



calc/

   Makefile:4

   integer.c:4

   button.c:4

At the moment, this working directory corresponds exactly to revision 4 in the repository. However, suppose you make a change to button.c, and commit that change. Assuming no other commits have taken place, your commit will create revision 5 of the repository, and your working copy will now look like this:



calc/

   Makefile:4

   integer.c:4

   button.c:5

Suppose that, at this point, Sally commits a change to integer.c, creating revision 6. If you use svn update to bring your working copy up to date, it will look like this:



calc/

   Makefile:6

   integer.c:6

   button.c:6

Sally's change to integer.c will appear in your working copy, and your change will still be present in button.c. In this example, the text of Makefile is identical in revisions 4, 5, and 6, but Subversion will mark your working copy of Makefile with revision 6 to indicate that it is still current. So, after you do a clean update at the top of your working copy, it will generally correspond to exactly one revision in the repository.

Updates and commits are separate

One of the fundamental rules of Subversion is that a push action does not cause a pull nor vice versa. Just because you're ready to submit new changes to the repository doesn't mean you're ready to receive changes from other people. And if you have new changes still in progress, svn update should gracefully merge repository changes into your own, rather than forcing you to publish them.

The main side effect of this rule is that it means a working copy has to do extra bookkeeping to track mixed revisions as well as be tolerant of the mixture. It's made more complicated by the fact that directories themselves are versioned.

For example, suppose you have a working copy entirely at revision 10. You edit the file foo.html and then perform an svn commit, which creates revision 15 in the repository. After the commit succeeds, many new users would expect the working copy to be entirely at revision 15, but that's not the case! Any number of changes might have happened in the repository between revisions 10 and 15. The client knows nothing of those changes in the repository, since you haven't yet run svn update, and svn commit doesn't pull down new changes. If, on the other hand, svn commit were to automatically download the newest changes, it would be possible to set the entire working copy to revision 15—but then we'd be breaking the fundamental rule of push and pull remaining separate actions. Therefore, the only safe thing the Subversion client can do is mark the one file—foo.html—as being at revision 15. The rest of the working copy remains at revision 10. Only by running svn update can the latest changes be downloaded and the whole working copy be marked as revision 15.

Mixed revisions are normal

The fact is, every time you run svn commit your working copy ends up with some mixture of revisions. The things you just committed are marked as having larger working revisions than everything else. After several commits (with no updates in between), your working copy will contain a whole mixture of revisions. Even if you're the only person using the repository, you will still see this phenomenon. To examine your mixture of working revisions, use the svn status command with the --verbose (-v) option (see the section called “See an overview of your changes” for more information).

Often, new users are completely unaware that their working copy contains mixed revisions. This can be confusing, because many client commands are sensitive to the working revision of the item they're examining. For example, the svn log command is used to display the history of changes to a file or directory (see the section called “Generating a List of Historical Changes”). When the user invokes this command on a working copy object, he expects to see the entire history of the object. But if the object's working revision is quite old (often because svn update hasn't been run in a long time), the history of the older version of the object is shown.

Mixed revisions are useful

If your project is sufficiently complex, you'll discover that it's sometimes nice to forcibly backdate (or update to a revision older than the one you already have) portions of your working copy to an earlier revision; you'll learn how to do that in Chapter 2, Basic Usage. Perhaps you'd like to test an earlier version of a submodule contained in a subdirectory, or perhaps you'd like to figure out when a bug first came into existence in a specific file. This is the time machine aspect of a version control system—the feature that allows you to move any portion of your working copy forward and backward in history.

Mixed revisions have limitations

However you make use of mixed revisions in your working copy, there are limitations to this flexibility.

First, you cannot commit the deletion of a file or directory that isn't fully up to date. If a newer version of the item exists in the repository, your attempt to delete will be rejected to prevent you from accidentally destroying changes you've not yet seen.

Second, you cannot commit a metadata change to a directory unless it's fully up to date. You'll learn about attaching properties to items in Chapter 3, Advanced Topics. A directory's working revision defines a specific set of entries and properties, and thus committing a property change to an out-of-date directory may destroy properties you've not yet seen.

Finally, beginning in Subversion 1.7, you cannot by default use a mixed-revision working copy as the target of a merge operation. (This new requirement was introduced to prevent common problems which stem from doing so.)

Summary

We covered a number of fundamental Subversion concepts in this chapter:

  • We introduced the notions of the central repository, the client working copy, and the array of repository revision trees.

  • We saw some simple examples of how two collaborators can use Subversion to publish and receive changes from one another, using the copy-modify-merge model.

  • We talked a bit about the way Subversion tracks and manages information in a working copy.

At this point, you should have a good idea of how Subversion works in the most general sense. Armed with this knowledge, you should now be ready to move into the next chapter, which is a detailed tour of Subversion's commands and features.




[5] The term working copy can be generally applied to any one file version's local instance. When most folks use the term, though, they are referring to a whole directory tree containing files and subdirectories managed by the version control system.

Chapter 2. Basic Usage

Theory is useful, but its application is just plain fun. Let's move now into the details of using Subversion. By the time you reach the end of this chapter, you will be able to perform all the tasks you need to use Subversion in a normal day's work. You'll start with getting your files into Subversion, followed by an initial checkout of your code. We'll then walk you through making changes and examining those changes. You'll also see how to bring changes made by others into your working copy, examine them, and work through any conflicts that might arise.

This chapter will not provide exhaustive coverage of all of Subversion's commands—rather, it's a conversational introduction to the most common Subversion tasks that you'll encounter. This chapter assumes that you've read and understood Chapter 1, Fundamental Concepts and are familiar with the general model of Subversion. For a complete reference of all commands, see Chapter 9, Subversion Complete Reference.

Also, this chapter assumes that the reader is seeking information about how to interact in a basic fashion with an existing Subversion repository. No repository means no working copy; no working copy means not much of interest in this chapter. There are many Internet sites which offer free or inexpensive Subversion repository hosting services. Or, if you'd prefer to set up and administer your own repositories, check out Chapter 5, Repository Administration. But don't expect the examples in this chapter to work without the user having access to a Subversion repository.

Finally, any Subversion operation that contacts the repository over a network may potentially require that the user authenticate. For the sake of simplicity, our examples throughout this chapter avoid demonstrating and discussing authentication. Be aware that if you hope to apply the knowledge herein to an existing, real-world Subversion instance, you'll probably be forced to provide at least a username and password to the server. See the section called “Client Credentials” for a detailed description of Subversion's handling of authentication and client credentials.

Help!

It goes without saying that this book exists to be a source of information and assistance for Subversion users new and old. Conveniently, though, the Subversion command-line is self-documenting, alleviating the need to grab a book off the shelf (wooden, virtual, or otherwise). The svn help command is your gateway to that built-in documentation:

$ svn help
Subversion command-line client, version 1.7.0.
Type 'svn help <subcommand>' for help on a specific subcommand.
Type 'svn --version' to see the program version and RA modules
  or 'svn --version --quiet' to see just the version number.

Most subcommands take file and/or directory arguments, recursing
on the directories.  If no arguments are supplied to such a
command, it recurses on the current directory (inclusive) by default.

Available subcommands:
   add
   blame (praise, annotate, ann)
   cat
…

As described in the previous output, you can ask for help on a particular subcommand by running svn help SUBCOMMAND. Subversion will respond with the full usage message for that subcommand, including its syntax, options, and behavior:

$ svn help help
help (?, h): Describe the usage of this program or its subcommands.
usage: help [SUBCOMMAND...]

Global options:
  --username ARG           : specify a username ARG
  --password ARG           : specify a password ARG
…

Many Unix-based distributions of Subversion include manual pages of the sort that can be invoked using the man program, but those tend to carry only pointers to other sources of real help, such as the project's website and to the website which hosts this book. Also, several companies offer Subversion help and support, too, usually via a mixture of web-based discussion forums and fee-based consulting. And of course, the Internet holds a decade's worth of Subversion-related discussions just begging to be located by your favorite search engine. Subversion help is never too far away.

Getting Data into Your Repository

You can get new files into your Subversion repository in two ways: svn import and svn add. We'll discuss svn import now and will discuss svn add later in this chapter when we review a typical day with Subversion.

Importing Files and Directories

The svn import command is a quick way to copy an unversioned tree of files into a repository, creating intermediate directories as necessary. svn import doesn't require a working copy, and your files are immediately committed to the repository. You typically use this when you have an existing tree of files that you want to begin tracking in your Subversion repository. For example:

$ svn import /path/to/mytree \
             http://svn.example.com/svn/repo/some/project \
             -m "Initial import"
Adding         mytree/foo.c
Adding         mytree/bar.c
Adding         mytree/subdir
Adding         mytree/subdir/quux.h

Committed revision 1.
$

The previous example copied the contents of the local directory mytree into the directory some/project in the repository. Note that you didn't have to create that new directory first—svn import does that for you. Immediately after the commit, you can see your data in the repository:

$ svn list http://svn.example.com/svn/repo/some/project
bar.c
foo.c
subdir/
$

Note that after the import is finished, the original local directory is not converted into a working copy. To begin working on that data in a versioned fashion, you still need to create a fresh working copy of that tree.

Recommended Repository Layout

Subversion provides the ultimate flexibility in terms of how you arrange your data. Because it simply versions directories and files, and because it ascribes no particular meaning to any of those objects, you may arrange the data in your repository in any way that you choose. Unfortunately, this flexibility also means that it's easy to find yourself lost without a roadmap as you attempt to navigate different Subversion repositories which may carry completely different and unpredictable arrangements of the data within them.

To counteract this confusion, we recommend that you follow a repository layout convention (established long ago, in the nascency of the Subversion project itself) in which a handful of strategically named Subversion repository directories convey valuable meaning about the data they hold. Most projects have a recognizable main line, or trunk, of development; some branches, which are divergent copies of development lines; and some tags, which are named, stable snapshots of a particular line of development. So we first recommend that each project have a recognizable project root in the repository, a directory under which all of the versioned information for that project—and only that project—lives. Secondly, we suggest that each project root contain a trunk subdirectory for the main development line, a branches subdirectory in which specific branches (or collections of branches) will be created, and a tags subdirectory in which specific tags (or collections of tags) will be created. Of course, if a repository houses only a single project, the root of the repository can serve as the project root, too.

Here are some examples:

$ svn list file:///var/svn/single-project-repo
trunk/
branches/
tags/
$ svn list file:///var/svn/multi-project-repo
project-A/
project-B/
$ svn list file:///var/svn/multi-project-repo/project-A
trunk/
branches/
tags/
$

We talk much more about tags and branches in Chapter 4, Branching and Merging. For details and some advice on how to set up repositories when you have multiple projects, see the section called “Repository Layout”. Finally, we discuss project roots more in the section called “Planning Your Repository Organization”.

What's In a Name?

Subversion tries hard not to limit the type of data you can place under version control. The contents of files and property values are stored and transmitted as binary data, and the section called “File Content Type” tells you how to give Subversion a hint that textual operations don't make sense for a particular file. There are a few places, however, where Subversion places restrictions on information it stores.

Subversion internally handles certain bits of data—for example, property names, pathnames, and log messages—as UTF-8-encoded Unicode. This is not to say that all your interactions with Subversion must involve UTF-8, though. As a general rule, Subversion clients will gracefully and transparently handle conversions between UTF-8 and the encoding system in use on your computer, if such a conversion can meaningfully be done (which is the case for most common encodings in use today).

In WebDAV exchanges and older versions of some of Subversion's administrative files, paths are used as XML attribute values, and property names in XML tag names. This means that pathnames can contain only legal XML (1.0) characters, and properties are further limited to ASCII characters. Subversion also prohibits TAB, CR, and LF characters in path names to prevent paths from being broken up in diffs or in the output of commands such as svn log or svn status.

While it may seem like a lot to remember, in practice these limitations are rarely a problem. As long as your locale settings are compatible with UTF-8 and you don't use control characters in path names, you should have no trouble communicating with Subversion. The command-line client adds an extra bit of help—to create legally correct versions for internal use it will automatically escape illegal path characters as needed in URLs that you type.

Creating a Working Copy

Most of the time, you will start using a Subversion repository by performing a checkout of your project. Checking out a directory from a repository creates a working copy of that directory on your local machine. Unless otherwise specified, this copy contains the youngest (that is, most recently created or modified) versions of the directory and its children found in the Subversion repository:

$ svn checkout http://svn.example.com/svn/repo/trunk
A    trunk/README
A    trunk/INSTALL
A    trunk/src/main.c
A    trunk/src/header.h
…
Checked out revision 8810.
$

Although the preceding example checks out the trunk directory, you can just as easily check out a deeper subdirectory of a repository by specifying that subdirectory's URL as the checkout URL:

$ svn checkout http://svn.example.com/svn/repo/trunk/src
A    src/main.c
A    src/header.h
A    src/lib/helpers.c
…
Checked out revision 8810.
$

Since Subversion uses a copy-modify-merge model instead of lock-modify-unlock (see the section called “Versioning Models”), you can immediately make changes to the files and directories in your working copy. Your working copy is just like any other collection of files and directories on your system. You can edit the files inside it, rename it, even delete the entire working copy and forget about it.

[Warning]Warning

While your working copy is just like any other collection of files and directories on your system, you can edit files at will, but you must tell Subversion about everything else that you do. For example, if you want to copy or move an item in a working copy, you should use svn copy or svn move instead of the copy and move commands provided by your operating system. We'll talk more about them later in this chapter.

Unless you're ready to commit the addition of a new file or directory or changes to existing ones, there's no need to further notify the Subversion server that you've done anything.

Notice that in the previous pair of examples, Subversion chose to create a working copy in a directory named for the final component of the checkout URL. This occurs only as a convenience to the user when the checkout URL is the only bit of information provided to the svn checkout command. Subversion's command-line client gives you additional flexibility, though, allowing you to optionally specify the local directory name that Subversion should use for the working copy it creates. For example:

$ svn checkout http://svn.example.com/svn/repo/trunk my-working-copy
A    my-working-copy/README
A    my-working-copy/INSTALL
A    my-working-copy/src/main.c
A    my-working-copy/src/header.h
…
Checked out revision 8810.
$

If the local directory you specify doesn't yet exist, that's okay—svn checkout will create it for you.

Basic Work Cycle

Subversion has numerous features, options, bells, and whistles, but on a day-to-day basis, odds are that you will use only a few of them. In this section, we'll run through the most common things that you might find yourself doing with Subversion in the course of a day's work.

The typical work cycle looks like this:

  1. Update your working copy. This involves the use of the svn update command.

  2. Make your changes. The most common changes that you'll make are edits to the contents of your existing files. But sometimes you need to add, remove, copy and move files and directories—the svn add, svn delete, svn copy, and svn move commands handle those sorts of structural changes within the working copy.

  3. Review your changes. The svn status and svn diff commands are critical to reviewing the changes you've made in your working copy.

  4. Fix your mistakes. Nobody's perfect, so as you review your changes, you may spot something that's not quite right. Sometimes the easiest way to fix a mistake is start all over again from scratch. The svn revert command restores a file or directory to its unmodified state.

  5. Resolve any conflicts (merge others' changes). In the time it takes you to make and review your changes, others might have made and published changes, too. You'll want to integrate their changes into your working copy to avoid the potential out-of-dateness scenarios when you attempt to publish your own. Again, the svn update command is the way to do this. If this results in local conflicts, you'll need to resolve those using the svn resolve command.

  6. Publish (commit) your changes. The svn commit command transmits your changes to the repository where, if they are accepted, they create the newest versions of all the things you modified. Now others can see your work, too!

Update Your Working Copy

When working on a project that is being modified via multiple working copies, you'll want to update your working copy to receive any changes committed from other working copies since your last update. These might be changes that other members of your project team have made, or they might simply be changes you've made yourself from a different computer. To protect your data, Subversion won't allow you commit new changes to out-of-date files and directories, so it's best to have the latest versions of all your project's files and directories before making new changes of your own.

Use svn update to bring your working copy into sync with the latest revision in the repository:

$ svn update
Updating '.':
U    foo.c
U    bar.c
Updated to revision 2.
$

In this case, it appears that someone checked in modifications to both foo.c and bar.c since the last time you updated, and Subversion has updated your working copy to include those changes.

When the server sends changes to your working copy via svn update, a letter code is displayed next to each item to let you know what actions Subversion performed to bring your working copy up to date. To find out what these letters mean, run svn help update or see svn update (up) in Chapter 9, Subversion Complete Reference.

Make Your Changes

Now you can get to work and make changes in your working copy. You can make two kinds of changes to your working copy: file changes and tree changes. You don't need to tell Subversion that you intend to change a file; just make your changes using your text editor, word processor, graphics program, or whatever tool you would normally use. Subversion automatically detects which files have been changed, and in addition, it handles binary files just as easily as it handles text files—and just as efficiently, too. Tree changes are different, and involve changes to a directory's structure. Such changes include adding and removing files, renaming files or directories, and copying files or directories to new locations. For tree changes, you use Subversion operations to schedule files and directories for removal, addition, copying, or moving. These changes may take place immediately in your working copy, but no additions or removals will happen in the repository until you commit them.

Here is an overview of the five Subversion subcommands that you'll use most often to make tree changes:

svn add FOO

Use this to schedule the file, directory, or symbolic link FOO to be added to the repository. When you next commit, FOO will become a child of its parent directory. Note that if FOO is a directory, everything underneath FOO will be scheduled for addition. If you want only to add FOO itself, pass the --depth=empty option.

svn delete FOO

Use this to schedule the file, directory, or symbolic link FOO to be deleted from the repository. If FOO is a file or link, it is immediately deleted from your working copy. If FOO is a directory, it is not deleted, but Subversion schedules it for deletion. When you commit your changes, FOO will be entirely removed from your working copy and the repository.[6]

svn copy FOO BAR

Create a new item BAR as a duplicate of FOO and automatically schedule BAR for addition. When BAR is added to the repository on the next commit, its copy history is recorded (as having originally come from FOO). svn copy does not create intermediate directories unless you pass the --parents option.

svn move FOO BAR

This command is exactly the same as running svn copy FOO BAR; svn delete FOO. That is, BAR is scheduled for addition as a copy of FOO, and FOO is scheduled for removal. svn move does not create intermediate directories unless you pass the --parents option.

svn mkdir FOO

This command is exactly the same as running mkdir FOO; svn add FOO. That is, a new directory named FOO is created and scheduled for addition.

Review Your Changes

Once you've finished making changes, you need to commit them to the repository, but before you do so, it's usually a good idea to take a look at exactly what you've changed. By examining your changes before you commit, you can compose a more accurate log message (a human-readable description of the committed changes stored alongside those changes in the repository). You may also discover that you've inadvertently changed a file, and that you need to undo that change before committing. Additionally, this is a good opportunity to review and scrutinize changes before publishing them. You can see an overview of the changes you've made by using the svn status command, and you can dig into the details of those changes by using the svn diff command.

See an overview of your changes

To get an overview of your changes, use the svn status command. You'll probably use svn status more than any other Subversion command.

[Tip]Tip

Because the cvs status command's output was so noisy, and because cvs update not only performs an update, but also reports the status of your local changes, most CVS users have grown accustomed to using cvs update to report their changes. In Subversion, the update and status reporting facilities are completely separate. See the section called “Distinction Between Status and Update” for more details.

If you run svn status at the top of your working copy with no additional arguments, it will detect and report all file and tree changes you've made.

$ svn status
?       scratch.c
A       stuff/loot
A       stuff/loot/new.c
D       stuff/old.c
M       bar.c
$

In its default output mode, svn status prints seven columns of characters, followed by several whitespace characters, followed by a file or directory name. The first column tells the status of a file or directory and/or its contents. Some of the most common codes that svn status displays are:

? item

The file, directory, or symbolic link item is not under version control.

A item

The file, directory, or symbolic link item has been scheduled for addition into the repository.

C item

The file item is in a state of conflict. That is, changes received from the server during an update overlap with local changes that you have in your working copy (and weren't resolved during the update). You must resolve this conflict before committing your changes to the repository.

D item

The file, directory, or symbolic link item has been scheduled for deletion from the repository.

M item

The contents of the file item have been modified.

If you pass a specific path to svn status, you get information about that item alone:

$ svn status stuff/fish.c
D       stuff/fish.c

svn status also has a --verbose (-v) option, which will show you the status of every item in your working copy, even if it has not been changed:

$ svn status -v
M               44        23    sally     README
                44        30    sally     INSTALL
M               44        20    harry     bar.c
                44        18    ira       stuff
                44        35    harry     stuff/trout.c
D               44        19    ira       stuff/fish.c
                44        21    sally     stuff/things
A                0         ?     ?        stuff/things/bloo.h
                44        36    harry     stuff/things/gloo.c

This is the long form output of svn status. The letters in the first column mean the same as before, but the second column shows the working revision of the item. The third and fourth columns show the revision in which the item last changed, and who changed it.

None of the prior invocations to svn status contact the repository—they merely report what is known about the working copy items based on the records stored in the working copy administrative area and on the timestamps and contents of modified files. But sometimes it is useful to see which of the items in your working copy have been modified in the repository since the last time you updated your working copy. For this, svn status offers the --show-updates (-u) option, which contacts the repository and adds information about items that are out of date:

$ svn status -u -v
M      *        44        23    sally     README
M               44        20    harry     bar.c
       *        44        35    harry     stuff/trout.c
D               44        19    ira       stuff/fish.c
A                0         ?     ?        stuff/things/bloo.h
Status against revision:   46

Notice in the previous example the two asterisks: if you were to run svn update at this point, you would receive changes to README and trout.c. This tells you some very useful information—because one of those items is also one that you have locally modified (the file README), you'll need to update and get the servers changes for that file before you commit, or the repository will reject your commit for being out of date. We discuss this in more detail later.

svn status can display much more information about the files and directories in your working copy than we've shown here—for an exhaustive description of svn status and its output, run svn help status or see svn status (stat, st) in Chapter 9, Subversion Complete Reference.

Examine the details of your local modifications

Another way to examine your changes is with the svn diff command, which displays differences in file content. When you run svn diff at the top of your working copy with no arguments, Subversion will print the changes you've made to human-readable files in your working copy. It displays those changes in unified diff format, a format which describes changes as hunks (or snippets) of a file's content where each line of text is prefixed with a single-character code: a space, which means the line was unchanged; a minus sign (-), which means the line was removed from the file; or a plus sign (+), which means the line was added to the file. In the context of svn diff, those minus-sign- and plus-sign-prefixed lines show how the lines looked before and after your modifications, respectively.

Here's an example:

$ svn diff
Index: bar.c
===================================================================
--- bar.c	(revision 3)
+++ bar.c	(working copy)
@@ -1,7 +1,12 @@
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+
+#include <stdio.h>

 int main(void) {
-  printf("Sixty-four slices of American Cheese...\n");
+  printf("Sixty-five slices of American Cheese...\n");
 return 0;
 }

Index: README
===================================================================
--- README	(revision 3)
+++ README	(working copy)
@@ -193,3 +193,4 @@
+Note to self:  pick up laundry.

Index: stuff/fish.c
===================================================================
--- stuff/fish.c	(revision 1)
+++ stuff/fish.c	(working copy)
-Welcome to the file known as 'fish'.
-Information on fish will be here soon.

Index: stuff/things/bloo.h
===================================================================
--- stuff/things/bloo.h	(revision 8)
+++ stuff/things/bloo.h	(working copy)
+Here is a new file to describe
+things about bloo.

The svn diff command produces this output by comparing your working files against its pristine text-base. Files scheduled for addition are displayed as files in which every line was added; files scheduled for deletion are displayed as if every line was removed from those files. The output from svn diff is somehwat compatible with the patch program—more so with the svn patch subcommand introduced in Subversion 1.7. Patch processing commands such as these read and apply patch files (or patches), which are files that describe differences made to one or more files. Because of this, you can share the changes you've made in your working copy with someone else without first committing those changes by creating a patch file from the redirected output of svn diff:

$ svn diff > patchfile
$

Subversion uses its internal diff engine, which produces unified diff format, by default. If you want diff output in a different format, specify an external diff program using --diff-cmd and pass any additional flags that it needs via the --extensions (-x) option. For example, you might want Subversion to defer its difference calculation and display to the GNU diff program, asking that program to print local modifications made to the file foo.c in context diff format (another flavor of difference format) while ignoring changes made only to the case of the letters used in the file's contents:

$ svn diff --diff-cmd /usr/bin/diff -x "-i" foo.c
…
$

Fix Your Mistakes

Suppose while viewing the output of svn diff you determine that all the changes you made to a particular file are mistakes. Maybe you shouldn't have changed the file at all, or perhaps it would be easier to make different changes starting from scratch. You could edit the file again and unmake all those changes. You could try to find a copy of how the file looked before you changed it, and then copy its contents atop your modified version. You could attempt to apply those changes to the file again in reverse using patch -R. And there are probably other approaches you could take.

Fortunately in Subversion, undoing your work and starting over from scratch doesn't require such acrobatics. Just use the svn revert command:

$ svn status README
M       README
$ svn revert README
Reverted 'README'
$ svn status README
$

In this example, Subversion has reverted the file to its premodified state by overwriting it with the pristine version of the file cached in the text-base area. But note that svn revert can undo any scheduled operation—for example, you might decide that you don't want to add a new file after all:

$ svn status new-file.txt
?       new-file.txt
$ svn add new-file.txt
A         new-file.txt
$ svn revert new-file.txt
Reverted 'new-file.txt'
$ svn status new-file.txt
?       new-file.txt
$

Or perhaps you mistakenly removed a file from version control:

$ svn status README
$ svn delete README
D         README
$ svn revert README
Reverted 'README'
$ svn status README
$

The svn revert command offers salvation for imperfect people. It can save you huge amounts of time and energy that would otherwise be spent manually unmaking changes or, worse, disposing of your working copy and checking out a fresh one just to have a clean slate to work with again.

Resolve Any Conflicts

We've already seen how svn status -u can predict conflicts, but dealing with those conflicts is still something that remains to be done. Conflicts can occur any time you attempt to merge or integrate (in a very general sense) changes from the repository into your working copy. By now you know that svn update creates exactly that sort of scenario—that command's very purpose is to bring your working copy up to date with the repository by merging all the changes made since your last update into your working copy. So how does Subversion report these conflicts to you, and how do you deal with them?

Suppose you run svn update and you see this sort of interesting output:

$ svn update
Updating '.':
U    INSTALL
G    README
Conflict discovered in 'bar.c'.
Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: 

The U (which stands for Updated) and G (for merGed) codes are no cause for concern; those files cleanly absorbed changes from the repository. A file marked with U contains no local changes but was updated with changes from the repository. One marked with G had local changes to begin with, but the changes coming from the repository didn't overlap with those local changes.

It's the next few lines which are interesting. First, Subversion reports to you that in its attempt to merge outstanding server changes into the file bar.c, it has detected that some of those changes clash with local modifications you've made to that file in your working copy but have not yet committed. Perhaps someone has changed the same line of text you also changed. Whatever the reason, Subversion instantly flags this file as being in a state of conflict. It then asks you what you want to do about the problem, allowing you to interactively choose an action to take toward resolving the conflict. The most commonly used options are displayed, but you can see all of the options by typing s:

…
Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: s

  (e)  edit             - change merged file in an editor
  (df) diff-full        - show all changes made to merged file
  (r)  resolved         - accept merged version of file

  (dc) display-conflict - show all conflicts (ignoring merged version)
  (mc) mine-conflict    - accept my version for all conflicts (same)
  (tc) theirs-conflict  - accept their version for all conflicts (same)

  (mf) mine-full        - accept my version of entire file (even non-conflicts)
  (tf) theirs-full      - accept their version of entire file (same)

  (p)  postpone         - mark the conflict to be resolved later
  (l)  launch           - launch external tool to resolve conflict
  (s)  show all         - show this list

Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options:

Let's briefly review each of these options before we go into detail on what each option means.

(e) edit

Open the file in conflict with your favorite editor, as set in the environment variable EDITOR.

(df) diff-full

Display the differences between the base revision and the conflicted file itself in unified diff format.

(r) resolved

After editing a file, tell svn that you've resolved the conflicts in the file and that it should accept the current contents—basically that you've resolved the conflict.

(dc) display-conflict

Display all conflicting regions of the file, ignoring changes which were successfully merged.

(mc) mine-conflict

Discard any newly received changes from the server which conflict with your local changes to the file under review. However, accept and merge all non-conflicting changes received from the server for that file.

(tc) theirs-conflict

Discard any local changes which conflict with incoming changes from the server for the file under review. However, preserve all non-conflicting local changes to that file.

(mf) mine-full

Discard all newly received changes from the server for the file under review, but preserve all your local changes for that file.

(tf) theirs-full

Discard all your local changes to the file under review and use only the newly received changes from the server for that file.

(p) postpone

Leave the file in a conflicted state for you to resolve after your update is complete.

(l) launch

Launch an external program to perform the conflict resolution. This requires a bit of preparation beforehand.

(s) show all

Show the list of all possible commands you can use in interactive conflict resolution.

We'll cover these commands in more detail now, grouping them together by related functionality.

Viewing conflict differences interactively

Before deciding how to attack a conflict interactively, odds are that you'd like to see exactly what is in conflict. Two of the commands available at the interactive conflict resolution prompt can assist you here. The first is the diff-full command (df), which displays all the local modifications to the file in question plus any conflict regions:

…
Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: df
--- .svn/text-base/sandwich.txt.svn-base      Tue Dec 11 21:33:57 2007
+++ .svn/tmp/tempfile.32.tmp     Tue Dec 11 21:34:33 2007
@@ -1 +1,5 @@
-Just buy a sandwich.
+<<<<<<< .mine
+Go pick up a cheesesteak.
+=======
+Bring me a taco!
+>>>>>>> .r32
…

The first line of the diff content shows the previous contents of the working copy (the BASE revision), the next content line is your change, and the last content line is the change that was just received from the server (usually the HEAD revision).

The second command is similar to the first, but the display-conflict (dc) command shows only the conflict regions, not all the changes made to the file. Additionally, this command uses a slightly different display format for the conflict regions which allows you to more easily compare the file's contents in those regions as they would appear in each of three states: original and unedited; with your local changes applied and the server's conflicting changes ignored; and with only the server's incoming changes applied and your local, conflicting changes reverted.

After reviewing the information provided by these commands, you're ready to move on to the next action.

Resolving conflict differences interactively

There are several different ways to resolve conflicts interactively—two of which allow you to selectively merge and edit changes, the rest of which allow you to simply pick a version of the file and move along.

If you wish to choose some combination of your local changes, you can use the edit command (e) to manually edit the file with conflict markers in a text editor (configured per the instructions in the section called “Using External Editors”). After you've edited the file, if you're satisfied with the changes you've made, you can tell Subversion that the edited file is no longer in conflict by using the resolved command (r).

Regardless of what your local Unix snob will likely tell you, editing the file by hand in your favorite text editor is a somewhat low-tech way of remedying conflicts (see the section called “Merging conflicts by hand” for a walkthrough). For this reason, Subversion provides the launch resolution command (l) to fire up a fancy graphical merge tool instead (see the section called “External merge”).

If you decide that you don't need to merge any changes, but just want to accept one version of the file or the other, you can either choose your changes (a.k.a. mine) by using the mine-full command (mf) or choose theirs by using the theirs-full command (tf).

Finally, there is also a pair of compromise options available. The mine-conflict (mc) and theirs-conflict (tc) commands instruct Subversion to select your local changes or the server's incoming changes, respectively, as the winner for all conflicts in the file. But, unlike the mine-full and theirs-full commands, these commands preserve both your local changes and changes received from the server in regions of the file where no conflict was detected.

Postponing conflict resolution

This may sound like an appropriate section for avoiding marital disagreements, but it's actually still about Subversion, so read on. If you're doing an update and encounter a conflict that you're not prepared to review or resolve, you can type p to postpone resolving a conflict on a file-by-file basis when you run svn update. If you know in advance that you don't want to resolve any conflicts interactively, you can pass the --non-interactive option to svn update, and any file in conflict will be marked with a C automatically.

The C (for Conflicted) means that the changes from the server overlapped with your own, and now you have to manually choose between them after the update has completed. When you postpone a conflict resolution, svn typically does three things to assist you in noticing and resolving that conflict:

  • Subversion prints a C during the update and remembers that the file is in a state of conflict.

  • If Subversion considers the file to be mergeable, it places conflict markers—special strings of text that delimit the sides of the conflict—into the file to visibly demonstrate the overlapping areas. (Subversion uses the svn:mime-type property to decide whether a file is capable of contextual, line-based merging. See the section called “File Content Type” to learn more.)

  • For every conflicted file, Subversion places three extra unversioned files in your working copy:

    filename.mine

    This is the file as it existed in your working copy before you began the update process. It contains any local modifications you had made to the file up to that point. (If Subversion considers the file to be unmergeable, the .mine file isn't created, since it would be identical to the working file.)

    filename.rOLDREV

    This is the file as it existed in the BASE revision—that is, the unmodified revision of the file in your working copy before you began the update process—where OLDREV is that base revision number.

    filename.rNEWREV

    This is the file that your Subversion client just received from the server via the update of your working copy, where NEWREV corresponds to the revision number to which you were updating (HEAD, unless otherwise requested).

For example, Sally makes changes to the file sandwich.txt, but does not yet commit those changes. Meanwhile, Harry commits changes to that same file. Sally updates her working copy before committing and she gets a conflict, which she postpones:

$ svn update
Updating '.':
Conflict discovered in 'sandwich.txt'.
Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: p
C    sandwich.txt
Updated to revision 2.
Summary of conflicts:
  Text conflicts: 1
$ ls -1
sandwich.txt
sandwich.txt.mine
sandwich.txt.r1
sandwich.txt.r2

At this point, Subversion will not allow Sally to commit the file sandwich.txt until the three temporary files are removed:

$ svn commit -m "Add a few more things"
svn: E155015: Commit failed (details follow):
svn: E155015: Aborting commit: '/home/sally/svn-work/sandwich.txt' remains in conflict

If you've postponed a conflict, you need to resolve the conflict before Subversion will allow you to commit your changes. You'll do this with the svn resolve command and one of several arguments to the --accept option.

If you want to choose the version of the file that you last checked out before making your edits, choose the base argument.

If you want to choose the version that contains only your edits, choose the mine-full argument.

If you want to choose the version that your most recent update pulled from the server (and thus discarding your edits entirely), choose the theirs-full argument.

However, if you want to pick and choose from your changes and the changes that your update fetched from the server, merge the conflicted text by hand (by examining and editing the conflict markers within the file) and then choose the working argument.

svn resolve removes the three temporary files and accepts the version of the file that you specified with the --accept option, and Subversion no longer considers the file to be in a state of conflict:

$ svn resolve --accept working sandwich.txt
Resolved conflicted state of 'sandwich.txt'

Merging conflicts by hand

Merging conflicts by hand can be quite intimidating the first time you attempt it, but with a little practice, it can become as easy as falling off a bike.

Here's an example. Due to a miscommunication, you and Sally, your collaborator, both edit the file sandwich.txt at the same time. Sally commits her changes, and when you go to update your working copy, you get a conflict and you're going to have to edit sandwich.txt to resolve the conflict. First, let's take a look at the file:

$ cat sandwich.txt
Top piece of bread
Mayonnaise
Lettuce
Tomato
Provolone
<<<<<<< .mine
Salami
Mortadella
Prosciutto
=======
Sauerkraut
Grilled Chicken
>>>>>>> .r2
Creole Mustard
Bottom piece of bread

The strings of less-than signs, equals signs, and greater-than signs are conflict markers and are not part of the actual data in conflict. You generally want to ensure that those are removed from the file before your next commit. The text between the first two sets of markers is composed of the changes you made in the conflicting area:

<<<<<<< .mine
Salami
Mortadella
Prosciutto
=======

The text between the second and third sets of conflict markers is the text from Sally's commit:

=======
Sauerkraut
Grilled Chicken
>>>>>>> .r2

Usually you won't want to just delete the conflict markers and Sally's changes—she's going to be awfully surprised when the sandwich arrives and it's not what she wanted. This is where you pick up the phone or walk across the office and explain to Sally that you can't get sauerkraut from an Italian deli.[7] Once you've agreed on the changes you will commit, edit your file and remove the conflict markers:

Top piece of bread
Mayonnaise
Lettuce
Tomato
Provolone
Salami
Mortadella
Prosciutto
Creole Mustard
Bottom piece of bread

Now use svn resolve, and you're ready to commit your changes:

$ svn resolve --accept working sandwich.txt
Resolved conflicted state of 'sandwich.txt'
$ svn commit -m "Go ahead and use my sandwich, discarding Sally's edits."

Note that svn resolve, unlike most of the other commands we deal with in this chapter, requires that you explicitly list any filenames that you wish to resolve. In any case, you want to be careful and use svn resolve only when you're certain that you've fixed the conflict in your file—once the temporary files are removed, Subversion will let you commit the file even if it still contains conflict markers.

If you ever get confused while editing the conflicted file, you can always consult the three files that Subversion creates for you in your working copy—including your file as it was before you updated. You can even use a third-party interactive merging tool to examine those three files.

Discarding your changes in favor of a newly fetched revision

If you get a conflict and decide that you want to throw out your changes, you can run svn resolve --accept theirs-full CONFLICTED-PATH and Subversion will discard your edits and remove the temporary files:

$ svn update
Updating '.':
Conflict discovered in 'sandwich.txt'.
Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: p
C    sandwich.txt
Updated to revision 2.
Summary of conflicts:
  Text conflicts: 1
$ ls sandwich.*
sandwich.txt  sandwich.txt.mine  sandwich.txt.r2  sandwich.txt.r1
$ svn resolve --accept theirs-full sandwich.txt
Resolved conflicted state of 'sandwich.txt'
$

Punting: using svn revert

If you decide that you want to throw out your changes and start your edits again (whether this occurs after a conflict or anytime), just revert your changes:

$ svn revert sandwich.txt
Reverted 'sandwich.txt'
$ ls sandwich.*
sandwich.txt
$

Note that when you revert a conflicted file, you don't have to use svn resolve.

Commit Your Changes

Finally! Your edits are finished, you've merged all changes from the server, and you're ready to commit your changes to the repository.

The svn commit command sends all of your changes to the repository. When you commit a change, you need to supply a log message describing your change. Your log message will be attached to the new revision you create. If your log message is brief, you may wish to supply it on the command line using the --message (-m) option:

$ svn commit -m "Corrected number of cheese slices."
Sending        sandwich.txt
Transmitting file data .
Committed revision 3.

However, if you've been composing your log message in some other text file as you work, you may want to tell Subversion to get the message from that file by passing its filename as the value of the --file (-F) option:

$ svn commit -F logmsg
Sending        sandwich.txt
Transmitting file data .
Committed revision 4.

If you fail to specify either the --message (-m) or --file (-F) option, Subversion will automatically launch your favorite editor (see the information on editor-cmd in the section called “Config”) for composing a log message.

[Tip]Tip

If you're in your editor writing a commit message and decide that you want to cancel your commit, you can just quit your editor without saving changes. If you've already saved your commit message, simply delete all the text, save again, and then abort:

$ svn commit
Waiting for Emacs...Done

Log message unchanged or not specified
(a)bort, (c)ontinue, (e)dit
a
$

The repository doesn't know or care whether your changes make any sense as a whole; it checks only to make sure nobody else has changed any of the same files that you did when you weren't looking. If somebody has done that, the entire commit will fail with a message informing you that one or more of your files are out of date:

$ svn commit -m "Add another rule"
Sending        rules.txt
svn: E155011: Commit failed (details follow):
svn: E155011: File '/home/sally/svn-work/sandwich.txt' is out of date
…

(The exact wording of this error message depends on the network protocol and server you're using, but the idea is the same in all cases.)

At this point, you need to run svn update, deal with any merges or conflicts that result, and attempt your commit again.

That covers the basic work cycle for using Subversion. Subversion offers many other features that you can use to manage your repository and working copy, but most of your day-to-day use of Subversion will involve only the commands that we've discussed so far in this chapter. We will, however, cover a few more commands that you'll use fairly often.

Examining History

Your Subversion repository is like a time machine. It keeps a record of every change ever committed and allows you to explore this history by examining previous versions of files and directories as well as the metadata that accompanies them. With a single Subversion command, you can check out the repository (or restore an existing working copy) exactly as it was at any date or revision number in the past. However, sometimes you just want to peer into the past instead of going into it.

Several commands can provide you with historical data from the repository:

svn diff

Shows line-level details of a particular change

svn log

Shows you broad information: log messages with date and author information attached to revisions and which paths changed in each revision

svn cat

Retrieves a file as it existed in a particular revision number and displays it on your screen

svn annotate

Retrieves a human-readable file as it existed in a particular revision number, displaying its contents in a tabular form with last-changed information attributed to each line of the file.

svn list

Displays the files in a directory for any given revision

Examining the Details of Historical Changes

We've already seen svn diff before—it displays file differences in unified diff format; we used it to show the local modifications made to our working copy before committing to the repository.

In fact, it turns out that there are three distinct uses of svn diff:

  • Examining local changes

  • Comparing your working copy to the repository

  • Comparing repository revisions

Examining local changes

As we've seen, invoking svn diff with no options will compare your working files to the cached pristine copies in the .svn area:

$ svn diff
Index: rules.txt
===================================================================
--- rules.txt	(revision 3)
+++ rules.txt	(working copy)
@@ -1,4 +1,5 @@
 Be kind to others
 Freedom = Responsibility
 Everything in moderation
-Chew with your mouth open
+Chew with your mouth closed
+Listen when others are speaking
$

Comparing working copy to repository

If a single --revision (-r) number is passed, your working copy is compared to the specified revision in the repository:

$ svn diff -r 3 rules.txt
Index: rules.txt
===================================================================
--- rules.txt	(revision 3)
+++ rules.txt	(working copy)
@@ -1,4 +1,5 @@
 Be kind to others
 Freedom = Responsibility
 Everything in moderation
-Chew with your mouth open
+Chew with your mouth closed
+Listen when others are speaking
$

Comparing repository revisions

If two revision numbers, separated by a colon, are passed via --revision (-r), the two revisions are directly compared:

$ svn diff -r 2:3 rules.txt
Index: rules.txt
===================================================================
--- rules.txt	(revision 2)
+++ rules.txt	(revision 3)
@@ -1,4 +1,4 @@
 Be kind to others
-Freedom = Chocolate Ice Cream
+Freedom = Responsibility
 Everything in moderation
 Chew with your mouth open
$

A more convenient way of comparing one revision to the previous revision is to use the --change (-c) option:

$ svn diff -c 3 rules.txt
Index: rules.txt
===================================================================
--- rules.txt	(revision 2)
+++ rules.txt	(revision 3)
@@ -1,4 +1,4 @@
 Be kind to others
-Freedom = Chocolate Ice Cream
+Freedom = Responsibility
 Everything in moderation
 Chew with your mouth open
$

Lastly, you can compare repository revisions even when you don't have a working copy on your local machine, just by including the appropriate URL on the command line:

$ svn diff -c 5 http://svn.example.com/repos/example/trunk/text/rules.txt
…
$

Generating a List of Historical Changes

To find information about the history of a file or directory, use the svn log command. svn log will provide you with a record of who made changes to a file or directory, at what revision it changed, the time and date of that revision, and—if it was provided—the log message that accompanied the commit:

$ svn log
------------------------------------------------------------------------
r3 | sally | 2008-05-15 23:09:28 -0500 (Thu, 15 May 2008) | 1 line

Added include lines and corrected # of cheese slices.
------------------------------------------------------------------------
r2 | harry | 2008-05-14 18:43:15 -0500 (Wed, 14 May 2008) | 1 line

Added main() methods.
------------------------------------------------------------------------
r1 | sally | 2008-05-10 19:50:31 -0500 (Sat, 10 May 2008) | 1 line

Initial import
------------------------------------------------------------------------

Note that the log messages are printed in reverse chronological order by default. If you wish to see a different range of revisions in a particular order or just a single revision, pass the --revision (-r) option:

Table 2.1. Common log requests

CommandDescription
svn log -r 5:19Display logs for revisions 5 through 19 in chronological order
svn log -r 19:5Display logs for revisions 5 through 19 in reverse chronological order
svn log -r 8Display logs for revision 8 only


You can also examine the log history of a single file or directory. For example:

$ svn log foo.c
…
$ svn log http://foo.com/svn/trunk/code/foo.c
…

These will display log messages only for those revisions in which the working file (or URL) changed.

If you want even more information about a file or directory, svn log also takes a --verbose (-v) option. Because Subversion allows you to move and copy files and directories, it is important to be able to track path changes in the filesystem. So, in verbose mode, svn log will include a list of changed paths in a revision in its output:

$ svn log -r 8 -v
------------------------------------------------------------------------
r8 | sally | 2008-05-21 13:19:25 -0500 (Wed, 21 May 2008) | 1 line
Changed paths:
   M /trunk/code/foo.c
   M /trunk/code/bar.h
   A /trunk/code/doc/README

Frozzled the sub-space winch.

------------------------------------------------------------------------

svn log also takes a --quiet (-q) option, which suppresses the body of the log message. When combined with --verbose (-v), it gives just the names of the changed files.

As of Subversion 1.7, users of the Subversion command-line can also take advantage of a special output mode for svn log which integrates a difference report such as is generated by the svn diff command we introduced earlier. When you invoke svn log with the --diff option, Subversion will append to each revision log chunk in the log report a diff-style difference report. This is a very convenient way to see both the high-level, semantic changes and the line-based modifications of a revision all at the same time!

Browsing the Repository

Using svn cat and svn list, you can view various revisions of files and directories without changing the working revision of your working copy. In fact, you don't even need a working copy to use either one.

Displaying file contents

If you want to examine an earlier version of a file and not necessarily the differences between two files, you can use svn cat:

$ svn cat -r 2 rules.txt
Be kind to others
Freedom = Chocolate Ice Cream
Everything in moderation
Chew with your mouth open
$

You can also redirect the output directly into a file:

$ svn cat -r 2 rules.txt > rules.txt.v2
$

Displaying line-by-line change attribution

Very similar to the svn cat command we discussed in the previous section is the svn annotate command. This command also displays the contents of a versioned file, but it does so using a tabular format. Each line of output shows not only a line of the file's content but also the username, the revision number and (optionally) the datestamp of the revision in which that line was last modified.

When used with a working copy file target, svn annotate will by default show line-by-line attribution of the file as it currently appears in the working copy.

$ svn annotate rules.txt
     1      harry Be kind to others
     3      sally Freedom = Responsibility
     1      harry Everything in moderation
     -          - Chew with your mouth closed
     -          - Listen when others are speaking

Notice that for some lines, there is no attribution provided. In this case, that's because those lines have been modified in the working copy's version of the file. In this way, svn annotate becomes another way for you to see which lines in the file you have changed. You can use the BASE revision keyword (see the section called “Revision Keywords”) to instead see the unmodified form of the file as it resides in your working copy.

$ svn annotate rules.txt@BASE
     1      harry Be kind to others
     3      sally Freedom = Responsibility
     1      harry Everything in moderation
     1      harry Chew with your mouth open

The --verbose (-v) option causes svn annotate to also include on each line the datestamp associated with that line's reported revision number. (This adds a significant amount of width to each line of ouput, so we'll skip the demonstration here.)

As with svn cat, you can also ask svn annotate to display previous versions of the file. This can be a handy trick when, after finding out who most recently modified a particular line of interest in the file, you then wish to see who modified the same line prior to that.

$ svn blame rules.txt -r 2
     1      harry Be kind to others
     1      harry Freedom = Chocolate Ice Cream
     1      harry Everything in moderation
     1      harry Chew with your mouth open

Unlike the svn cat command, the functionality of svn annotate is tied heavily to the idea of lines of text in a human-readable file. As such, if you attempt to run the command on a file that Subversion has determined is not human-readable (per the file's svn:mime-type property—see the section called “File Content Type” for details), you'll get an error message.

$ svn annotate images/logo.png
Skipping binary file: 'images/logo.png'
$

As revealed in the error message, you can use the --force option to disable this check and proceed with the annotation as if the file's contents are, in fact, human-readable and line-based. Naturally, if you force Subversion to try to perform line-based annotation on a nontextual file, you'll get what you asked for: a screenful of nonsense.

$ svn annotate images/logo.png --force
     6      harry \211PNG
     6      harry ^Z
     6      harry 
     7      harry \274\361\MI\300\365\353^X\300…
[Tip]Tip

Depending on your mood at the time you execute this command and your reasons for doing so, you may find yourself typing svn blame … or svn praise … instead of using the canonical svn annotate command form. That's okay—the Subversion developers anticipated as much, so those particular command aliases work, too!

Finally, as with many of Subversion's informational commands, you can also reference files in your svn annotate command invocations by their repository URLs, allowing access to this information even when you don't have ready access to a working copy.

Listing versioned directories

The svn list command shows you what files are in a repository directory without actually downloading the files to your local machine:

$ svn list http://svn.example.com/repo/project
README
branches/
tags/
trunk/

If you want a more detailed listing, pass the --verbose (-v) flag to get output like this:

$ svn list -v http://svn.example.com/repo/project
  23351 sally                 Feb 05 13:26 ./
  20620 harry            1084 Jul 13  2006 README
  23339 harry                 Feb 04 01:40 branches/
  23198 harry                 Jan 23 17:17 tags/
  23351 sally                 Feb 05 13:26 trunk/

The columns tell you the revision at which the file or directory was last modified, the user who modified it, the size if it is a file, the date it was last modified, and the item's name.

[Warning]Warning

The svn list command with no arguments defaults to the repository URL of the current working directory, not the local working copy directory. After all, if you want a listing of your local directory, you could use just plain ls (or any reasonable non-Unixy equivalent).

Fetching Older Repository Snapshots

In addition to all of the previous commands, you can use the --revision (-r) option with svn update to take an entire working copy back in time:[8]

# Make the current directory look like it did in r1729.
$ svn update -r 1729
Updating '.':
…
$
[Tip]Tip

Many Subversion newcomers attempt to use the preceding svn update example to undo committed changes, but this won't work as you can't commit changes that you obtain from backdating a working copy if the changed files have newer revisions. See the section called “Resurrecting Deleted Items” for a description of how to undo a commit.

If you'd prefer to create a whole new working copy from an older snapshot, you can do so by modifying the typical svn checkout command. As with svn update, you can provide the --revision (-r) option. But for reasons that we cover in the section called “Peg and Operative Revisions”, you might instead want to specify the target revision as part of Subversion's expanded URL syntax.

# Checkout the trunk from r1729.
$ svn checkout http://svn.example.com/svn/repo/trunk@1729 trunk-1729
…
# Checkout the current trunk as it looked in r1729.
$ svn checkout http://svn.example.com/svn/repo/trunk -r 1729 trunk-1729
…
$

Lastly, if you're building a release and wish to bundle up your versioned files and directories, you can use svn export to create a local copy of all or part of your repository without any .svn administrative directories included. The basic syntax of this subcommand is identical to that of svn checkout:

# Export the trunk from the latest revision.
$ svn export http://svn.example.com/svn/repo/trunk trunk-export
…
# Export the trunk from r1729.
$ svn export http://svn.example.com/svn/repo/trunk@1729 trunk-1729
…
# Export the current trunk as it looked in r1729. 
$ svn export http://svn.example.com/svn/repo/trunk -r 1729 trunk-1729
…
$

Sometimes You Just Need to Clean Up

Now that we've covered the day-to-day tasks that you'll frequently use Subversion for, we'll review a few administrative tasks relating to your working copy.

Disposing of a Working Copy

Subversion doesn't track either the state or the existence of working copies on the server, so there's no server overhead to keeping working copies around. Likewise, there's no need to let the server know that you're going to delete a working copy.

If you're likely to use a working copy again, there's nothing wrong with just leaving it on disk until you're ready to use it again, at which point all it takes is an svn update to bring it up to date and ready for use.

However, if you're definitely not going to use a working copy again, you can safely delete the entire thing using whatever directory removal capabilities your operating system offers. We recommend that before you do so you run svn status and review any files listed in its output that are prefixed with a ? to make certain that they're not of importance.

Recovering from an Interruption

When Subversion modifies your working copy—either your files or its own administrative state—it tries to do so as safely as possible. Before changing the working copy, Subversion logs its intentions in a private to-do list, of sorts. Next, it performs those actions to effect the desired change, holding a lock on the relevant part of the working copy while it works. This prevents other Subversion clients from accessing the working copy mid-change. Finally, Subversion releases its lock and cleans up its private to-do list. Architecturally, this is similar to a journaled filesystem. If a Subversion operation is interrupted (e.g, if the process is killed or if the machine crashes), the private to-do list remains on disk. This allows Subversion to return to that list later to complete any unfinished operations and return your working copy to a consistent state.

This is exactly what svn cleanup does: it searches your working copy and runs any leftover to-do items, removing working copy locks as it completes those operations. If Subversion ever tells you that some part of your working copy is locked, run svn cleanup to remedy the problem. The svn status command will inform you about administrative locks in the working copy, too, by displaying an L next to those locked paths:

$ svn status
  L     somedir
M       somedir/foo.c
$ svn cleanup
$ svn status
M       somedir/foo.c

Don't confuse these working copy administrative locks with the user-managed locks that Subversion users create when using the lock-modify-unlock model of concurrent version control; see the sidebar The Three Meanings of Lock for clarification.

Dealing with Structural Conflicts

So far, we have only talked about conflicts at the level of file content. When you and your collaborators make overlapping changes within the same file, Subversion forces you to merge those changes before you can commit.[9]

But what happens if your collaborators move or delete a file that you are still working on? Maybe there was a miscommunication, and one person thinks the file should be deleted, while another person still wants to commit changes to the file. Or maybe your collaborators did some refactoring, renaming files and moving around directories in the process. If you were still working on these files, those modifications may need to be applied to the files at their new location. Such conflicts manifest themselves at the directory tree structure level rather than at the file content level, and are known as tree conflicts.

As with textual conflicts, tree conflicts prevent a commit from being made from the conflicted state, giving the user the opportunity to examine the state of the working copy for potential problems arising from the tree conflict, and resolving any such problems before committing.

An Example Tree Conflict

Suppose a software project you were working on currently looked like this:

$ svn list -Rv svn://svn.example.com/trunk/
     13 harry                 Sep 06 10:34 ./
     13 harry              27 Sep 06 10:34 COPYING
     13 harry              41 Sep 06 10:32 Makefile
     13 harry              53 Sep 06 10:34 README
     13 harry                 Sep 06 10:32 code/
     13 harry              54 Sep 06 10:32 code/bar.c
     13 harry             130 Sep 06 10:32 code/foo.c
$

Later, in revision 14, your collaborator Harry renames the file bar.c to baz.c. Unfortunately, you don't realize this yet. As it turns out, you are busy in your working copy composing a different set of changes, some of which also involve modifications to bar.c:

$ svn diff
Index: code/foo.c
===================================================================
--- code/foo.c	(revision 13)
+++ code/foo.c	(working copy)
@@ -3,5 +3,5 @@
 int main(int argc, char *argv[])
 {
     printf("I don't like being moved around!\n%s", bar());
-    return 0;
+    return 1;
 }
Index: code/bar.c
===================================================================
--- code/bar.c	(revision 13)
+++ code/bar.c	(working copy)
@@ -1,4 +1,4 @@
 const char *bar(void)
 {
-    return "Me neither!\n";
+    return "Well, I do like being moved around!\n";
 }
$

You first realize that someone else has changed bar.c when your own commit attempt fails:

$ svn commit -m "Small fixes"
Sending        code/bar.c
svn: E155011: Commit failed (details follow):
svn: E155011: File '/home/svn/project/code/bar.c' is out of date
svn: E160013: File not found: transaction '14-e', path '/code/bar.c'
$

At this point, you need to run svn update. Besides bringing our working copy up to date so that you can see Harry's changes, this also flags a tree conflict so you have the opportunity to evaluate and properly resolve it.

$ svn update
Updating '.':
   C code/bar.c
A    code/baz.c
U    Makefile
Updated to revision 14.
Summary of conflicts:
  Tree conflicts: 1
$

In its output, svn update signifies tree conflicts using a capital C in the fourth output column. svn status reveals additional details of the conflict:

$ svn status
M       code/foo.c
A  +  C code/bar.c
      >   local edit, incoming delete upon update
Summary of conflicts:
  Tree conflicts: 1
$

Note how bar.c is automatically scheduled for re-addition in your working copy, which simplifies things in case you want to keep the file.

Because a move in Subversion is implemented as a copy operation followed by a delete operation, and these two operations cannot be easily related to one another during an update, all Subversion can warn you about is an incoming delete operation on a locally modified file. This delete operation may be part of a move, or it could be a genuine delete operation. Determining exactly what semantic change was made to the repository is important—you want to know just how your own edits fit into the overall trajectory of the project. So read log messages, talk to your collaborators, study the line-based differences—do whatever you must do—to determine your best course of action.

In this case, Harry's commit log message tells you what you need to know.

$ svn log -r14 ^/trunk
------------------------------------------------------------------------
r14 | harry | 2011-09-06 10:38:17 -0400 (Tue, 06 Sep 2011) | 1 line
Changed paths:
   M /Makefile
   D /code/bar.c
   A /code/baz.c (from /code/bar.c:13)

Rename bar.c to baz.c, and adjust Makefile accordingly.
------------------------------------------------------------------------
$

svn info shows the URLs of the items involved in the conflict. The left URL shows the source of the local side of the conflict, while the right URL shows the source of the incoming side of the conflict. These URLs indicate where you should start searching the repository's history for the change which conflicts with your local change.

$ svn info code/bar.c
Path: code/bar.c
Name: bar.c
URL: http://svn.example.com/svn/repo/trunk/code/bar.c
…
Tree conflict: local edit, incoming delete upon update
  Source  left: (file) ^/trunk/code/bar.c@4
  Source right: (none) ^/trunk/code/bar.c@5

$

bar.c is now said to be the victim of a tree conflict. It cannot be committed until the conflict is resolved:

$ svn commit -m "Small fixes" 
svn: E155015: Commit failed (details follow):
svn: E155015: Aborting commit: '/home/svn/project/code/bar.c' remains in confl
ict
$

To resolve this conflict, you must either agree or disagree with the move that Harry made.

If you agree with the move, your bar.c is superfluous. You'll want to delete it and mark the tree conflict as resolved. But wait: you made changes to that file! Before deleting bar.c, you need to decide if the changes you made to it need to be applied elsewhere, for example to the new baz.c file where all of bar.c's code now lives. Let's assume that your changes do need to follow the move. Subversion isn't smart enough to do this work for you[10], so you need to migrate your changes manually.

In our example, you could manually re-make your change to bar.c pretty easily—it was, after all, a single-line change. That's not always the case, though, so we'll show a more scalable approach. We'll first use svn diff to create a patch file. Then we'll edit the headers of that patch file to point to the new name of our renamed file. Finally, we re-apply the modified patch to our working copy.

$ svn diff code/bar.c > PATCHFILE
$ cat PATCHFILE
Index: code/bar.c
===================================================================
--- code/bar.c	(working copy)
+++ code/bar.c	(working copy)
@@ -1,4 +1,4 @@
 const char *bar(void)
 {
-    return "Me neither!\n";
+    return "Well, I do like being moved around!\n";
 }
$ ### Edit PATCHFILE to refer to code/baz.c instead of code/bar.c
$ cat PATCHFILE
Index: code/baz.c
===================================================================
--- code/baz.c	(working copy)
+++ code/baz.c	(working copy)
@@ -1,4 +1,4 @@
 const char *bar(void)
 {
-    return "Me neither!\n";
+    return "Well, I do like being moved around!\n";
 }
$ svn patch PATCHFILE
U         code/baz.c
$

Now that the changes you originally made to bar.c have been successfully reproduced in baz.c, you can delete bar.c and resolve the conflict, instructing the resolution logic to accept what is currently in the working copy as the desired result.

$ svn delete --force code/bar.c
D         code/bar.c
$ svn resolve --accept=working code/bar.c
Resolved conflicted state of 'code/bar.c'
$ svn status
M       code/foo.c
M       code/baz.c
$ svn diff
Index: code/foo.c
===================================================================
--- code/foo.c  (revision 14)
+++ code/foo.c  (working copy)
@@ -3,5 +3,5 @@
 int main(int argc, char *argv[])
 {
     printf("I don't like being moved around!\n%s", bar());
-    return 0;
+    return 1;
 }
Index: code/baz.c
===================================================================
--- code/baz.c  (revision 14)
+++ code/baz.c  (working copy)
@@ -1,4 +1,4 @@
 const char *bar(void)
 {
-    return "Me neither!\n";
+    return "Well, I do like being moved around!\n";
 }
$

But what if you do not agree with the move? Well, in that case, you can delete baz.c instead, after making sure any changes made to it after it was renamed are either preserved or not worth keeping. (Do not forget to also revert the changes Harry made to Makefile.) Since bar.c is already scheduled for re-addition, there is nothing else left to do, and the conflict can be marked resolved:

$ svn delete --force code/baz.c
D         code/baz.c
$ svn resolve --accept=working code/bar.c
Resolved conflicted state of 'code/bar.c'
$ svn status
M       code/foo.c
A  +    code/bar.c
D       code/baz.c
M       Makefile
$ svn diff
Index: code/foo.c
===================================================================
--- code/foo.c	(revision 14)
+++ code/foo.c	(working copy)
@@ -3,5 +3,5 @@
 int main(int argc, char *argv[])
 {
     printf("I don't like being moved around!\n%s", bar());
-    return 0;
+    return 1;
 }
Index: code/bar.c
===================================================================
--- code/bar.c	(revision 14)
+++ code/bar.c	(working copy)
@@ -1,4 +1,4 @@
 const char *bar(void)
 {
-    return "Me neither!\n";
+    return "Well, I do like being moved around!\n";
 }
Index: code/baz.c
===================================================================
--- code/baz.c	(revision 14)
+++ code/baz.c	(working copy)
@@ -1,4 +0,0 @@
-const char *bar(void)
-{
-    return "Me neither!\n";
-}
Index: Makefile
===================================================================
--- Makefile	(revision 14)
+++ Makefile	(working copy)
@@ -1,2 +1,2 @@
 foo: 
-	$(CC) -o $@ code/foo.c code/baz.c
+	$(CC) -o $@ code/foo.c code/bar.c

You've now resolved your first tree conflict! You can commit your changes and tell Harry during tea break about all the extra work he caused for you.

Summary

Now we've covered most of the Subversion client commands. Notable exceptions are those dealing with branching and merging (see Chapter 4, Branching and Merging) and properties (see the section called “Properties”). However, you may want to take a moment to skim through Chapter 9, Subversion Complete Reference to get an idea of all the different commands that Subversion has—and how you can use them to make your work easier.




[6] Of course, nothing is ever totally deleted from the repository—just from its HEAD revision. You may continue to access the deleted item in previous revisions. Should you desire to resurrect the item so that it is again present in HEAD, see the section called “Resurrecting Deleted Items”.

[7] And if you ask them for it, they may very well ride you out of town on a rail.

[8] See? We told you that Subversion was a time machine.

[9] Well, you could mark files containing conflict markers as resolved and commit them, if you really wanted to. But this is rarely done in practice.

[10] In some cases, Subversion 1.5 and 1.6 would actually handle this for you, but this somewhat hit-or-miss functionality was removed in Subversion 1.7.

Chapter 3. Advanced Topics

If you've been reading this book chapter by chapter, from start to finish, you should by now have acquired enough knowledge to use the Subversion client to perform the most common version control operations. You understand how to check out a working copy from a Subversion repository. You are comfortable with submitting and receiving changes using the svn commit and svn update operations. You've probably even developed a reflex that causes you to run the svn status command almost unconsciously. For all intents and purposes, you are ready to use Subversion in a typical environment.

But the Subversion feature set doesn't stop at common version control operations. It has other bits of functionality besides just communicating file and directory changes to and from a central repository.

This chapter highlights some of Subversion's features that, while important, may not be part of the typical user's daily routine. It assumes that you are familiar with Subversion's basic file and directory versioning capabilities. If you aren't, you'll want to first read Chapter 1, Fundamental Concepts and Chapter 2, Basic Usage. Once you've mastered those basics and consumed this chapter, you'll be a Subversion power user!

Revision Specifiers

As we described in the section called “Revisions”, revision numbers in Subversion are pretty straightforward—integers that keep getting larger as you commit more changes to your versioned data. Still, it doesn't take long before you can no longer remember exactly what happened in each and every revision. Fortunately, the typical Subversion workflow doesn't often demand that you supply arbitrary revisions to the Subversion operations you perform. For operations that do require a revision specifier, you generally supply a revision number that you saw in a commit email, in the output of some other Subversion operation, or in some other context that would give meaning to that particular number.

[Note]Note

Referring to revision numbers with an r prefix (r314, for example) is an established practice in Subversion communities, and is both supported and encouraged by many Subversion-related tools. In most places where you would specify a bare revision number on the command line, you may also use the rNNN syntax.

But occasionally, you need to pinpoint a moment in time for which you don't already have a revision number memorized or handy. So besides the integer revision numbers, svn allows as input some additional forms of revision specifiers: revision keywords and revision dates.

[Note]Note

The various forms of Subversion revision specifiers can be mixed and matched when used to specify revision ranges. For example, you can use -r REV1:REV2 where REV1 is a revision keyword and REV2 is a revision number, or where REV1 is a date and REV2 is a revision keyword, and so on. The individual revision specifiers are independently evaluated, so you can put whatever you want on the opposite sides of that colon.

Revision Keywords

The Subversion client understands a number of revision keywords. These keywords can be used instead of integer arguments to the --revision (-r) option, and are resolved into specific revision numbers by Subversion:

HEAD

The latest (or youngest) revision in the repository.

BASE

The revision number of an item in a working copy. If the item has been locally modified, this refers to the way the item appears without those local modifications.

COMMITTED

The most recent revision prior to, or equal to, BASE, in which an item changed.

PREV

The revision immediately before the last revision in which an item changed. Technically, this boils down to COMMITTED-1.

As can be derived from their descriptions, the PREV, BASE, and COMMITTED revision keywords are used only when referring to a working copy path—they don't apply to repository URLs. HEAD, on the other hand, can be used in conjunction with both of these path types.

Here are some examples of revision keywords in action:

$ svn diff -r PREV:COMMITTED foo.c
# shows the last change committed to foo.c

$ svn log -r HEAD
# shows log message for the latest repository commit

$ svn diff -r HEAD
# compares your working copy (with all of its local changes) to the
# latest version of that tree in the repository

$ svn diff -r BASE:HEAD foo.c
# compares the unmodified version of foo.c with the latest version of
# foo.c in the repository

$ svn log -r BASE:HEAD
# shows all commit logs for the current versioned directory since you
# last updated

$ svn update -r PREV foo.c
# rewinds the last change on foo.c, decreasing foo.c's working revision

$ svn diff -r BASE:14 foo.c
# compares the unmodified version of foo.c with the way foo.c looked
# in revision 14

Revision Dates

Revision numbers reveal nothing about the world outside the version control system, but sometimes you need to correlate a moment in real time with a moment in version history. To facilitate this, the --revision (-r) option can also accept as input date specifiers wrapped in curly braces ({ and }). Subversion accepts the standard ISO-8601 date and time formats, plus a few others. Here are some examples.

$ svn update -r {2006-02-17}
$ svn update -r {15:30}
$ svn update -r {15:30:00.200000}
$ svn update -r {"2006-02-17 15:30"}
$ svn update -r {"2006-02-17 15:30 +0230"}
$ svn update -r {2006-02-17T15:30}
$ svn update -r {2006-02-17T15:30Z}
$ svn update -r {2006-02-17T15:30-04:00}
$ svn update -r {20060217T1530}
$ svn update -r {20060217T1530Z}
$ svn update -r {20060217T1530-0500}
…
[Note]Note

Keep in mind that most shells will require you to, at a minimum, quote or otherwise escape any spaces that are included as part of revision date specifiers. Certain shells may also take issue with the unescaped use of curly braces, too. Consult your shell's documentation for the requirements specific to your environment.

When you specify a date, Subversion resolves that date to the most recent revision of the repository as of that date, and then continues to operate against that resolved revision number:

$ svn log -r {2006-11-28}
------------------------------------------------------------------------
r12 | ira | 2006-11-27 12:31:51 -0600 (Mon, 27 Nov 2006) | 6 lines
…

You can also use a range of dates. Subversion will find all revisions between both dates, inclusive:

$ svn log -r {2006-11-20}:{2006-11-29}
…
[Warning]Warning

Since the timestamp of a revision is stored as an unversioned, modifiable property of the revision (see the section called “Properties”), revision timestamps can be changed to represent complete falsifications of true chronology, or even removed altogether. Subversion's ability to correctly convert revision dates into real revision numbers depends on revision datestamps maintaining a sequential ordering—the younger the revision, the younger its timestamp. If this ordering isn't maintained, you will likely find that trying to use dates to specify revision ranges in your repository doesn't always return the data you might have expected.

Peg and Operative Revisions

We copy, move, rename, and completely replace files and directories on our computers all the time. And your version control system shouldn't get in the way of your doing these things with your version-controlled files and directories, either. Subversion's file management support is quite liberating, affording almost as much flexibility for versioned files as you'd expect when manipulating your unversioned ones. But that flexibility means that across the lifetime of your repository, a given versioned object might have many paths, and a given path might represent several entirely different versioned objects. This introduces a certain level of complexity to your interactions with those paths and objects.

Subversion is pretty smart about noticing when an object's version history includes such changes of address. For example, if you ask for the revision history log of a particular file that was renamed last week, Subversion happily provides all those logs—the revision in which the rename itself happened, plus the logs of relevant revisions both before and after that rename. So, most of the time, you don't even have to think about such things. But occasionally, Subversion needs your help to clear up ambiguities.

The simplest example of this occurs when a directory or file is deleted from version control, and then a new directory or file is created with the same name and added to version control. The thing you deleted and the thing you later added aren't the same thing. They merely happen to have had the same path—/trunk/object, for example. What, then, does it mean to ask Subversion about the history of /trunk/object? Are you asking about the thing currently at that location, or the old thing you deleted from that location? Are you asking about the operations that have happened to all the objects that have ever lived at that path? Subversion needs a hint about what you really want.

And thanks to moves, versioned object history can get far more twisted than even that. For example, you might have a directory named concept, containing some nascent software project you've been toying with. Eventually, though, that project matures to the point that the idea seems to actually have some wings, so you do the unthinkable and decide to give the project a name.[11] Let's say you called your software Frabnaggilywort. At this point, it makes sense to rename the directory to reflect the project's new name, so concept is renamed to frabnaggilywort. Life goes on, Frabnaggilywort releases a 1.0 version and is downloaded and used daily by hordes of people aiming to improve their lives.

It's a nice story, really, but it doesn't end there. Entrepreneur that you are, you've already got another think in the tank. So you make a new directory, concept, and the cycle begins again. In fact, the cycle begins again many times over the years, each time starting with that old concept directory, then sometimes seeing that directory renamed as the idea cures, sometimes seeing it deleted when you scrap the idea. Or, to get really sick, maybe you rename concept to something else for a while, but later rename the thing back to concept for some reason.

In scenarios like these, attempting to instruct Subversion to work with these reused paths can be a little like instructing a motorist in Chicago's West Suburbs to drive east down Roosevelt Road and turn left onto Main Street. In a mere 20 minutes, you can cross Main Street in Wheaton, Glen Ellyn, and Lombard. And no, they aren't the same street. Our motorist—and our Subversion—need a little more detail to do the right thing.

Fortunately, Subversion allows you to tell it exactly which Main Street you meant. The mechanism used is called a peg revision, and you provide these to Subversion for the sole purpose of identifying unique lines of history. Because at most one versioned object may occupy a path at any given time—or, more precisely, in any one revision—the combination of a path and a peg revision is all that is needed to unambiguously identify a specific line of history. Peg revisions are specified to the Subversion command-line client using at syntax, so called because the syntax involves appending an at sign (@) and the peg revision to the end of the path with which the revision is associated.

But what of the --revision (-r) of which we've spoken so much in this book? That revision (or set of revisions) is called the operative revision (or operative revision range). Once a particular line of history has been identified using a path and peg revision, Subversion performs the requested operation using the operative revision(s). To map this to our Chicagoland streets analogy, if we are told to go to 606 N. Main Street in Wheaton,[12] we can think of Main Street as our path and Wheaton as our peg revision. These two pieces of information identify a unique path that can be traveled (north or south on Main Street), and they keep us from traveling up and down the wrong Main Street in search of our destination. Now we throw in 606 N. as our operative revision of sorts, and we know exactly where to go.

Say that long ago we created our repository, and in revision 1 we added our first concept directory, plus an IDEA file in that directory talking about the concept. After several revisions in which real code was added and tweaked, we, in revision 20, renamed this directory to frabnaggilywort. By revision 27, we had a new concept, a new concept directory to hold it, and a new IDEA file to describe it. And then five years and thousands of revisions flew by, just like they would in any good romance story.

Now, years later, we wonder what the IDEA file looked like back in revision 1. But Subversion needs to know whether we are asking about how the current file looked back in revision 1, or whether we are asking for the contents of whatever file lived at concepts/IDEA in revision 1. Certainly those questions have different answers, and because of peg revisions, you can ask those questions. To find out how the current IDEA file looked in that old revision, you run:

$ svn cat -r 1 concept/IDEA 
svn: E195012: Unable to find repository location for 'concept/IDEA' in revision 1

Of course, in this example, the current IDEA file didn't exist yet in revision 1, so Subversion gives an error. The previous command is shorthand for a longer notation which explicitly lists a peg revision. The expanded notation is:

$ svn cat -r 1 concept/IDEA@BASE
svn: E195012: Unable to find repository location for 'concept/IDEA' in revision 1

And when executed, it has the expected results.

The perceptive reader is probably wondering at this point whether the peg revision syntax causes problems for working copy paths or URLs that actually have at signs in them. After all, how does svn know whether news@11 is the name of a directory in my tree or just a syntax for revision 11 of news? Thankfully, while svn will always assume the latter, there is a trivial workaround. You need only append an at sign to the end of the path, such as news@11@. svn cares only about the last at sign in the argument, and it is not considered illegal to omit a literal peg revision specifier after that at sign. This workaround even applies to paths that end in an at sign—you would use filename@@ to talk about a file named filename@.

Let's ask the other question, then—in revision 1, what were the contents of whatever file occupied the address concepts/IDEA at the time? We'll use an explicit peg revision to help us out.

$ svn cat concept/IDEA@1
The idea behind this project is to come up with a piece of software
that can frab a naggily wort.  Frabbing naggily worts is tricky
business, and doing it incorrectly can have serious ramifications, so
we need to employ over-the-top input validation and data verification
mechanisms.

Notice that we didn't provide an operative revision this time. That's because when no operative revision is specified, Subversion assumes a default operative revision that's the same as the peg revision.

As you can see, the output from our operation appears to be correct. The text even mentions frabbing naggily worts, so this is almost certainly the file that describes the software now called Frabnaggilywort. In fact, we can verify this using the combination of an explicit peg revision and explicit operative revision. We know that in HEAD, the Frabnaggilywort project is located in the frabnaggilywort directory. So we specify that we want to see how the line of history identified in HEAD as the path frabnaggilywort/IDEA looked in revision 1.

$ svn cat -r 1 frabnaggilywort/IDEA@HEAD
The idea behind this project is to come up with a piece of software
that can frab a naggily wort.  Frabbing naggily worts is tricky
business, and doing it incorrectly can have serious ramifications, so
we need to employ over-the-top input validation and data verification
mechanisms.

And the peg and operative revisions need not be so trivial, either. For example, say frabnaggilywort had been deleted from HEAD, but we know it existed in revision 20, and we want to see the diffs for its IDEA file between revisions 4 and 10. We can use peg revision 20 in conjunction with the URL that would have held Frabnaggilywort's IDEA file in revision 20, and then use 4 and 10 as our operative revision range.

$ svn diff -r 4:10 http://svn.red-bean.com/projects/frabnaggilywort/IDEA@20
Index: frabnaggilywort/IDEA
===================================================================
--- frabnaggilywort/IDEA	(revision 4)
+++ frabnaggilywort/IDEA	(revision 10)
@@ -1,5 +1,5 @@
-The idea behind this project is to come up with a piece of software
-that can frab a naggily wort.  Frabbing naggily worts is tricky
-business, and doing it incorrectly can have serious ramifications, so
-we need to employ over-the-top input validation and data verification
-mechanisms.
+The idea behind this project is to come up with a piece of
+client-server software that can remotely frab a naggily wort.
+Frabbing naggily worts is tricky business, and doing it incorrectly
+can have serious ramifications, so we need to employ over-the-top
+input validation and data verification mechanisms.

Fortunately, most folks aren't faced with such complex situations. But when you are, remember that peg revisions are that extra hint Subversion needs to clear up ambiguity.

Properties

We've already covered in detail how Subversion stores and retrieves various versions of files and directories in its repository. Whole chapters have been devoted to this most fundamental piece of functionality provided by the tool. And if the versioning support stopped there, Subversion would still be complete from a version control perspective.

But it doesn't stop there.

In addition to versioning your directories and files, Subversion provides interfaces for adding, modifying, and removing versioned metadata on each of your versioned directories and files. We refer to this metadata as properties, and they can be thought of as two-column tables that map property names to arbitrary values attached to each item in your working copy. Generally speaking, the names and values of the properties can be whatever you want them to be, with the constraint that the names must contain only ASCII characters. And the best part about these properties is that they, too, are versioned, just like the textual contents of your files. You can modify, commit, and revert property changes as easily as you can file content changes. And the sending and receiving of property changes occurs as part of your typical commit and update operations—you don't have to change your basic processes to accommodate them.

[Note]Note

Subversion has reserved the set of properties whose names begin with svn: as its own. While there are only a handful of such properties in use today, you should avoid creating custom properties for your own needs whose names begin with this prefix. Otherwise, you run the risk that a future release of Subversion will grow support for a feature or behavior driven by a property of the same name but with perhaps an entirely different interpretation.

Properties show up elsewhere in Subversion, too. Just as files and directories may have arbitrary property names and values attached to them, each revision as a whole may have arbitrary properties attached to it. The same constraints apply—human-readable names and anything-you-want binary values. The main difference is that revision properties are not versioned. In other words, if you change the value of, or delete, a revision property, there's no way, within the scope of Subversion's functionality, to recover the previous value.

Subversion has no particular policy regarding the use of properties. It asks only that you do not use property names that begin with the prefix svn: as that's the namespace that it sets aside for its own use. And Subversion does, in fact, use properties—both the versioned and unversioned variety. Certain versioned properties have special meaning or effects when found on files and directories, or they house a particular bit of information about the revisions on which they are found. Certain revision properties are automatically attached to revisions by Subversion's commit process, and they carry information about the revision. Most of these properties are mentioned elsewhere in this or other chapters as part of the more general topics to which they are related. For an exhaustive list of Subversion's predefined properties, see the section called “Subversion Properties” in Chapter 9, Subversion Complete Reference.

[Note]Note

While Subversion automatically attaches properties (svn:date, svn:author, svn:log, and so on) to revisions, it does not presume thereafter the existence of those properties, and neither should you or the tools you use to interact with your repository. Revision properties can be deleted programmatically or via the client (if allowed by the repository hooks) without damaging Subversion's ability to function. So, when writing scripts which operate on your Subversion repository data, do not make the mistake of assuming that any particular revision property exists on a revision.

In this section, we will examine the utility—both to users of Subversion and to Subversion itself—of property support. You'll learn about the property-related svn subcommands and how property modifications affect your normal Subversion workflow.

Why Properties?

Just as Subversion uses properties to store extra information about the files, directories, and revisions that it contains, you might also find properties to be of similar use. You might find it useful to have a place close to your versioned data to hang custom metadata about that data.

Say you wish to design a web site that houses many digital photos and displays them with captions and a datestamp. Now, your set of photos is constantly changing, so you'd like to have as much of this site automated as possible. These photos can be quite large, so as is common with sites of this nature, you want to provide smaller thumbnail images to your site visitors.

Now, you can get this functionality using traditional files. That is, you can have your image123.jpg and an image123-thumbnail.jpg side by side in a directory. Or if you want to keep the filenames the same, you might have your thumbnails in a different directory, such as thumbnails/image123.jpg. You can also store your captions and datestamps in a similar fashion, again separated from the original image file. But the problem here is that your collection of files multiplies with each new photo added to the site.

Now consider the same web site deployed in a way that makes use of Subversion's file properties. Imagine having a single image file, image123.jpg, with properties set on that file that are named caption, datestamp, and even thumbnail. Now your working copy directory looks much more manageable—in fact, it looks to the casual browser like there are nothing but image files in it. But your automation scripts know better. They know that they can use svn (or better yet, they can use the Subversion language bindings—see the section called “Using the APIs”) to dig out the extra information that your site needs to display without having to read an index file or play path manipulation games.

[Note]Note

While Subversion places few restrictions on the names and values you use for properties, it has not been designed to optimally carry large property values or large sets of properties on a given file or directory. Subversion commonly holds all the property names and values associated with a single item in memory at the same time, which can cause detrimental performance or failed operations when extremely large property sets are used.

Custom revision properties are also frequently used. One common such use is a property whose value contains an issue tracker ID with which the revision is associated, perhaps because the change made in that revision fixes a bug filed in the tracker issue with that ID. Other uses include hanging more friendly names on the revision—it might be hard to remember that revision 1935 was a fully tested revision. But if there's, say, a test-results property on that revision with the value all passing, that's meaningful information to have. And Subversion allows you to easily do this via the --with-revprop option of the svn commit command:

$ svn commit -m "Fix up the last remaining known regression bug." \
             --with-revprop "test-results=all passing"
Sending        lib/crit_bits.c
Transmitting file data .
Committed revision 912.
$

Manipulating Properties

The svn program affords a few ways to add or modify file and directory properties. For properties with short, human-readable values, perhaps the simplest way to add a new property is to specify the property name and value on the command line of the svn propset subcommand:

$ svn propset copyright '(c) 2006 Red-Bean Software' calc/button.c
property 'copyright' set on 'calc/button.c'
$

But we've been touting the flexibility that Subversion offers for your property values. And if you are planning to have a multiline textual, or even binary, property value, you probably do not want to supply that value on the command line. So the svn propset subcommand takes a --file (-F) option for specifying the name of a file that contains the new property value.

$ svn propset license -F /path/to/LICENSE calc/button.c
property 'license' set on 'calc/button.c'
$

There are some restrictions on the names you can use for properties. A property name must start with a letter, a colon (:), or an underscore (_); after that, you can also use digits, hyphens (-), and periods (.).[13]

In addition to the propset command, the svn program supplies the propedit command. This command uses the configured editor program (see the section called “Config”) to add or modify properties. When you run the command, svn invokes your editor program on a temporary file that contains the current value of the property (or that is empty, if you are adding a new property). Then, you just modify that value in your editor program until it represents the new value you wish to store for the property, save the temporary file, and then exit the editor program. If Subversion detects that you've actually changed the existing value of the property, it will accept that as the new property value. If you exit your editor without making any changes, no property modification will occur:

$ svn propedit copyright calc/button.c  ### exit the editor without changes
No changes to property 'copyright' on 'calc/button.c'
$

We should note that, as with other svn subcommands, those related to properties can act on multiple paths at once. This enables you to modify properties on whole sets of files with a single command. For example, we could have done the following:

$ svn propset copyright '(c) 2006 Red-Bean Software' calc/*
property 'copyright' set on 'calc/Makefile'
property 'copyright' set on 'calc/button.c'
property 'copyright' set on 'calc/integer.c'
…
$

All of this property adding and editing isn't really very useful if you can't easily get the stored property value. So the svn program supplies two subcommands for displaying the names and values of properties stored on files and directories. The svn proplist command will list the names of properties that exist on a path. Once you know the names of the properties on the node, you can request their values individually using svn propget. This command will, given a property name and a path (or set of paths), print the value of the property to the standard output stream.

$ svn proplist calc/button.c
Properties on 'calc/button.c':
  copyright
  license
$ svn propget copyright calc/button.c
(c) 2006 Red-Bean Software

There's even a variation of the proplist command that will list both the name and the value for all of the properties. Simply supply the --verbose (-v) option.

$ svn proplist -v calc/button.c
Properties on 'calc/button.c':
  copyright
    (c) 2006 Red-Bean Software
  license
    ================================================================
    Copyright (c) 2006 Red-Bean Software.  All rights reserved.

    Redistribution and use in source and binary forms, with or without
    modification, are permitted provided that the following conditions 
    are met:

    1. Redistributions of source code must retain the above copyright
    notice, this list of conditions, and the recipe for Fitz's famous
    red-beans-and-rice.
    …

The last property-related subcommand is propdel. Since Subversion allows you to store properties with empty values, you can't remove a property altogether using svn propedit or svn propset. For example, this command will not yield the desired effect:

$ svn propset license "" calc/button.c
property 'license' set on 'calc/button.c'
$ svn proplist -v calc/button.c
Properties on 'calc/button.c':
  copyright
    (c) 2006 Red-Bean Software
  license
    
$

You need to use the propdel subcommand to delete properties altogether. The syntax is similar to the other property commands:

$ svn propdel license calc/button.c
property 'license' deleted from 'calc/button.c'.
$ svn proplist -v calc/button.c
Properties on 'calc/button.c':
  copyright
    (c) 2006 Red-Bean Software
$

Remember those unversioned revision properties? You can modify those, too, using the same svn subcommands that we just described. Simply add the --revprop command-line parameter and specify the revision whose property you wish to modify. Since revisions are global, you don't need to specify a target path to these property-related commands so long as you are positioned in a working copy of the repository whose revision property you wish to modify. Otherwise, you can simply provide the URL of any path in the repository of interest (including the repository's root URL). For example, you might want to replace the commit log message of an existing revision.[14] If your current working directory is part of a working copy of your repository, you can simply run the svn propset command with no target path:

$ svn propset svn:log "* button.c: Fix a compiler warning." -r11 --revprop
property 'svn:log' set on repository revision '11'
$

But even if you haven't checked out a working copy from that repository, you can still effect the property change by providing the repository's root URL:

$ svn propset svn:log "* button.c: Fix a compiler warning." -r11 --revprop \
              http://svn.example.com/repos/project
property 'svn:log' set on repository revision '11'
$

Note that the ability to modify these unversioned properties must be explicitly added by the repository administrator (see the section called “Commit Log Message Correction”). That's because the properties aren't versioned, so you run the risk of losing information if you aren't careful with your edits. The repository administrator can set up methods to protect against this loss, and by default, modification of unversioned properties is disabled.

[Tip]Tip

Users should, where possible, use svn propedit instead of svn propset. While the end result of the commands is identical, the former will allow them to see the current value of the property that they are about to change, which helps them to verify that they are, in fact, making the change they think they are making. This is especially true when modifying unversioned revision properties. Also, it is significantly easier to modify multiline property values in a text editor than at the command line.

Properties and the Subversion Workflow

Now that you are familiar with all of the property-related svn subcommands, let's see how property modifications affect the usual Subversion workflow. As we mentioned earlier, file and directory properties are versioned, just like your file contents. As a result, Subversion provides the same opportunities for merging—cleanly or with conflicts—someone else's modifications into your own.

As with file contents, your property changes are local modifications, made permanent only when you commit them to the repository with svn commit. Your property changes can be easily unmade, too—the svn revert command will restore your files and directories to their unedited states—contents, properties, and all. Also, you can receive interesting information about the state of your file and directory properties by using the svn status and svn diff commands.

$ svn status calc/button.c
 M      calc/button.c
$ svn diff calc/button.c
Property changes on: calc/button.c
___________________________________________________________________
Added: copyright
## -0,0 +1 ##
+(c) 2006 Red-Bean Software
$

Notice how the status subcommand displays M in the second column instead of the first. That is because we have modified the properties on calc/button.c, but not its textual contents. Had we changed both, we would have seen M in the first column, too. (We cover svn status in the section called “See an overview of your changes”).

You might also have noticed the nonstandard way that Subversion currently displays property differences. You can still use svn diff and redirect its output to create a usable patch file. The patch program will ignore property patches—as a rule, it ignores any noise it can't understand. This does, unfortunately, mean that to fully apply a patch generated by svn diff using patch, any property modifications will need to be applied by hand.

Subversion 1.7 improves this situation in two ways. First, its nonstandard display of property differences is at least machine-readable—an improvement over the display of properties in versions prior to 1.7. But Subversion 1.7 also introduces the svn patch subcommand, designed specifically to handle the additional information which svn diff's output can carry, applying those changes to the Subversion working copy. Of specific relevance to our topic, property differences present in patch files generated by svn diff in Subversion 1.7 or better can be automatically applied to a working copy by the svn patch command. For more about svn patch, see svn patch in Chapter 9, Subversion Complete Reference.

[Note]Note

There's one exception to how property changes are reported by svn diff: changes to Subversion's special svn:mergeinfo property—used to track information about merges which have been performed in your repository—are described in a more human-readable fashion. This is quite helpful to the humans who have to read those descriptions. But it also serves to cause patching programs (including svn patch) to skip those change descriptions as noise. This might sound like a bug, but it really isn't because this property is intended to be managed solely by the svn merge subcommand. For more about merge tracking, see Chapter 4, Branching and Merging.

Automatic Property Setting

Properties are a powerful feature of Subversion, acting as key components of many Subversion features discussed elsewhere in this and other chapters—textual diff and merge support, keyword substitution, newline translation, and so on. But to get the full benefit of properties, they must be set on the right files and directories. Unfortunately, that step can be easily forgotten in the routine of things, especially since failing to set a property doesn't usually result in an obvious error (at least compared to, say, failing to add a file to version control). To help your properties get applied to the places that need them, Subversion provides a couple of simple but useful features.

Whenever you introduce a file to version control using the svn add or svn import commands, Subversion tries to assist by setting some common file properties automatically. First, on operating systems whose filesystems support an execute permission bit, Subversion will automatically set the svn:executable property on newly added or imported files whose execute bit is enabled. (See the section called “File Executability” later in this chapter for more about this property.)

Second, Subversion tries to determine the file's MIME type. If you've configured a mime-types-files runtime configuration parameter, Subversion will try to find a MIME type mapping in that file for your file's extension. If it finds such a mapping, it will set your file's svn:mime-type property to the MIME type it found. If no mapping file is configured, or no mapping for your file's extension could be found, Subversion will fall back to heuristic algorithms to determine the file's MIME type. Depending on how it is built, Subversion 1.7 can make use of file scanning libraries[15] to detect a file's type based on its content. Failing all else, Subversion will employ its own very basic heuristic to determine whether the file contains nontextual content. If so, it automatically sets the svn:mime-type property on that file to application/octet-stream (the generic this is a collection of bytes MIME type). Of course, if Subversion guesses incorrectly, or if you wish to set the svn:mime-type property to something more precise—perhaps image/png or application/x-shockwave-flash—you can always remove or edit that property. (For more on Subversion's use of MIME types, see the section called “File Content Type” later in this chapter.)

[Note]Note

UTF-16 is commonly used to encode files whose semantic content is textual in nature, but the encoding itself makes heavy use of bytes which are outside the typical ASCII character byte range. As such, Subversion will tend to classify such files as binary files, much to the chagrin of users who desire line-based differencing and merging, keyword substitution, and other behaviors for those files.

Subversion also provides, via its runtime configuration system (see the section called “Runtime Configuration Area”), a more flexible automatic property setting feature that allows you to create mappings of filename patterns to property names and values. Once again, these mappings affect adds and imports, and can not only override the default MIME type decision made by Subversion during those operations, but can also set additional Subversion or custom properties, too. For example, you might create a mapping that says that anytime you add JPEG files—ones whose names match the pattern *.jpg—Subversion should automatically set the svn:mime-type property on those files to image/jpeg. Or perhaps any files that match *.cpp should have svn:eol-style set to native, and svn:keywords set to Id. Automatic property support is perhaps the handiest property-related tool in the Subversion toolbox. See the section called “Config” for more about configuring that support.

[Note]Note

Subversion administrators commonly ask if it is possible to configure, on the server side, a set of property definitions which all connecting clients will automatically consider when operating on working copies checked out from that server. Unfortunately, Subversion doesn't offer this feature. Administrators can use hook scripts to validate that the properties added to and modified on files and directories match the administrator's preferred policies, rejecting commits which are non-compliant in this fashion. (See the section called “Implementing Repository Hooks” for more about hook scripts.) But there's no way to automatically dictate those preferences to Subversion clients beforehand.

File Portability

Fortunately for Subversion users who routinely find themselves on different computers with different operating systems, Subversion's command-line program behaves almost identically on all those systems. If you know how to wield svn on one platform, you know how to wield it everywhere.

However, the same is not always true of other general classes of software or of the actual files you keep in Subversion. For example, on a Windows machine, the definition of a text file would be similar to that used on a Linux box, but with a key difference—the character sequences used to mark the ends of the lines of those files. There are other differences, too. Unix platforms have (and Subversion supports) symbolic links; Windows does not. Unix platforms use filesystem permission to determine executability; Windows uses filename extensions.

Because Subversion is in no position to unite the whole world in common definitions and implementations of all of these things, the best it can do is to try to help make your life simpler when you need to work with your versioned files and directories on multiple computers and operating systems. This section describes some of the ways Subversion does this.

File Content Type

Subversion joins the ranks of the many applications that recognize and make use of Multipurpose Internet Mail Extensions (MIME) content types. Besides being a general-purpose storage location for a file's content type, the value of the svn:mime-type file property determines some behavioral characteristics of Subversion itself.

For example, one of the benefits that Subversion typically provides is contextual, line-based merging of changes received from the server during an update into your working file. But for files containing nontextual data, there is often no concept of a line. So, for versioned files whose svn:mime-type property is set to a nontextual MIME type (generally, something that doesn't begin with text/, though there are exceptions), Subversion does not attempt to perform contextual merges during updates. Instead, any time you have locally modified a binary working copy file that is also being updated, your file is left untouched and Subversion creates two new files. One file has a .oldrev extension and contains the BASE revision of the file. The other file has a .newrev extension and contains the contents of the updated revision of the file. This behavior is really for the protection of the user against failed attempts at performing contextual merges on files that simply cannot be contextually merged.

Additionally, since the acts of displaying line-based differences and line-based change attribution are, rather obviously, dependent on there being a meaningful definition of line for a given file, files with nontextual MIME types will by default trigger errors when used as the targets of svn diff and svn annotate operations. This can be especially frustrating for users with XML files whose svn:mime-type property is set to something such as application/xml which is not unambiguously human-readable and as such is treated as nontextual by Subversion. Fortunately, those subcommands offer a --force option for forcing Subversion to attempt the operations in spite of the apparent non-human-readability of the files.

[Warning]Warning

The svn:mime-type property, when set to a value that does not indicate textual file contents, can cause some unexpected behaviors with respect to other properties. For example, since the idea of line endings (and therefore, line-ending conversion) makes no sense when applied to nontextual files, Subversion will prevent you from setting the svn:eol-style property on such files. This is obvious when attempted on a single file target—svn propset will error out. But it might not be as clear if you perform a recursive property set, where Subversion will silently skip over files that it deems unsuitable for a given property.

Subversion provides a number of mechanisms by which to automatically set the svn:mime-type property on a versioned file. See the section called “Automatic Property Setting” for details.

Also, if the svn:mime-type property is set, then the Subversion Apache module will use its value to populate the Content-type: HTTP header when responding to GET requests. This gives your web browser a crucial clue about how to display a file when you use it to peruse your Subversion repository's contents.

File Executability

On many operating systems, the ability to execute a file as a command is governed by the presence of an execute permission bit. This bit usually defaults to being disabled, and must be explicitly enabled by the user for each file that needs it. But it would be a monumental hassle to have to remember exactly which files in a freshly checked-out working copy were supposed to have their executable bits toggled on, and then to have to do that toggling. So, Subversion provides the svn:executable property as a way to specify that the executable bit for the file on which that property is set should be enabled, and Subversion honors that request when populating working copies with such files.

This property has no effect on filesystems that have no concept of an executable permission bit, such as FAT32 and NTFS.[17] Also, although it has no defined values, Subversion will force its value to * when setting this property. Finally, this property is valid only on files, not on directories.

End-of-Line Character Sequences

Unless otherwise noted using a versioned file's svn:mime-type property, Subversion assumes the file contains human-readable data. Generally speaking, Subversion uses this knowledge only to determine whether contextual difference reports for that file are possible. Otherwise, to Subversion, bytes are bytes.

This means that by default, Subversion doesn't pay any attention to the type of end-of-line (EOL) markers used in your files. Unfortunately, different operating systems have different conventions about which character sequences represent the end of a line of text in a file. For example, the usual line-ending token used by software on the Windows platform is a pair of ASCII control characters—a carriage return (CR) followed by a line feed (LF). Unix software, however, just uses the LF character to denote the end of a line.

Not all of the various tools on these operating systems understand files that contain line endings in a format that differs from the native line-ending style of the operating system on which they are running. So, typically, Unix programs treat the CR character present in Windows files as a regular character (usually rendered as ^M), and Windows programs combine all of the lines of a Unix file into one giant line because no carriage return-linefeed (or CRLF) character combination was found to denote the ends of the lines.

This sensitivity to foreign EOL markers can be frustrating for folks who share a file across different operating systems. For example, consider a source code file, and developers who edit this file on both Windows and Unix systems. If all the developers always use tools that preserve the line-ending style of the file, no problems occur.

But in practice, many common tools either fail to properly read a file with foreign EOL markers, or convert the file's line endings to the native style when the file is saved. If the former is true for a developer, he has to use an external conversion utility (such as dos2unix or its companion, unix2dos) to prepare the file for editing. The latter case requires no extra preparation. But both cases result in a file that differs from the original quite literally on every line! Prior to committing his changes, the user has two choices. Either he can use a conversion utility to restore the modified file to the same line-ending style that it was in before his edits were made, or he can simply commit the file—new EOL markers and all.

The result of scenarios like these include wasted time and unnecessary modifications to committed files. Wasted time is painful enough. But when commits change every line in a file, this complicates the job of determining which of those lines were changed in a nontrivial way. Where was that bug really fixed? On what line was a syntax error introduced?

The solution to this problem is the svn:eol-style property. When this property is set to a valid value, Subversion uses it to determine what special processing to perform on the file so that the file's line-ending style isn't flip-flopping with every commit that comes from a different operating system. The valid values are:

native

This causes the file to contain the EOL markers that are native to the operating system on which Subversion was run. In other words, if a user on a Windows machine checks out a working copy that contains a file with an svn:eol-style property set to native, that file will contain CRLF EOL markers. A Unix user checking out a working copy that contains the same file will see LF EOL markers in his copy of the file.

Note that Subversion will actually store the file in the repository using normalized LF EOL markers regardless of the operating system. This is basically transparent to the user, though.

CRLF

This causes the file to contain CRLF sequences for EOL markers, regardless of the operating system in use.

LF

This causes the file to contain LF characters for EOL markers, regardless of the operating system in use.

CR

This causes the file to contain CR characters for EOL markers, regardless of the operating system in use. This line-ending style is not very common.

Ignoring Unversioned Items

In any given working copy, there is a good chance that alongside all those versioned files and directories are other files and directories that are neither versioned nor intended to be. Text editors litter directories with backup files. Software compilers generate intermediate—or even final—files that you typically wouldn't bother to version. And users themselves drop various other files and directories wherever they see fit, often in version control working copies.

It's ludicrous to expect Subversion working copies to be somehow impervious to this kind of clutter and impurity. In fact, Subversion counts it as a feature that its working copies are just typical directories, just like unversioned trees. But these not-to-be-versioned files and directories can cause some annoyance for Subversion users. For example, because the svn add and svn import commands act recursively by default and don't know which files in a given tree you do and don't wish to version, it's easy to accidentally add stuff to version control that you didn't mean to. And because svn status reports, by default, every item of interest in a working copy—including unversioned files and directories—its output can get quite noisy where many of these things exist.

So Subversion provides two ways for telling it which files you would prefer that it simply disregard. One of the ways involves the use of Subversion's runtime configuration system (see the section called “Runtime Configuration Area”), and therefore applies to all the Subversion operations that make use of that runtime configuration—generally those performed on a particular computer or by a particular user of a computer. The other way makes use of Subversion's directory property support and is more tightly bound to the versioned tree itself, and therefore affects everyone who has a working copy of that tree. Both of the mechanisms use file patterns (strings of literal and special wildcard characters used to match against filenames) to decide which files to ignore.

The Subversion runtime configuration system provides an option, global-ignores, whose value is a whitespace-delimited collection of file patterns. The Subversion client checks these patterns against the names of the files that are candidates for addition to version control, as well as to unversioned files that the svn status command notices. If any file's name matches one of the patterns, Subversion will basically act as if the file didn't exist at all. This is really useful for the kinds of files that you almost never want to version, such as editor backup files such as Emacs' *~ and .*~ files.

When found on a versioned directory, the svn:ignore property is expected to contain a list of newline-delimited file patterns that Subversion should use to determine ignorable objects in that same directory. These patterns do not override those found in the global-ignores runtime configuration option, but are instead appended to that list. And it's worth noting again that, unlike the global-ignores option, the patterns found in the svn:ignore property apply only to the directory on which that property is set, and not to any of its subdirectories. The svn:ignore property is a good way to tell Subversion to ignore files that are likely to be present in every user's working copy of that directory, such as compiler output or—to use an example more appropriate to this book—the HTML, PDF, or PostScript files generated as the result of a conversion of some source DocBook XML files to a more legible output format.

[Note]Note

Subversion's support for ignorable file patterns extends only to the one-time process of adding unversioned files and directories to version control. Once an object is under Subversion's control, the ignore pattern mechanisms no longer apply to it. In other words, don't expect Subversion to avoid committing changes you've made to a versioned file simply because that file's name matches an ignore pattern—Subversion always notices all of its versioned objects.

The global list of ignore patterns tends to be more a matter of personal taste and ties more closely to a user's particular tool chain than to the details of any particular working copy's needs. So, the rest of this section will focus on the svn:ignore property and its uses.

Say you have the following output from svn status:

$ svn status calc
 M      calc/button.c
?       calc/calculator
?       calc/data.c
?       calc/debug_log
?       calc/debug_log.1
?       calc/debug_log.2.gz
?       calc/debug_log.3.gz

In this example, you have made some property modifications to button.c, but in your working copy, you also have some unversioned files: the latest calculator program that you've compiled from your source code, a source file named data.c, and a set of debugging output logfiles. Now, you know that your build system always results in the calculator program being generated.[18] And you know that your test suite always leaves those debugging logfiles lying around. These facts are true for all working copies of this project, not just your own. And you know that you aren't interested in seeing those things every time you run svn status, and you are pretty sure that nobody else is interested in them either. So you use svn propedit svn:ignore calc to add some ignore patterns to the calc directory.

$ svn propget svn:ignore calc
calculator
debug_log*
$

After you've added this property, you will now have a local property modification on the calc directory. But notice what else is different about your svn status output:

$ svn status
 M      calc
 M      calc/button.c
?       calc/data.c

Now, all that cruft is missing from the output! Your calculator compiled program and all those logfiles are still in your working copy; Subversion just isn't constantly reminding you that they are present and unversioned. And now with all the uninteresting noise removed from the display, you are left with more intriguing items—such as that source code file data.c that you probably forgot to add to version control.

Of course, this less-verbose report of your working copy status isn't the only one available. If you actually want to see the ignored files as part of the status report, you can pass the --no-ignore option to Subversion:

$ svn status --no-ignore
 M      calc
 M      calc/button.c
I       calc/calculator
?       calc/data.c
I       calc/debug_log
I       calc/debug_log.1
I       calc/debug_log.2.gz
I       calc/debug_log.3.gz

As mentioned earlier, the list of file patterns to ignore is also used by svn add and svn import. Both of these operations involve asking Subversion to begin managing some set of files and directories. Rather than force the user to pick and choose which files in a tree she wishes to start versioning, Subversion uses the ignore patterns—both the global and the per-directory lists—to determine which files should not be swept into the version control system as part of a larger recursive addition or import operation. And here again, you can use the --no-ignore option to tell Subversion to disregard its ignores list and operate on all the files and directories present.

[Tip]Tip

Even if svn:ignore is set, you may run into problems if you use shell wildcards in a command. Shell wildcards are expanded into an explicit list of targets before Subversion operates on them, so running svn SUBCOMMAND * is just like running svn SUBCOMMAND file1 file2 file3 …. In the case of the svn add command, this has an effect similar to passing the --no-ignore option. So instead of using a wildcard, use svn add --force . to do a bulk scheduling of unversioned things for addition. The explicit target will ensure that the current directory isn't overlooked because of being already under version control, and the --force option will cause Subversion to crawl through that directory, adding unversioned files while still honoring the svn:ignore property and global-ignores runtime configuration variable. Be sure to also provide the --depth files option to the svn add command if you don't want a fully recursive crawl for things to add.

Keyword Substitution

Subversion has the ability to substitute keywords—pieces of useful, dynamic information about a versioned file—into the contents of the file itself. Keywords generally provide information about the last modification made to the file. Because this information changes each time the file changes, and more importantly, just after the file changes, it is a hassle for any process except the version control system to keep the data completely up to date. Left to human authors, the information would inevitably grow stale.

For example, say you have a document in which you would like to display the last date on which it was modified. You could burden every author of that document to, just before committing their changes, also tweak the part of the document that describes when it was last changed. But sooner or later, someone would forget to do that. Instead, simply ask Subversion to perform keyword substitution on the LastChangedDate keyword. You control where the keyword is inserted into your document by placing a keyword anchor at the desired location in the file. This anchor is just a string of text formatted as $KeywordName$.

All keywords are case-sensitive where they appear as anchors in files: you must use the correct capitalization for the keyword to be expanded. You should consider the value of the svn:keywords property to be case-sensitive, too—certain keyword names will be recognized regardless of case, but this behavior is deprecated.

Subversion defines the list of keywords available for substitution. That list contains the following keywords, some of which have aliases that you can also use:

Date

This keyword describes the last time the file was known to have been changed in the repository, and is of the form $Date: 2006-07-22 21:42:37 -0700 (Sat, 22 Jul 2006) $. It may also be specified as LastChangedDate. Unlike the Id keyword, which uses UTC, the Date keyword displays dates using the local time zone.

Revision

This keyword describes the last known revision in which this file changed in the repository, and looks something like $Revision: 144 $. It may also be specified as LastChangedRevision or Rev.

Author

This keyword describes the last known user to change this file in the repository, and looks something like $Author: harry $. It may also be specified as LastChangedBy.

HeadURL

This keyword describes the full URL to the latest version of the file in the repository, and looks something like $HeadURL: http://svn.example.com/repos/trunk/calc.c $. It may be abbreviated as URL.

Id

This keyword is a compressed combination of the other keywords. Its substitution looks something like $Id: calc.c 148 2006-07-28 21:30:43Z sally $, and is interpreted to mean that the file calc.c was last changed in revision 148 on the evening of July 28, 2006 by the user sally. The date displayed by this keyword is in UTC, unlike that of the Date keyword (which uses the local time zone).

Header

This keyword is similar to the Id keyword but contains the full URL of the latest revision of the item, identical to HeadURL. Its substitution looks something like $Header: http://svn.example.com/repos/trunk/calc.c 148 2006-07-28 21:30:43Z sally $.

Several of the preceding descriptions use the phrase last known or similar wording. Keep in mind that keyword expansion is a client-side operation, and your client knows only about changes that have occurred in the repository when you update your working copy to include those changes. If you never update your working copy, your keywords will never expand to different values even if those versioned files are being changed regularly in the repository.

Simply adding keyword anchor text to your file does nothing special. Subversion will never attempt to perform textual substitutions on your file contents unless explicitly asked to do so. After all, you might be writing a document[19] about how to use keywords, and you don't want Subversion to substitute your beautiful examples of unsubstituted keyword anchors!

To tell Subversion whether to substitute keywords on a particular file, we again turn to the property-related subcommands. The svn:keywords property, when set on a versioned file, controls which keywords will be substituted on that file. The value is a space-delimited list of keyword names or aliases.

For example, say you have a versioned file named weather.txt that looks like this:

Here is the latest report from the front lines.
$LastChangedDate$
$Rev$
Cumulus clouds are appearing more frequently as summer approaches.

With no svn:keywords property set on that file, Subversion will do nothing special. Now, let's enable substitution of the LastChangedDate keyword.

$ svn propset svn:keywords "Date Author" weather.txt
property 'svn:keywords' set on 'weather.txt'
$

Now you have made a local property modification on the weather.txt file. You will see no changes to the file's contents (unless you made some of your own prior to setting the property). Notice that the file contained a keyword anchor for the Rev keyword, yet we did not include that keyword in the property value we set. Subversion will happily ignore requests to substitute keywords that are not present in the file and will not substitute keywords that are not present in the svn:keywords property value.

Immediately after you commit this property change, Subversion will update your working file with the new substitute text. Instead of seeing your keyword anchor $LastChangedDate$, you'll see its substituted result. That result also contains the name of the keyword and continues to be delimited by the dollar sign ($) characters. And as we predicted, the Rev keyword was not substituted because we didn't ask for it to be.

Note also that we set the svn:keywords property to Date Author, yet the keyword anchor used the alias $LastChangedDate$ and still expanded correctly:

Here is the latest report from the front lines.
$LastChangedDate: 2006-07-22 21:42:37 -0700 (Sat, 22 Jul 2006) $
$Rev$
Cumulus clouds are appearing more frequently as summer approaches.

If someone else now commits a change to weather.txt, your copy of that file will continue to display the same substituted keyword value as before—until you update your working copy. At that time, the keywords in your weather.txt file will be resubstituted with information that reflects the most recent known commit to that file.

You can also instruct Subversion to maintain a fixed length (in terms of the number of bytes consumed) for the substituted keyword. By using a double colon (::) after the keyword name, followed by a number of space characters, you define that fixed width. When Subversion goes to substitute your keyword for the keyword and its value, it will essentially replace only those space characters, leaving the overall width of the keyword field unchanged. If the substituted value is shorter than the defined field width, there will be extra padding characters (spaces) at the end of the substituted field; if it is too long, it is truncated with a special hash (#) character just before the final dollar sign terminator.

For example, say you have a document in which you have some section of tabular data reflecting the document's Subversion keywords. Using the original Subversion keyword substitution syntax, your file might look something like:

$Rev$:     Revision of last commit
$Author$:  Author of last commit
$Date$:    Date of last commit

Now, that looks nice and tabular at the start of things. But when you then commit that file (with keyword substitution enabled, of course), you see:

$Rev: 12 $:     Revision of last commit
$Author: harry $:  Author of last commit
$Date: 2006-03-15 02:33:03 -0500 (Wed, 15 Mar 2006) $:    Date of last commit

The result is not so beautiful. And you might be tempted to then adjust the file after the substitution so that it again looks tabular. But that holds only as long as the keyword values are the same width. If the last committed revision rolls into a new place value (say, from 99 to 100), or if another person with a longer username commits the file, stuff gets all crooked again. However, if you are using Subversion 1.2 or later, you can use the new fixed-length keyword syntax and define some field widths that seem sane, so your file might look like this:

$Rev::               $:  Revision of last commit
$Author::            $:  Author of last commit
$Date::              $:  Date of last commit

You commit this change to your file. This time, Subversion notices the new fixed-length keyword syntax and maintains the width of the fields as defined by the padding you placed between the double colon and the trailing dollar sign. After substitution, the width of the fields is completely unchanged—the short values for Rev and Author are padded with spaces, and the long Date field is truncated by a hash character:

$Rev:: 13            $:  Revision of last commit
$Author:: harry      $:  Author of last commit
$Date:: 2006-03-15 0#$:  Date of last commit

The use of fixed-length keywords is especially handy when performing substitutions into complex file formats that themselves use fixed-length fields for data, or for which the stored size of a given data field is overbearingly difficult to modify from outside the format's native application. Of course, where binary file formats are concerned, you must always take great care that any keyword substitution you introduce—fixed-length or otherwise—does not violate the integrity of that format. While it might sound easy enough, this can be an astonishingly difficult task for most of the popular binary file formats in use today, and not something to be undertaken by the faint of heart!

[Warning]Warning

Be aware that because the width of a keyword field is measured in bytes, the potential for corruption of multibyte values exists. For example, a username that contains some multibyte UTF-8 characters might suffer truncation in the middle of the string of bytes that make up one of those characters. The result will be a mere truncation when viewed at the byte level, but will likely appear as a string with an incorrect or garbled final character when viewed as UTF-8 text. It is conceivable that certain applications, when asked to load the file, would notice the broken UTF-8 text and deem the entire file corrupt, refusing to operate on the file altogether. So, when limiting keywords to a fixed size, choose a size that allows for this type of byte-wise expansion.

Sparse Directories

By default, most Subversion operations on directories act in a recursive manner. For example, svn checkout creates a working copy with every file and directory in the specified area of the repository, descending recursively through the repository tree until the entire structure is copied to your local disk. Subversion 1.5 introduces a feature called sparse directories (or shallow checkouts) that allows you to easily check out a working copy—or a portion of a working copy—more shallowly than full recursion, with the freedom to bring in previously ignored files and subdirectories at a later time.

For example, say we have a repository with a tree of files and directories with names of the members of a human family with pets. (It's an odd example, to be sure, but bear with us.) A regular svn checkout operation will give us a working copy of the whole tree:

$ svn checkout file:///var/svn/repos mom
A    mom/son
A    mom/son/grandson
A    mom/daughter
A    mom/daughter/granddaughter1
A    mom/daughter/granddaughter1/bunny1.txt
A    mom/daughter/granddaughter1/bunny2.txt
A    mom/daughter/granddaughter2
A    mom/daughter/fishie.txt
A    mom/kitty1.txt
A    mom/doggie1.txt
Checked out revision 1.
$

Now, let's check out the same tree again, but this time we'll ask Subversion to give us only the topmost directory with none of its children at all:

$ svn checkout file:///var/svn/repos mom-empty --depth empty
Checked out revision 1
$

Notice that we added to our original svn checkout command line a new --depth option. This option is present on many of Subversion's subcommands and is similar to the --non-recursive (-N) and --recursive (-R) options. In fact, it combines, improves upon, supercedes, and ultimately obsoletes these two older options. For starters, it expands the supported degrees of depth specification available to users, adding some previously unsupported (or inconsistently supported) depths. Here are the depth values that you can request for a given Subversion operation:

--depth empty

Include only the immediate target of the operation, not any of its file or directory children.

--depth files

Include the immediate target of the operation and any of its immediate file children.

--depth immediates

Include the immediate target of the operation and any of its immediate file or directory children. The directory children will themselves be empty.

--depth infinity

Include the immediate target, its file and directory children, its children's children, and so on to full recursion.

Of course, merely combining two existing options into one hardly constitutes a new feature worthy of a whole section in our book. Fortunately, there is more to this story. This idea of depth extends not just to the operations you perform with your Subversion client, but also as a description of a working copy citizen's ambient depth, which is the depth persistently recorded by the working copy for that item. Its key strength is this very persistence—the fact that it is sticky. The working copy remembers the depth you've selected for each item in it until you later change that depth selection; by default, Subversion commands operate on the working copy citizens present, regardless of their selected depth settings.

[Tip]Tip

You can check the recorded ambient depth of a working copy using the svn info command. If the ambient depth is anything other than infinite recursion, svn info will display a line describing that depth value:

$ svn info mom-immediates | grep "^Depth:"
Depth: immediates
$

Our previous examples demonstrated checkouts of infinite depth (the default for svn checkout) and empty depth. Let's look now at examples of the other depth values:

$ svn checkout file:///var/svn/repos mom-files --depth files
A    mom-files/kitty1.txt
A    mom-files/doggie1.txt
Checked out revision 1.
$ svn checkout file:///var/svn/repos mom-immediates --depth immediates
A    mom-immediates/son
A    mom-immediates/daughter
A    mom-immediates/kitty1.txt
A    mom-immediates/doggie1.txt
Checked out revision 1.
$

As described, each of these depths is something more than only the target, but something less than full recursion.

We've used svn checkout as an example here, but you'll find the --depth option present on many other Subversion commands, too. In those other commands, depth specification is a way to limit the scope of an operation to some depth, much like the way the older --non-recursive (-N) and --recursive (-R) options behave. This means that when operating on a working copy of some depth, while requesting an operation of a shallower depth, the operation is limited to that shallower depth. In fact, we can make an even more general statement: given a working copy of any arbitrary—even mixed—ambient depth, and a Subversion command with some requested operational depth, the command will maintain the ambient depth of the working copy members while still limiting the scope of the operation to the requested (or default) operational depth.

In addition to the --depth option, the svn update and svn switch subcommands also accept a second depth-related option: --set-depth. It is with this option that you can change the sticky depth of a working copy item. Watch what happens as we take our empty-depth checkout and gradually telescope it deeper using svn update --set-depth NEW-DEPTH TARGET:

$ svn update --set-depth files mom-empty
Updating 'mom-empty':
A    mom-empty/kittie1.txt
A    mom-empty/doggie1.txt
Updated to revision 1.
$ svn update --set-depth immediates mom-empty
Updating 'mom-empty':
A    mom-empty/son
A    mom-empty/daughter
Updated to revision 1.
$ svn update --set-depth infinity mom-empty
Updating 'mom-empty':
A    mom-empty/son/grandson
A    mom-empty/daughter/granddaughter1
A    mom-empty/daughter/granddaughter1/bunny1.txt
A    mom-empty/daughter/granddaughter1/bunny2.txt
A    mom-empty/daughter/granddaughter2
A    mom-empty/daughter/fishie1.txt
Updated to revision 1.
$

As we gradually increased our depth selection, the repository gave us more pieces of our tree.

In our example, we operated only on the root of our working copy, changing its ambient depth value. But we can independently change the ambient depth value of any subdirectory inside the working copy, too. Careful use of this ability allows us to flesh out only certain portions of the working copy tree, leaving other portions absent altogether (hence the sparse bit of the feature's name). Here's an example of how we might build out a portion of one branch of our family's tree, enable full recursion on another branch, and keep still other pieces pruned (absent from disk).

$ rm -rf mom-empty
$ svn checkout file:///var/svn/repos mom-empty --depth empty
Checked out revision 1.
$ svn update --set-depth empty mom-empty/son
Updating 'mom-empty/son':
A    mom-empty/son
Updated to revision 1.
$ svn update --set-depth empty mom-empty/daughter
Updating 'mom-empty/daughter':
A    mom-empty/daughter
Updated to revision 1.
$ svn update --set-depth infinity mom-empty/daughter/granddaughter1
Updating 'mom-empty/daughter/granddaughter1':
A    mom-empty/daughter/granddaughter1
A    mom-empty/daughter/granddaughter1/bunny1.txt
A    mom-empty/daughter/granddaughter1/bunny2.txt
Updated to revision 1.
$

Fortunately, having a complex collection of ambient depths in a single working copy doesn't complicate the way you interact with that working copy. You can still make, revert, display, and commit local modifications in your working copy without providing any new options (including --depth and --set-depth) to the relevant subcommands. Even svn update works as it does elsewhere when no specific depth is provided—it updates the working copy targets that are present while honoring their sticky depths.

You might at this point be wondering, So what? When would I use this? One scenario where this feature finds utility is tied to a particular repository layout, specifically where you have many related or codependent projects or software modules living as siblings in a single repository location (trunk/project1, trunk/project2, trunk/project3, etc.). In such scenarios, it might be the case that you personally care about only a handful of those projects—maybe some primary project and a few other modules on which it depends. You can check out individual working copies of all of these things, but those working copies are disjoint and, as a result, it can be cumbersome to perform operations across several or all of them at the same time. The alternative is to use the sparse directories feature, building out a single working copy that contains only the modules you care about. You'd start with an empty-depth checkout of the common parent directory of the projects, and then update with infinite depth only the items you wish to have, like we demonstrated in the previous example. Think of it like an opt-in system for working copy citizens.

The original (Subversion 1.5) implementation of shallow checkouts was good, but didn't support de-telescoping of working copy items. Subversion 1.6 remedied this problem. For example, running svn update --set-depth empty in an infinite-depth working copy will discard everything but the topmost directory.[20] Subversion 1.6 also introduced another supported value for the --set-depth option: exclude. Using --set-depth exclude with svn update will cause the update target to be removed from the working copy entirely—a directory target won't even be left present-but-empty. This is especially handy when there are more things that you'd like to keep in a working copy than things you'd like to not keep.

Consider a directory with hundreds of subdirectories, one of which you would like to omit from your working copy. Using an additive approach to sparse directories, you might check out the directory with an empty depth, then explicitly telescope (using svn update --set-depth infinity) each and every subdirectory of the directory except the one you don't care about.

$ svn checkout http://svn.example.com/repos/many-dirs --depth empty
…
$ svn update --set-depth infinity many-dirs/wanted-dir-1
…
$ svn update --set-depth infinity many-dirs/wanted-dir-2
…
$ svn update --set-depth infinity many-dirs/wanted-dir-3
…
### and so on, and so on, ...

This could be quite tedious, especially since you don't even have stubs of these directories in your working copy to deal with. Such a working copy would also have another characteristic that you might not expect or desire: if someone else creates any new subdirectories in this top-level directory, you won't receive those when you update your working copy.

Beginning with Subversion 1.6, you can take a different approach. First, check out the directory in full. Then run svn update --set-depth exclude on the one subdirectory you don't care about.

$ svn checkout http://svn.example.com/repos/many-dirs
…
$ svn update --set-depth exclude many-dirs/unwanted-dir
D         many-dirs/unwanted-dir
$

This approach leaves your working copy with the same stuff as in the first approach, but any new subdirectories which appear in the top-level directory would also show up when you update your working copy. The downside of this approach is that you have to actually check out that whole subdirectory that you don't even want just so you can tell Subversion that you don't want it. This might not even be possible if that subdirectory is too large to fit on your disk (which might, after all, be the very reason you don't want it in your working copy).

[Note]Note

While the functionality for excluding an existing item from a working copy was hung off of the svn update command, you might have noticed that the output from svn update --set-depth exclude differs from that of a normal update operation. This output betrays the fact that, under the hood, exclusion is a completely client-side operation, very much unlike a typical update.

In such a situation, you might consider a compromise approach. First, check out the top-level directory with --depth immediates. Then, exclude the directory you don't want using svn update --set-depth exclude. Finally, telescope all the items that remain to infinite depth, which should be fairly easy to do because they are all addressable in your shell.

$ svn checkout http://svn.example.com/repos/many-dirs --depth immediates
…
$ svn update --set-depth exclude many-dirs/unwanted-dir
D         many-dirs/unwanted-dir
$ svn update --set-depth infinity many-dirs/*
…
$

Once again, your working copy will have the same stuff as in the previous two scenarios. But now, any time a new file or subdirectory is committed to the top-level directory, you'll receive it—at an empty depth—when you update your working copy. You can now decide what to do with such newly appearing working copy items: expand them into infinite depth, or exclude them altogether.

Locking

Subversion's copy-modify-merge version control model lives and dies on its data merging algorithms—specifically on how well those algorithms perform when trying to resolve conflicts caused by multiple users modifying the same file concurrently. Subversion itself provides only one such algorithm: a three-way differencing algorithm that is smart enough to handle data at a granularity of a single line of text. Subversion also allows you to supplement its content merge processing with external differencing utilities (as described in the section called “External diff3” and the section called “External merge”), some of which may do an even better job, perhaps providing granularity of a word or a single character of text. But common among those algorithms is that they generally work only on text files. The landscape starts to look pretty grim when you start talking about content merges of nontextual file formats. And when you can't find a tool that can handle that type of merging, you begin to run into problems with the copy-modify-merge model.

Let's look at a real-life example of where this model runs aground. Harry and Sally are both graphic designers working on the same project, a bit of marketing collateral for an automobile mechanic. Central to the design of a particular poster is an image of a car in need of some bodywork, stored in a file using the PNG image format. The poster's layout is almost finished, and both Harry and Sally are pleased with the particular photo they chose for their damaged car—a baby blue 1967 Ford Mustang with an unfortunate bit of crumpling on the left front fender.

Now, as is common in graphic design work, there's a change in plans, which causes the car's color to be a concern. So Sally updates her working copy to HEAD, fires up her photo-editing software, and sets about tweaking the image so that the car is now cherry red. Meanwhile, Harry, feeling particularly inspired that day, decides that the image would have greater impact if the car also appears to have suffered greater impact. He, too, updates to HEAD, and then draws some cracks on the vehicle's windshield. He manages to finish his work before Sally finishes hers, and after admiring the fruits of his undeniable talent, he commits the modified image. Shortly thereafter, Sally is finished with the car's new finish and tries to commit her changes. But, as expected, Subversion fails the commit, informing Sally that her version of the image is now out of date.

Here's where the difficulty sets in. If Harry and Sally were making changes to a text file, Sally would simply update her working copy, receiving Harry's changes in the process. In the worst possible case, they would have modified the same region of the file, and Sally would have to work out by hand the proper resolution to the conflict. But these aren't text files—they are binary images. And while it's a simple matter to describe what one would expect the results of this content merge to be, there is precious little chance that any software exists that is smart enough to examine the common baseline image that each of these graphic artists worked against, the changes that Harry made, and the changes that Sally made, and then spit out an image of a busted-up red Mustang with a cracked windshield!

Of course, things would have gone more smoothly if Harry and Sally had serialized their modifications to the image—if, say, Harry had waited to draw his windshield cracks on Sally's now-red car, or if Sally had tweaked the color of a car whose windshield was already cracked. As is discussed in the section called “The copy-modify-merge solution”, most of these types of problems go away entirely where perfect communication between Harry and Sally exists.[21] But as one's version control system is, in fact, one form of communication, it follows that having that software facilitate the serialization of nonparallelizable editing efforts is no bad thing. This is where Subversion's implementation of the lock-modify-unlock model steps into the spotlight. This is where we talk about Subversion's locking feature, which is similar to the reserved checkouts mechanisms of other version control systems.

Subversion's locking feature exists ultimately to minimize wasted time and effort. By allowing a user to programmatically claim the exclusive right to change a file in the repository, that user can be reasonably confident that any energy he invests on unmergeable changes won't be wasted—his commit of those changes will succeed. Also, because Subversion communicates to other users that serialization is in effect for a particular versioned object, those users can reasonably expect that the object is about to be changed by someone else. They, too, can then avoid wasting their time and energy on unmergeable changes that won't be committable due to eventual out-of-dateness.

When referring to Subversion's locking feature, one is actually talking about a fairly diverse collection of behaviors, which include the ability to lock a versioned file[22] (claiming the exclusive right to modify the file), to unlock that file (yielding that exclusive right to modify), to see reports about which files are locked and by whom, to annotate files for which locking before editing is strongly advised, and so on. In this section, we'll cover all of these facets of the larger locking feature.

Creating Locks

In the Subversion repository, a lock is a piece of metadata that grants exclusive access to one user to change a file. This user is said to be the lock owner. Each lock also has a unique identifier, typically a long string of characters, known as the lock token. The repository manages locks, ultimately handling their creation, enforcement, and removal. If any commit transaction attempts to modify or delete a locked file (or delete one of the parent directories of the file), the repository will demand two pieces of information—that the client performing the commit be authenticated as the lock owner, and that the lock token has been provided as part of the commit process as a form of proof that the client knows which lock it is using.

To demonstrate lock creation, let's refer back to our example of multiple graphic designers working on the same binary image files. Harry has decided to change a JPEG image. To prevent other people from committing changes to the file while he is modifying it (as well as alerting them that he is about to change it), he locks the file in the repository using the svn lock command.

$ svn lock banana.jpg -m "Editing file for tomorrow's release."
'banana.jpg' locked by user 'harry'.
$

The preceding example demonstrates a number of new things. First, notice that Harry passed the --message (-m) option to svn lock. Similar to svn commit, the svn lock command can take comments—via either --message (-m) or --file (-F)—to describe the reason for locking the file. Unlike svn commit, however, svn lock will not demand a message by launching your preferred text editor. Lock comments are optional, but still recommended to aid communication.

Second, the lock attempt succeeded. This means that the file wasn't already locked, and that Harry had the latest version of the file. If Harry's working copy of the file had been out of date, the repository would have rejected the request, forcing Harry to svn update and reattempt the locking command. The locking command would also have failed if the file had already been locked by someone else.

As you can see, the svn lock command prints confirmation of the successful lock. At this point, the fact that the file is locked becomes apparent in the output of the svn status and svn info reporting subcommands.

$ svn status
     K  banana.jpg

$ svn info banana.jpg
Path: banana.jpg
Name: banana.jpg
Working Copy Root Path: /home/harry/project
URL: http://svn.example.com/repos/project/banana.jpg
Repository Root: http://svn.example.com/repos/project
Repository UUID: edb2f264-5ef2-0310-a47a-87b0ce17a8ec
Revision: 2198
Node Kind: file
Schedule: normal
Last Changed Author: frank
Last Changed Rev: 1950
Last Changed Date: 2006-03-15 12:43:04 -0600 (Wed, 15 Mar 2006)
Text Last Updated: 2006-06-08 19:23:07 -0500 (Thu, 08 Jun 2006)
Properties Last Updated: 2006-06-08 19:23:07 -0500 (Thu, 08 Jun 2006)
Checksum: 3b110d3b10638f5d1f4fe0f436a5a2a5
Lock Token: opaquelocktoken:0c0f600b-88f9-0310-9e48-355b44d4a58e
Lock Owner: harry
Lock Created: 2006-06-14 17:20:31 -0500 (Wed, 14 Jun 2006)
Lock Comment (1 line):
Editing file for tomorrow's release.

$

The fact that the svn info command, which does not contact the repository when run against working copy paths, can display the lock token reveals an important piece of information about those tokens: they are cached in the working copy. The presence of the lock token is critical. It gives the working copy authorization to make use of the lock later on. Also, the svn status command shows a K next to the file (short for locKed), indicating that the lock token is present.

Now that Harry has locked banana.jpg, Sally is unable to change or delete that file:

$ svn delete banana.jpg
D         banana.jpg
$ svn commit -m "Delete useless file."
Deleting       banana.jpg
svn: E175002: Commit failed (details follow):
svn: E175002: Server sent unexpected return value (423 Locked) in response to 
DELETE request for '/repos/project/!svn/wrk/64bad3a9-96f9-0310-818a-df4224ddc
35d/banana.jpg'
$

But Harry, after touching up the banana's shade of yellow, is able to commit his changes to the file. That's because he authenticates as the lock owner and also because his working copy holds the correct lock token:

$ svn status
M    K  banana.jpg
$ svn commit -m "Make banana more yellow"
Sending        banana.jpg
Transmitting file data .
Committed revision 2201.
$ svn status
$

Notice that after the commit is finished, svn status shows that the lock token is no longer present in the working copy. This is the standard behavior of svn commit—it searches the working copy (or list of targets, if you provide such a list) for local modifications and sends all the lock tokens it encounters during this walk to the server as part of the commit transaction. After the commit completes successfully, all of the repository locks that were mentioned are released—even on files that weren't committed. This is meant to discourage users from being sloppy about locking or from holding locks for too long. If Harry haphazardly locks 30 files in a directory named images because he's unsure of which files he needs to change, yet changes only four of those files, when he runs svn commit images, the process will still release all 30 locks.

This behavior of automatically releasing locks can be overridden with the --no-unlock option to svn commit. This is best used for those times when you want to commit changes, but still plan to make more changes and thus need to retain existing locks. You can also make this your default behavior by setting the no-unlock runtime configuration option (see the section called “Runtime Configuration Area”).

Of course, locking a file doesn't oblige one to commit a change to it. The lock can be released at any time with a simple svn unlock command:

$ svn unlock banana.c
'banana.c' unlocked.

Discovering Locks

When a commit fails due to someone else's locks, it's fairly easy to learn about them. The easiest way is to run svn status -u:

$ svn status -u
M               23   bar.c
M    O          32   raisin.jpg
        *       72   foo.h
Status against revision:     105
$

In this example, Sally can see not only that her copy of foo.h is out of date, but also that one of the two modified files she plans to commit is locked in the repository. The O symbol stands for Other, meaning that a lock exists on the file and was created by somebody else. If she were to attempt a commit, the lock on raisin.jpg would prevent it. Sally is left wondering who made the lock, when, and why. Once again, svn info has the answers:

$ svn info ^/raisin.jpg
Path: raisin.jpg
Name: raisin.jpg
URL: http://svn.example.com/repos/project/raisin.jpg
Repository Root: http://svn.example.com/repos/project
Repository UUID: edb2f264-5ef2-0310-a47a-87b0ce17a8ec
Revision: 105
Node Kind: file
Last Changed Author: sally
Last Changed Rev: 32
Last Changed Date: 2006-01-25 12:43:04 -0600 (Sun, 25 Jan 2006)
Lock Token: opaquelocktoken:fc2b4dee-98f9-0310-abf3-653ff3226e6b
Lock Owner: harry
Lock Created: 2006-02-16 13:29:18 -0500 (Thu, 16 Feb 2006)
Lock Comment (1 line):
Need to make a quick tweak to this image.
$

Just as you can use svn info to examine objects in the working copy, you can also use it to examine objects in the repository. If the main argument to svn info is a working copy path, then all of the working copy's cached information is displayed; any mention of a lock means that the working copy is holding a lock token (if a file is locked by another user or in another working copy, svn info on a working copy path will show no lock information at all). If the main argument to svn info is a URL, the information reflects the latest version of an object in the repository, and any mention of a lock describes the current lock on the object.

So in this particular example, Sally can see that Harry locked the file on February 16 to make a quick tweak. It being June, she suspects that he probably forgot all about the lock. She might phone Harry to complain and ask him to release the lock. If he's unavailable, she might try to forcibly break the lock herself or ask an administrator to do so.

Breaking and Stealing Locks

A repository lock isn't sacred—in Subversion's default configuration state, locks can be released not only by the person who created them, but by anyone. When somebody other than the original lock creator destroys a lock, we refer to this as breaking the lock.

From the administrator's chair, it's simple to break locks. The svnlook and svnadmin programs have the ability to display and remove locks directly from the repository. (For more information about these tools, see the section called “An Administrator's Toolkit”.)

$ svnadmin lslocks /var/svn/repos
Path: /project2/images/banana.jpg
UUID Token: opaquelocktoken:c32b4d88-e8fb-2310-abb3-153ff1236923
Owner: frank
Created: 2006-06-15 13:29:18 -0500 (Thu, 15 Jun 2006)
Expires: 
Comment (1 line):
Still improving the yellow color.

Path: /project/raisin.jpg
UUID Token: opaquelocktoken:fc2b4dee-98f9-0310-abf3-653ff3226e6b
Owner: harry
Created: 2006-02-16 13:29:18 -0500 (Thu, 16 Feb 2006)
Expires: 
Comment (1 line):
Need to make a quick tweak to this image.

$ svnadmin rmlocks /var/svn/repos /project/raisin.jpg
Removed lock on '/project/raisin.jpg'.
$

The more interesting option is to allow users to break each other's locks over the network. To do this, Sally simply needs to pass the --force to the svn unlock command:

$ svn status -u
M               23   bar.c
M    O          32   raisin.jpg
        *       72   foo.h
Status against revision:     105
$ svn unlock raisin.jpg
svn: E195013: 'raisin.jpg' is not locked in this working copy
$ svn info raisin.jpg | grep URL
URL: http://svn.example.com/repos/project/raisin.jpg
$ svn unlock http://svn.example.com/repos/project/raisin.jpg
svn: warning: W160039: Unlock failed on 'raisin.jpg' (403 Forbidden)
$ svn unlock --force http://svn.example.com/repos/project/raisin.jpg
'raisin.jpg' unlocked.
$

Now, Sally's initial attempt to unlock failed because she ran svn unlock directly on her working copy of the file, and no lock token was present. To remove the lock directly from the repository, she needs to pass a URL to svn unlock. Her first attempt to unlock the URL fails, because she can't authenticate as the lock owner (nor does she have the lock token). But when she passes --force, the authentication and authorization requirements are ignored, and the remote lock is broken.

Simply breaking a lock may not be enough. In the running example, Sally may not only want to break Harry's long-forgotten lock, but relock the file for her own use. She can accomplish this by using svn unlock with --force and then svn lock back-to-back, but there's a small chance that somebody else might lock the file between the two commands. The simpler thing to do is to steal the lock, which involves breaking and relocking the file all in one atomic step. To do this, Sally passes the --force option to svn lock:

$ svn lock raisin.jpg
svn: warning: W160035: Path '/project/raisin.jpg' is already locked by user 'h
arry' in filesystem '/var/svn/repos/db'
$ svn lock --force raisin.jpg
'raisin.jpg' locked by user 'sally'.
$

In any case, whether the lock is broken or stolen, Harry may be in for a surprise. Harry's working copy still contains the original lock token, but that lock no longer exists. The lock token is said to be defunct. The lock represented by the lock token has either been broken (no longer in the repository) or stolen (replaced with a different lock). Either way, Harry can see this by asking svn status to contact the repository:

$ svn status
     K  raisin.jpg
$ svn status -u
     B          32   raisin.jpg
Status against revision:     105
$ svn update
Updating '.':
  B  raisin.jpg
Updated to revision 105.
$ svn status
$

If the repository lock was broken, then svn status --show-updates (-u) displays a B (Broken) symbol next to the file. If a new lock exists in place of the old one, then a T (sTolen) symbol is shown. Finally, svn update notices any defunct lock tokens and removes them from the working copy.

Lock Communication

We've seen how svn lock and svn unlock can be used to create, release, break, and steal locks. This satisfies the goal of serializing commit access to a file. But what about the larger problem of preventing wasted time?

For example, suppose Harry locks an image file and then begins editing it. Meanwhile, miles away, Sally wants to do the same thing. She doesn't think to run svn status -u, so she has no idea that Harry has already locked the file. She spends hours editing the file, and when she tries to commit her change, she discovers that either the file is locked or that she's out of date. Regardless, her changes aren't mergeable with Harry's. One of these two people has to throw away his or her work, and a lot of time has been wasted.

Subversion's solution to this problem is to provide a mechanism to remind users that a file ought to be locked before the editing begins. The mechanism is a special property: svn:needs-lock. If that property is attached to a file (regardless of its value, which is irrelevant), Subversion will try to use filesystem-level permissions to make the file read-only—unless, of course, the user has explicitly locked the file. When a lock token is present (as a result of using svn lock), the file becomes read/write. When the lock is released, the file becomes read-only again.

The theory, then, is that if the image file has this property attached, Sally would immediately notice something is strange when she opens the file for editing: many applications alert users immediately when a read-only file is opened for editing, and nearly all would prevent her from saving changes to the file. This reminds her to lock the file before editing, whereby she discovers the preexisting lock:

$ /usr/local/bin/gimp raisin.jpg
gimp: error: file is read-only!
$ ls -l raisin.jpg
-r--r--r--   1 sally   sally   215589 Jun  8 19:23 raisin.jpg
$ svn lock raisin.jpg
svn: warning: W160035: Path '/project/raisin.jpg' is already locked by user 'h
arry' in filesystem '/var/svn/repos/db'
$ svn info http://svn.example.com/repos/project/raisin.jpg | grep Lock
Lock Token: opaquelocktoken:fc2b4dee-98f9-0310-abf3-653ff3226e6b
Lock Owner: harry
Lock Created: 2006-06-08 07:29:18 -0500 (Thu, 08 June 2006)
Lock Comment (1 line):
Making some tweaks.  Locking for the next two hours.
$
[Tip]Tip

Users and administrators alike are encouraged to attach the svn:needs-lock property to any file that cannot be contextually merged. This is the primary technique for encouraging good locking habits and preventing wasted effort.

Note that this property is a communication tool that works independently from the locking system. In other words, any file can be locked, whether or not this property is present. And conversely, the presence of this property doesn't make the repository require a lock when committing.

Unfortunately, the system isn't flawless. It's possible that even when a file has the property, the read-only reminder won't always work. Sometimes applications misbehave and hijack the read-only file, silently allowing users to edit and save the file anyway. There's not much that Subversion can do in this situation—at the end of the day, there's simply no substitution for good interpersonal communication.[23]

Externals Definitions

Sometimes it is useful to construct a working copy that is made out of a number of different checkouts. For example, you may want different subdirectories to come from different locations in a repository or perhaps from different repositories altogether. You could certainly set up such a scenario by hand—using svn checkout to create the sort of nested working copy structure you are trying to achieve. But if this layout is important for everyone who uses your repository, every other user will need to perform the same checkout operations that you did.

Fortunately, Subversion provides support for externals definitions. An externals definition is a mapping of a local directory to the URL—and ideally a particular revision—of a versioned directory. In Subversion, you declare externals definitions in groups using the svn:externals property. You can create or modify this property using svn propset or svn propedit (see the section called “Manipulating Properties”). It can be set on any versioned directory, and its value describes both the external repository location and the client-side directory to which that location should be checked out.

The convenience of the svn:externals property is that once it is set on a versioned directory, everyone who checks out a working copy with that directory also gets the benefit of the externals definition. In other words, once one person has made the effort to define the nested working copy structure, no one else has to bother—Subversion will, after checking out the original working copy, automatically also check out the external working copies.

[Warning]Warning

The relative target subdirectories of externals definitions must not already exist on your or other users' systems—Subversion will create them when it checks out the external working copy.

You also get in the externals definition design all the regular benefits of Subversion properties. The definitions are versioned. If you need to change an externals definition, you can do so using the regular property modification subcommands. When you commit a change to the svn:externals property, Subversion will synchronize the checked-out items against the changed externals definition when you next run svn update. The same thing will happen when others update their working copies and receive your changes to the externals definition.

[Tip]Tip

Because the svn:externals property has a multiline value, we strongly recommend that you use svn propedit instead of svn propset.

Subversion releases prior to 1.5 honor an externals definition format that is a multiline table of subdirectories (relative to the versioned directory on which the property is set), optional revision flags, and fully qualified, absolute Subversion repository URLs. An example of this might look as follows:

$ svn propget svn:externals calc
third-party/sounds             http://svn.example.com/repos/sounds
third-party/skins -r148        http://svn.example.com/skinproj
third-party/skins/toolkit -r21 http://svn.example.com/skin-maker

When someone checks out a working copy of the calc directory referred to in the previous example, Subversion also continues to check out the items found in its externals definition.

$ svn checkout http://svn.example.com/repos/calc
A    calc
A    calc/Makefile
A    calc/integer.c
A    calc/button.c
Checked out revision 148.

Fetching external item into calc/third-party/sounds
A    calc/third-party/sounds/ding.ogg
A    calc/third-party/sounds/dong.ogg
A    calc/third-party/sounds/clang.ogg
…
A    calc/third-party/sounds/bang.ogg
A    calc/third-party/sounds/twang.ogg
Checked out revision 14.

Fetching external item into calc/third-party/skins
…

As of Subversion 1.5, though, a new format of the svn:externals property is supported. Externals definitions are still multiline, but the order and format of the various pieces of information have changed. The new syntax more closely mimics the order of arguments you might pass to svn checkout: the optional revision flags come first, then the external Subversion repository URL, and finally the relative local subdirectory. Notice, though, that this time we didn't say fully qualified, absolute Subversion repository URLs. That's because the new format supports relative URLs and URLs that carry peg revisions. The previous example of an externals definition might, in Subversion 1.5, look like the following:

$ svn propget svn:externals calc
      http://svn.example.com/repos/sounds third-party/sounds
-r148 http://svn.example.com/skinproj third-party/skins
-r21  http://svn.example.com/skin-maker third-party/skins/toolkit

Or, making use of the peg revision syntax (which we describe in detail in the section called “Peg and Operative Revisions”), it might appear as:

$ svn propget svn:externals calc
http://svn.example.com/repos/sounds third-party/sounds
http://svn.example.com/skinproj@148 third-party/skins
http://svn.example.com/skin-maker@21 third-party/skins/toolkit
[Tip]Tip

You should seriously consider using explicit revision numbers in all of your externals definitions. Doing so means that you get to decide when to pull down a different snapshot of external information, and exactly which snapshot to pull. Besides avoiding the surprise of getting changes to third-party repositories that you might not have any control over, using explicit revision numbers also means that as you backdate your working copy to a previous revision, your externals definitions will also revert to the way they looked in that previous revision, which in turn means that the external working copies will be updated to match the way they looked back when your repository was at that previous revision. For software projects, this could be the difference between a successful and a failed build of an older snapshot of your complex codebase.

For most repositories, these three ways of formatting the externals definitions have the same ultimate effect. They all bring the same benefits. Unfortunately, they all bring the same annoyances, too. Since the definitions shown use absolute URLs, moving or copying a directory to which they are attached will not affect what gets checked out as an external (though the relative local target subdirectory will, of course, move with the renamed directory). This can be confusing—even frustrating—in certain situations. For example, say you have a top-level directory named my-project, and you've created an externals definition on one of its subdirectories (my-project/some-dir) that tracks the latest revision of another of its subdirectories (my-project/external-dir).

$ svn checkout http://svn.example.com/projects .
A    my-project
A    my-project/some-dir
A    my-project/external-dir
…
Fetching external item into 'my-project/some-dir/subdir'
Checked out external at revision 11.

Checked out revision 11.
$ svn propget svn:externals my-project/some-dir
subdir http://svn.example.com/projects/my-project/external-dir

$

Now you use svn move to rename the my-project directory. At this point, your externals definition will still refer to a path under the my-project directory, even though that directory no longer exists.

$ svn move -q my-project renamed-project
$ svn commit -m "Rename my-project to renamed-project."
Deleting       my-project
Adding         renamed-project

Committed revision 12.
$ svn update
Updating '.':

svn: warning: W200000: Error handling externals definition for 'renamed-projec
t/some-dir/subdir':
svn: warning: W170000: URL 'http://svn.example.com/projects/my-project/externa
l-dir' at revision 12 doesn't exist
At revision 12.
svn: E205011: Failure occurred processing one or more externals definitions
$

Also, absolute URLs can cause problems with repositories that are available via multiple URL schemes. For example, if your Subversion server is configured to allow everyone to check out the repository over http:// or https://, but only allow commits to come in via https://, you have an interesting problem on your hands. If your externals definitions use the http:// form of the repository URLs, you won't be able to commit anything from the working copies created by those externals. On the other hand, if they use the https:// form of the URLs, anyone who might be checking out via http:// because his client doesn't support https:// will be unable to fetch the external items. Be aware, too, that if you need to reparent your working copy (using svn relocate), externals definitions will not also be reparented.

Subversion 1.5 takes a huge step in relieving these frustrations. As mentioned earlier, the URLs used in the new externals definition format can be relative, and Subversion provides syntax magic for specifying multiple flavors of URL relativity.

../

Relative to the URL of the directory on which the svn:externals property is set

^/

Relative to the root of the repository in which the svn:externals property is versioned

//

Relative to the scheme of the URL of the directory on which the svn:externals property is set

/

Relative to the root URL of the server on which the svn:externals property is versioned

^/../REPO-NAME

Relative to a sibling repository beneath the same SVNParentPath location as the repository in which the svn:externals is defined.

So, looking a fourth time at our previous externals definition example, and making use of the new absolute URL syntax in various ways, we might now see:

$ svn propget svn:externals calc
^/sounds third-party/sounds
/skinproj@148 third-party/skins
//svn.example.com/skin-maker@21 third-party/skins/toolkit
$

Subversion 1.6 brought two more improvements to externals definitions. First, it added a quoting and escape mechanism to the syntax so that the path of the external working copy may contain whitespace. This was previously problematic, of course, because whitespace is used to delimit the fields in an externals definition. Now you need only wrap such a path specification in double-quote (") characters or escape the problematic characters in the path with a backslash (\) character. Of course, if you have spaces in the URL portion of the external definition, you should use the standard URI-encoding mechanism to represent those.

$ svn propget svn:externals paint
http://svn.thirdparty.com/repos/My%20Project "My Project"
http://svn.thirdparty.com/repos/%22Quotes%20Too%22 \"Quotes\ Too\"
$

Subversion 1.6 also introduced support for external definitions for files. File externals are configured just like externals for directories and appear as a versioned file in the working copy.

For example, let's say you had the file /trunk/bikeshed/blue.html in your repository, and you wanted this file, as it appeared in revision 40, to appear in your working copy of /trunk/www/ as green.html.

The externals definition required to achieve this should look familiar by now:

$ svn propget svn:externals www/
^/trunk/bikeshed/blue.html@40 green.html
$ svn update
Updating '.':

Fetching external item into 'www'
E    www/green.html
Updated external to revision 40.

Update to revision 103.
$ svn status
    X   www/green.html
$

As you can see in the previous output, Subversion denotes file externals with the letter E when they are fetched into the working copy, and with the letter X when showing the working copy status.

[Warning]Warning

While directory externals can place the external directory at any depth, and any missing intermediate directories will be created, file externals must be placed into a working copy that is already checked out.

When examining the file external with svn info, you can see the URL and revision the external is coming from:

$ svn info www/green.html 
Path: www/green.html
Name: green.html
Working Copy Root Path: /home/harry/projects/my-project
URL: http://svn.example.com/projects/my-project/trunk/bikeshed/blue.html
Repository Root: http://svn.example.com/projects/my-project
Repository UUID: b2a368dc-7564-11de-bb2b-113435390e17
Revision: 40
Node kind: file
Schedule: normal
Last Changed Author: harry
Last Changed Rev: 40
Last Changed Date: 2009-07-20 20:38:20 +0100 (Mon, 20 Jul 2009)
Text Last Updated: 2009-07-20 23:22:36 +0100 (Mon, 20 Jul 2009)
Checksum: 01a58b04617b92492d99662c3837b33b
$

Because file externals appear in the working copy as versioned files, they can be modified and even committed if they reference a file at the HEAD revision. The committed changes will then appear in the external as well as the file referenced by the external. However, in our example, we pinned the external to an older revision, so attempting to commit the external fails:

$ svn status
M   X   www/green.html
$ svn commit -m "change the color" www/green.html
Sending        www/green.html
svn: E155011: Commit failed (details follow):
svn: E155011: File '/trunk/bikeshed/blue.html' is out of date
$

Keep this in mind when defining file externals. If you need the external to refer to a certain revision of a file you will not be able to modify the external. If you want to be able to modify the external, you cannot specify a revision other than the HEAD revision, which is implied if no revision is specified.

Unfortunately, the support which exists for externals definitions in Subversion remains less than ideal. Both file and directory externals have shortcomings. For either type of external, the local subdirectory part of the definition cannot contain .. parent directory indicators (such as ../../skins/myskin). File externals cannot refer to files from other repositories. A file external's URL must always be in the same repository as the URL that the file external will be inserted into. Also, file externals cannot be moved or deleted. The svn:externals property must be modified instead. However, file externals can be copied.

Perhaps most disappointingly, the working copies created via the externals definition support are still disconnected from the primary working copy (on whose versioned directories the svn:externals property was actually set). And Subversion still truly operates only on nondisjoint working copies. So, for example, if you want to commit changes that you've made in one or more of those external working copies, you must run svn commit explicitly on those working copies—committing on the primary working copy will not recurse into any external ones.

We've already mentioned some of the additional shortcomings of the old svn:externals format and how the newer Subversion 1.5 format improves upon it. But be careful when making use of the new format that you don't inadvertently introduce new problems. For example, while the latest clients will continue to recognize and support the original externals definition format, pre-1.5 clients will not be able to correctly parse the new format. If you change all your externals definitions to the newer format, you effectively force everyone who uses those externals to upgrade their Subversion clients to a version that can parse them. Also, be careful to avoid naively relocating the -rNNN portion of the definition—the older format uses that revision as a peg revision, but the newer format uses it as an operative revision (with a peg revision of HEAD unless otherwise specified; see the section called “Peg and Operative Revisions” for a full explanation of the distinction here).

[Warning]Warning

External working copies are still completely self-sufficient working copies. You can operate directly on them as you would any other working copy. This can be a handy feature, allowing you to examine an external working copy independently of any primary working copy whose svn:externals property caused its instantiation. Be careful, though, that you don't inadvertently modify your external working copy in subtle ways that cause problems. For example, while an externals definition might specify that the external working copy should be held at a particular revision number, if you run svn update directly on the external working copy, Subversion will oblige, and now your external working copy is out of sync with its declaration in the primary working copy. Using svn switch to directly switch the external working copy (or some portion thereof) to another URL could cause similar problems if the contents of the primary working copy are expecting particular contents in the external content.

Besides the svn checkout, svn update, svn switch, and svn export commands which actually manage the disjoint (or disconnected) subdirectories into which externals are checked out, the svn status command also recognizes externals definitions. It displays a status code of X for the disjoint external subdirectories, and then recurses into those subdirectories to display the status of the external items themselves. You can pass the --ignore-externals option to any of these subcommands to disable externals definition processing.

Changelists

It is commonplace for a developer to find himself working at any given time on multiple different, distinct changes to a particular bit of source code. This isn't necessarily due to poor planning or some form of digital masochism. A software engineer often spots bugs in his peripheral vision while working on some nearby chunk of source code. Or perhaps he's halfway through some large change when he realizes the solution he's working on is best committed as several smaller logical units. Often, these logical units aren't nicely contained in some module, safely separated from other changes. The units might overlap, modifying different files in the same module, or even modifying different lines in the same file.

Developers can employ various work methodologies to keep these logical changes organized. Some use separate working copies of the same repository to hold each individual change in progress. Others might choose to create short-lived feature branches in the repository and use a single working copy that is constantly switched to point to one such branch or another. Still others use diff and patch tools to back up and restore uncommitted changes to and from patch files associated with each change. Each of these methods has its pros and cons, and to a large degree, the details of the changes being made heavily influence the methodology used to distinguish them.

Subversion provides a changelists feature that adds yet another method to the mix. Changelists are basically arbitrary labels (currently at most one per file) applied to working copy files for the express purpose of associating multiple files together. Users of many of Google's software offerings are familiar with this concept already. For example, Gmail doesn't provide the traditional folders-based email organization mechanism. In Gmail, you apply arbitrary labels to emails, and multiple emails can be said to be part of the same group if they happen to share a particular label. Viewing only a group of similarly labeled emails then becomes a simple user interface trick. Many other Web 2.0 sites have similar mechanisms—consider the tags used by sites such as YouTube and Flickr, categories applied to blog posts, and so on. Folks understand today that organization of data is critical, but that how that data is organized needs to be a flexible concept. The old files-and-folders paradigm is too rigid for some applications.

Subversion's changelist support allows you to create changelists by applying labels to files you want to be associated with that changelist, remove those labels, and limit the scope of the files on which its subcommands operate to only those bearing a particular label. In this section, we'll look in detail at how to do these things.

Creating and Modifying Changelists

You can create, modify, and delete changelists using the svn changelist command. More accurately, you use this command to set or unset the changelist association of a particular working copy file. A changelist is effectively created the first time you label a file with that changelist; it is deleted when you remove that label from the last file that had it. Let's examine a usage scenario that demonstrates these concepts.

Harry is fixing some bugs in the calculator application's mathematics logic. His work leads him to change a couple of files:

$ svn status
M       integer.c
M       mathops.c
$

While testing his bug fix, Harry notices that his changes bring to light a tangentially related bug in the user interface logic found in button.c. Harry decides that he'll go ahead and fix that bug, too, as a separate commit from his math fixes. Now, in a small working copy with only a handful of files and few logical changes, Harry can probably keep his two logical change groupings mentally organized without any problem. But today he's going to use Subversion's changelists feature as a special favor to the authors of this book.

Harry first creates a changelist and associates with it the two files he's already changed. He does this by using the svn changelist command to assign the same arbitrary changelist name to those files:

$ svn changelist math-fixes integer.c mathops.c
A [math-fixes] integer.c
A [math-fixes] mathops.c
$ svn status

--- Changelist 'math-fixes':
M       integer.c
M       mathops.c
$

As you can see, the output of svn status reflects this new grouping.

Harry now sets off to fix the secondary UI problem. Since he knows which file he'll be changing, he assigns that path to a changelist, too. Unfortunately, Harry carelessly assigns this third file to the same changelist as the previous two files:

$ svn changelist math-fixes button.c
A [math-fixes] button.c
$ svn status

--- Changelist 'math-fixes':
        button.c
M       integer.c
M       mathops.c
$

Fortunately, Harry catches his mistake. At this point, he has two options. He can remove the changelist association from button.c, and then assign a different changelist name:

$ svn changelist --remove button.c
D [math-fixes] button.c
$ svn changelist ui-fix button.c
A [ui-fix] button.c
$

Or, he can skip the removal and just assign a new changelist name. In this case, Subversion will first warn Harry that button.c is being removed from the first changelist:

$ svn changelist ui-fix button.c
D [math-fixes] button.c
A [ui-fix] button.c
$ svn status

--- Changelist 'ui-fix':
        button.c

--- Changelist 'math-fixes':
M       integer.c
M       mathops.c
$

Harry now has two distinct changelists present in his working copy, and svn status will group its output according to these changelist determinations. Notice that even though Harry hasn't yet modified button.c, it still shows up in the output of svn status as interesting because it has a changelist assignment. Changelists can be added to and removed from files at any time, regardless of whether they contain local modifications.

Harry now fixes the user interface problem in button.c.

$ svn status

--- Changelist 'ui-fix':
M       button.c

--- Changelist 'math-fixes':
M       integer.c
M       mathops.c
$

Changelists As Operation Filters

The visual grouping that Harry sees in the output of svn status as shown in our previous section is nice, but not entirely useful. The status command is but one of many operations that he might wish to perform on his working copy. Fortunately, many of Subversion's other operations understand how to operate on changelists via the use of the --changelist option.

When provided with a --changelist option, Subversion commands will limit the scope of their operation to only those files to which a particular changelist name is assigned. If Harry now wants to see the actual changes he's made to the files in his math-fixes changelist, he could explicitly list only the files that make up that changelist on the svn diff command line.

$ svn diff integer.c mathops.c
Index: integer.c
===================================================================
--- integer.c	(revision 1157)
+++ integer.c	(working copy)
…
Index: mathops.c
===================================================================
--- mathops.c	(revision 1157)
+++ mathops.c	(working copy)
…
$

That works okay for a few files, but what if Harry's change touched 20 or 30 files? That would be an annoyingly long list of explicitly named files. Now that he's using changelists, though, Harry can avoid explicitly listing the set of files in his changelist from now on, and instead provide just the changelist name:

$ svn diff --changelist math-fixes
Index: integer.c
===================================================================
--- integer.c	(revision 1157)
+++ integer.c	(working copy)
…
Index: mathops.c
===================================================================
--- mathops.c	(revision 1157)
+++ mathops.c	(working copy)
…
$

And when it's time to commit, Harry can again use the --changelist option to limit the scope of the commit to files in a certain changelist. He might commit his user interface fix by doing the following:

$ svn commit -m "Fix a UI bug found while working on math logic." \
             --changelist ui-fix
Sending        button.c
Transmitting file data .
Committed revision 1158.
$

In fact, the svn commit command provides a second changelists-related option: --keep-changelists. Normally, changelist assignments are removed from files after they are committed. But if --keep-changelists is provided, Subversion will leave the changelist assignment on the committed (and now unmodified) files. In any case, committing files assigned to one changelist leaves other changelists undisturbed.

$ svn status

--- Changelist 'math-fixes':
M       integer.c
M       mathops.c
$
[Note]Note

The --changelist option acts only as a filter for Subversion command targets, and will not add targets to an operation. For example, on a commit operation specified as svn commit /path/to/dir, the target is the directory /path/to/dir and its children (to infinite depth). If you then add a changelist specifier to that command, only those files in and under /path/to/dir that are assigned that changelist name will be considered as targets of the commit—the commit will not include files located elsewhere (such as in /path/to/another-dir), regardless of their changelist assignment, even if they are part of the same working copy as the operation's target(s).

Even the svn changelist command accepts the --changelist option. This allows you to quickly and easily rename or remove a changelist:

$ svn changelist math-bugs --changelist math-fixes --depth infinity .
D [math-fixes] integer.c
A [math-bugs] integer.c
D [math-fixes] mathops.c
A [math-bugs] mathops.c
$ svn changelist --remove --changelist math-bugs --depth infinity .
D [math-bugs] integer.c
D [math-bugs] mathops.c
$

Finally, you can specify multiple instances of the --changelist option on a single command line. Doing so limits the operation you are performing to files found in any of the specified changesets.

Changelist Limitations

Subversion's changelist feature is a handy tool for grouping working copy files, but it does have a few limitations. Changelists are artifacts of a particular working copy, which means that changelist assignments cannot be propagated to the repository or otherwise shared with other users. Changelists can be assigned only to files—Subversion doesn't currently support the use of changelists with directories. Finally, you can have at most one changelist assignment on a given working copy file. Here is where the blog post category and photo service tag analogies break down—if you find yourself needing to assign a file to multiple changelists, you're out of luck.

Network Model

At some point, you're going to need to understand how your Subversion client communicates with its server. Subversion's networking layer is abstracted, meaning that Subversion clients exhibit the same general behaviors no matter what sort of server they are operating against. Whether speaking the HTTP protocol (http://) with the Apache HTTP Server or speaking the custom Subversion protocol (svn://) with svnserve, the basic network model is the same. In this section, we'll explain the basics of that network model, including how Subversion manages authentication and authorization matters.

Requests and Responses

The Subversion client spends most of its time managing working copies. When it needs information from a remote repository, however, it makes a network request, and the server responds with an appropriate answer. The details of the network protocol are hidden from the user—the client attempts to access a URL, and depending on the URL scheme, a particular protocol is used to contact the server (see the section called “Addressing the Repository”).

[Tip]Tip

Run svn --version to see which URL schemes and protocols the client knows how to use.

When the server process receives a client request, it often demands that the client identify itself. It issues an authentication challenge to the client, and the client responds by providing credentials back to the server. Once authentication is complete, the server responds with the original information that the client asked for. Notice that this system is different from systems such as CVS, where the client preemptively offers credentials (logs in) to the server before ever making a request. In Subversion, the server pulls credentials by challenging the client at the appropriate moment, rather than the client pushing them. This makes certain operations more elegant. For example, if a server is configured to allow anyone in the world to read a repository, the server will never issue an authentication challenge when a client attempts to svn checkout.

If the particular network requests issued by the client result in a new revision being created in the repository (e.g., svn commit), Subversion uses the authenticated username associated with those requests as the author of the revision. That is, the authenticated user's name is stored as the value of the svn:author property on the new revision (see the section called “Subversion Properties” in Chapter 9, Subversion Complete Reference). If the client was not authenticated (i.e., if the server never issued an authentication challenge), the revision's svn:author property is empty.

Client Credentials

Many Subversion servers are configured to require authentication. Sometimes anonymous read operations are allowed, while write operations must be authenticated. In other cases, reads and writes alike require authentication. Subversion's different server options understand different authentication protocols, but from the user's point of view, authentication typically boils down to usernames and passwords. Subversion clients offer several different ways to retrieve and store a user's authentication credentials, from interactive prompting for usernames and passwords to encrypted and non-encrypted on-disk data caches.

The security-conscious reader will suspect immediately that there is reason for concern here. Caching passwords on disk? That's terrible! You should never do that! Don't worry—it's not as bad as it sounds. The following sections discuss the various types of credential caches that Subversion uses, when it uses them, and how to disable that functionality in whole or in part.

Caching credentials

Subversion offers a remedy for the annoyance caused when users are forced to type their usernames and passwords over and over again. By default, whenever the command-line client successfully responds to a server's authentication challenge, credentials are cached on disk and keyed on a combination of the server's hostname, port, and authentication realm. This cache will then be automatically consulted in the future, avoiding the need for the user to re-type his or her authentication credentials. If seemingly suitable credentials are not present in the cache, or if the cached credentials ultimately fail to authenticate, the client will, by default, fall back to prompting the user for the necessary information.

The Subversion developers recognize that on-disk caches of authentication credentials can be a security risk. To offset this, Subversion works with available mechanisms provided by the operating system and environment to try to minimize the risk of leaking this information.

  • On Windows, the Subversion client stores passwords in the %APPDATA%/Subversion/auth/ directory. On Windows 2000 and later, the standard Windows cryptography services are used to encrypt the password on disk. Because the encryption key is managed by Windows and is tied to the user's own login credentials, only the user can decrypt the cached password. (Note that if the user's Windows account password is reset by an administrator, all of the cached passwords become undecipherable. The Subversion client will behave as though they don't exist, prompting for passwords when required.)

  • Similarly, on Mac OS X, the Subversion client stores all repository passwords in the login keyring (managed by the Keychain service), which is protected by the user's account password. User preference settings can impose additional policies, such as requiring that the user's account password be entered each time the Subversion password is used.

  • For other Unix-like operating systems, no single standard keychain service exists. However, the Subversion client knows how to store passwords securely using the GNOME Keyring and KDE Wallet services. Also, before storing unencrypted passwords in the ~/.subversion/auth/ caching area, the Subversion client will ask the user for permission to do so. Note that the auth/ caching area is still permission-protected so that only the user (owner) can read data from it, not the world at large. The operating system's own file permissions protect the passwords from other non-administrative users on the same system, provided they have no direct physical access to the storage media of the home directory, or backups thereof.

Of course, for the truly paranoid, none of these mechanisms meets the test of perfection. So for those folks willing to sacrifice convenience for the ultimate in security, Subversion provides various ways of disabling its credentials caching system altogether.

Disabling password caching

When you perform a Subversion operation that requires you to authenticate, by default Subversion tries to cache your authentication credentials on disk in encrypted form. On some systems, Subversion may be unable to encrypt your authentication data. In those situations, Subversion will ask whether you want to cache your credentials to disk in plaintext:

$ svn checkout https://host.example.com:443/svn/private-repo
-----------------------------------------------------------------------
ATTENTION!  Your password for authentication realm:

   <https://host.example.com:443> Subversion Repository

can only be stored to disk unencrypted!  You are advised to configure
your system so that Subversion can store passwords encrypted, if
possible.  See the documentation for details.

You can avoid future appearances of this warning by setting the value
of the 'store-plaintext-passwords' option to either 'yes' or 'no' in
'/tmp/servers'.
-----------------------------------------------------------------------
Store password unencrypted (yes/no)? 

If you want the convenience of not having to continually reenter your password for future operations, you can answer yes to this prompt. If you're concerned about caching your Subversion passwords in plaintext and do not want to be asked about it again and again, you can disable caching of plaintext passwords either permanently, or on a server-per-server basis.

[Warning]Warning

When considering how to use Subversion's password caching system, you'll want to consult any governing policies that are in place for your client computer—many companies have strict rules about the ways that their employees' authentication credentials should be stored.

To permanently disable caching of passwords in plaintext, add the line store-plaintext-passwords = no to the [global] section in the servers configuration file on the local machine. To disable plaintext password caching for a particular server, use the same setting in the appropriate group section in the servers configuration file. (See the section called “Configuration Options” in Chapter 7, Customizing Your Subversion Experience for details.)

To disable password caching entirely for any single Subversion command-line operation, pass the --no-auth-cache option to that command line. To permanently disable caching entirely, add the line store-passwords = no to your local machine's Subversion configuration file.

Removing cached credentials

Sometimes users will want to remove specific credentials from the disk cache. To do this, you need to navigate into the auth/ area and manually delete the appropriate cache file. Credentials are cached in individual files; if you look inside each file, you will see keys and values. The svn:realmstring key describes the particular server realm that the file is associated with:

$ ls ~/.subversion/auth/svn.simple/
5671adf2865e267db74f09ba6f872c28
3893ed123b39500bca8a0b382839198e
5c3c22968347b390f349ff340196ed39

$ cat ~/.subversion/auth/svn.simple/5671adf2865e267db74f09ba6f872c28

K 8
username
V 3
joe
K 8
password
V 4
blah
K 15
svn:realmstring
V 45
<https://svn.domain.com:443> Joe's repository
END

Once you have located the proper cache file, just delete it.

Command-line authentication

All Subversion command-line operations accept the --username and --password options, which allow you to specify your username and password, respectively, so that Subversion isn't forced to prompt you for that information. This is especially handy if you need to invoke Subversion from a script and cannot rely on Subversion being able to locate valid cached credentials for you. These options are also helpful when Subversion has already cached authentication credentials for you, but you know they aren't the ones you want it to use. Perhaps several system users share a login to the system, but each have distinct Subversion identities. You can omit the --password option from this pair if you wish Subversion to use only the provided username, but still prompt you for that username's password.

Authentication wrap-up

One last word about svn's authentication behavior, specifically regarding the --username and --password options. Many client subcommands accept these options, but it is important to understand that using these options does not automatically send credentials to the server. As discussed earlier, the server pulls credentials from the client when it deems necessary; the client cannot push them at will. If a username and/or password are passed as options, they will be presented to the server only if the server requests them. These options are typically used to authenticate as a different user than Subversion would have chosen by default (such as your system login name) or when trying to avoid interactive prompting (such as when calling svn from a script).

[Note]Note

A common mistake is to misconfigure a server so that it never issues an authentication challenge. When users pass --username and --password options to the client, they're surprised to see that they're never used; that is, new revisions still appear to have been committed anonymously!

Here is a final summary that describes how a Subversion client behaves when it receives an authentication challenge.

  1. First, the client checks whether the user specified any credentials as command-line options (--username and/or --password). If so, the client will try to use those credentials to authenticate against the server.

  2. If no command-line credentials were provided, or the provided ones were invalid, the client looks up the server's hostname, port, and realm in the runtime configuration's auth/ area, to see whether appropriate credentials are cached there. If so, it attempts to use those credentials to authenticate.

  3. Finally, if the previous mechanisms failed to successfully authenticate the user against the server, the client resorts to interactively prompting the user for valid credentials (unless instructed not to do so via the --non-interactive option or its client-specific equivalents).

If the client successfully authenticates by any of these methods, it will attempt to cache the credentials on disk (unless the user has disabled this behavior, as mentioned earlier).

Summary

After reading this chapter, you should have a firm grasp on some of Subversion's features that, while perhaps not used every time you interact with your version control system, are certainly handy to know about. But don't stop here! Read on to the following chapter, where you'll learn about branches, tags, and merging. Then you'll have nearly full mastery of the Subversion client. Though our lawyers won't allow us to promise you anything, this additional knowledge could make you measurably more cool.[24]




[11] You're not supposed to name it. Once you name it, you start getting attached to it.—Mike Wazowski

[12] 606 N. Main Street, Wheaton, Illinois, is the home of the Wheaton History Center. It seemed appropriate….

[13] If you're familiar with XML, this is pretty much the ASCII subset of the syntax for XML Name.

[14] Fixing spelling errors, grammatical gotchas, and just-plain-wrongness in commit log messages is perhaps the most common use case for the --revprop option.

[15] Currently, libmagic is the support library used to accomplish this.

[16] You think that was rough? During that same era, WordPerfect also used .DOC for their proprietary file format's preferred extension!

[17] The Windows filesystems use file extensions (such as .EXE, .BAT, and .COM) to denote executable files.

[18] Isn't that the whole point of a build system?

[19] … or maybe even a section of a book …

[20] Safely, of course. As in other situations, Subversion will leave on disk any files you've modified or which aren't versioned.

[21] Communication wouldn't have been such bad medicine for Harry and Sally's Hollywood namesakes, either, for that matter.

[22] Subversion does not currently allow locks on directories.

[23] Except, perhaps, a classic Vulcan mind-meld.

[24] No purchase necessary. Certains terms and conditions apply. No guarantee of coolness—implicit or otherwise—exists. Mileage may vary.

Chapter 4. Branching and Merging

 

君子务本 (It is upon the Trunk that a gentleman works.)

 
 --Confucius

Branching and merging are fundamental aspects of version control, simple enough to explain conceptually but offering just enough complexity and nuance to merit their own chapter in this book. Herein, we'll introduce you to the general ideas behind these operations as well as Subversion's somewhat unique approach to them. If you've not familiarized yourself with Subversion's basic concepts (found in Chapter 1, Fundamental Concepts), we recommmend that you do so before reading this chapter.

What's a Branch?

Suppose it's your job to maintain a document for a division in your company—a handbook of some sort. One day a different division asks you for the same handbook, but with a few parts tweaked for them, since they do things slightly differently.

What do you do in this situation? You do the obvious: make a second copy of your document and begin maintaining the two copies separately. As each department asks you to make small changes, you incorporate them into one copy or the other.

You often want to make the same change to both copies. For example, if you discover a typo in the first copy, it's very likely that the same typo exists in the second copy. The two documents are almost the same, after all; they differ only in small, specific ways.

This is the basic concept of a branch—namely, a line of development that exists independently of another line, yet still shares a common history if you look far enough back in time. A branch always begins life as a copy of something, and moves on from there, generating its own history (see Figure 4.1, “Branches of development”).

Figure 4.1. Branches of development

Branches of development


Subversion has commands to help you maintain parallel branches of your files and directories. It allows you to create branches by copying your data, and remembers that the copies are related to one another. It also helps you duplicate changes from one branch to another. Finally, it can make portions of your working copy reflect different branches so that you can mix and match different lines of development in your daily work.

Using Branches

At this point, you should understand how each commit creates a new state of the filesystem tree (called a revision) in the repository. If you don't, go back and read about revisions in the section called “Revisions”.

Let's revisit the example from Chapter 1, Fundamental Concepts. Remember that you and your collaborator, Sally, are sharing a repository that contains two projects, paint and calc. Notice that in Figure 4.2, “Starting repository layout”, however, each project directory now contains subdirectories named trunk and branches. The reason for this will soon become clear.

Figure 4.2. Starting repository layout

Starting repository layout


As before, assume that Sally and you both have working copies of the calc project. Specifically, you each have a working copy of /calc/trunk. All the files for the project are in this subdirectory rather than in /calc itself, because your team has decided that /calc/trunk is where the main line of development is going to take place.

Let's say that you've been given the task of implementing a large software feature. It will take a long time to write, and will affect all the files in the project. The immediate problem is that you don't want to interfere with Sally, who is in the process of fixing small bugs here and there. She's depending on the fact that the latest version of the project (in /calc/trunk) is always usable. If you start committing your changes bit by bit, you'll surely break things for Sally (and other team members as well).

One strategy is to crawl into a hole: you can stop sharing information for a week or two, gutting and reorganizing all the files in your private working copy but not committing or updating until you're completely finished with your task. There are a number of problems with this, though. First, it's not very safe. Should something bad happen to your working copy or computer, you risk losing all your changes. Second, it's not very flexible. Unless you manually replicate your changes across different working copies or computers, you're stuck trying to make your changes in a single working copy. Similarly, it's difficult to share your work-in-progress with anyone else. A common software development best practice is to allow your peers to review your work as you go. If nobody sees your intermediate commits, you lose potential feedback and may end up going down the wrong path for weeks before another person on your team notices. Finally, when you're finished with all your changes, you might find it very difficult to merge your completed work with the rest of the company's main body of code. Sally (or others) may have made many other changes in the repository that are difficult to incorporate into your working copy when you eventually run svn update after weeks of isolation.

The better solution is to create your own branch, or line of development, in the repository. This allows you to save your not-yet-completed work frequently without interfering with others' changes and while still selectively sharing information with your collaborators. You'll see exactly how this works as we continue.

Creating a Branch

Creating a branch is very simple—you make a copy of your project tree in the repository using the svn copy command. Since your project's source code is rooted in the /calc/trunk directory, it's that directory that you'll copy. Where should the new copy live? Wherever you wish. The repository location in which branches are stashed is left by Subversion as a matter of project policy. Finally, your branch will need a name to distinguish it from other branches. Once again, the name you choose is unimportant to Subversion—you can use whatever name works best for you and your team.

Let's assume that your team (like most) has a policy of creating branches in the branches directory that is a sibling of the project's trunk (the /calc/branches directory in our scenario). Lacking inspiration, you settle on my-calc-branch as the name you wish to give your branch. This means that you'll create a new directory, /calc/branches/my-calc-branch, which begins its life as a copy of /calc/trunk.

You may already have seen svn copy used to copy one file to another within a working copy. But it can also be used to do a remote copy—a copy that immediately results in a newly committed repository revision and for which no working copy is required at all. Just copy one URL to another:

$ svn copy http://svn.example.com/repos/calc/trunk \
           http://svn.example.com/repos/calc/branches/my-calc-branch \
           -m "Creating a private branch of /calc/trunk."

Committed revision 341.
$

This command causes a near-instantaneous commit in the repository, creating a new directory in revision 341. The new directory is a copy of /calc/trunk. This is shown in Figure 4.3, “Repository with new copy”.[25] While it's also possible to create a branch by using svn copy to duplicate a directory within the working copy, this technique isn't recommended. It can be quite slow, in fact! Copying a directory on the client side is a linear-time operation, in that it actually has to duplicate every file and subdirectory within that working copy directory on the local disk. Copying a directory on the server, however, is a constant-time operation, and it's the way most people create branches.

Figure 4.3. Repository with new copy

Repository with new copy


Working with Your Branch

Now that you've created a branch of the project, you can check out a new working copy to start using it:

$ svn checkout http://svn.example.com/repos/calc/branches/my-calc-branch
A  my-calc-branch/Makefile
A  my-calc-branch/integer.c
A  my-calc-branch/button.c
Checked out revision 341.
$

There's nothing special about this working copy; it simply mirrors a different directory in the repository. When you commit changes, however, Sally won't see them when she updates, because her working copy is of /calc/trunk. (Be sure to read the section called “Traversing Branches” later in this chapter: the svn switch command is an alternative way of creating a working copy of a branch.)

Let's pretend that a week goes by, and the following commits happen:

  • You make a change to /calc/branches/my-calc-branch/button.c, which creates revision 342.

  • You make a change to /calc/branches/my-calc-branch/integer.c, which creates revision 343.

  • Sally makes a change to /calc/trunk/integer.c, which creates revision 344.

Now two independent lines of development (shown in Figure 4.4, “The branching of one file's history”) are happening on integer.c.

Figure 4.4. The branching of one file's history

The branching of one file's history


Things get interesting when you look at the history of changes made to your copy of integer.c:

$ pwd
/home/user/my-calc-branch

$ svn log -v integer.c
------------------------------------------------------------------------
r343 | user | 2002-11-07 15:27:56 -0600 (Thu, 07 Nov 2002) | 2 lines
Changed paths:
   M /calc/branches/my-calc-branch/integer.c

* integer.c:  frozzled the wazjub.

------------------------------------------------------------------------
r341 | user | 2002-11-03 15:27:56 -0600 (Thu, 07 Nov 2002) | 2 lines
Changed paths:
   A /calc/branches/my-calc-branch (from /calc/trunk:340)

Creating a private branch of /calc/trunk.

------------------------------------------------------------------------
r303 | sally | 2002-10-29 21:14:35 -0600 (Tue, 29 Oct 2002) | 2 lines
Changed paths:
   M /calc/trunk/integer.c

* integer.c:  changed a docstring.

------------------------------------------------------------------------
r98 | sally | 2002-02-22 15:35:29 -0600 (Fri, 22 Feb 2002) | 2 lines
Changed paths:
   A /calc/trunk/integer.c

* integer.c:  adding this file to the project.

------------------------------------------------------------------------

Notice that Subversion is tracing the history of your branch's integer.c all the way back through time, even traversing the point where it was copied. It shows the creation of the branch as an event in the history, because integer.c was implicitly copied when all of /calc/trunk/ was copied. Now look at what happens when Sally runs the same command on her copy of the file:

$ pwd
/home/sally/calc

$ svn log -v integer.c
------------------------------------------------------------------------
r344 | sally | 2002-11-07 15:27:56 -0600 (Thu, 07 Nov 2002) | 2 lines
Changed paths:
   M /calc/trunk/integer.c

* integer.c:  fix a bunch of spelling errors.

------------------------------------------------------------------------
r303 | sally | 2002-10-29 21:14:35 -0600 (Tue, 29 Oct 2002) | 2 lines
Changed paths:
   M /calc/trunk/integer.c

* integer.c:  changed a docstring.

------------------------------------------------------------------------
r98 | sally | 2002-02-22 15:35:29 -0600 (Fri, 22 Feb 2002) | 2 lines
Changed paths:
   A /calc/trunk/integer.c

* integer.c:  adding this file to the project.

------------------------------------------------------------------------

Sally sees her own revision 344 change, but not the change you made in revision 343. As far as Subversion is concerned, these two commits affected different files in different repository locations. However, Subversion does show that the two files share a common history. Before the branch copy was made in revision 341, the files used to be the same file. That's why you and Sally both see the changes made in revisions 303 and 98.

The Key Concepts Behind Branching

You should remember two important lessons from this section. First, Subversion has no internal concept of a branch—it knows only how to make copies. When you copy a directory, the resultant directory is only a branch because you attach that meaning to it. You may think of the directory differently, or treat it differently, but to Subversion it's just an ordinary directory that happens to carry some extra historical information.

Second, because of this copy mechanism, Subversion's branches exist as normal filesystem directories in the repository. This is different from other version control systems, where branches are typically defined by adding extra-dimensional labels to collections of files. The location of your branch directory doesn't matter to Subversion. Most teams follow a convention of putting all branches into a /branches directory, but you're free to invent any policy you wish.

Basic Merging

Now you and Sally are working on parallel branches of the project: you're working on a private branch, and Sally is working on the trunk, or main line of development.

For projects that have a large number of contributors, it's common for most people to have working copies of the trunk. Whenever someone needs to make a long-running change that is likely to disrupt the trunk, a standard procedure is to create a private branch and commit changes there until all the work is complete.

So, the good news is that you and Sally aren't interfering with each other. The bad news is that it's very easy to drift too far apart. Remember that one of the problems with the crawl in a hole strategy is that by the time you're finished with your branch, it may be near-impossible to merge your changes back into the trunk without a huge number of conflicts.

Instead, you and Sally might continue to share changes as you work. It's up to you to decide which changes are worth sharing; Subversion gives you the ability to selectively copy changes between branches. And when you're completely finished with your branch, your entire set of branch changes can be copied back into the trunk. In Subversion terminology, the general act of replicating changes from one branch to another is called merging, and it is performed using various invocations of the svn merge subcommand.

In the examples that follow, we're assuming that both your Subversion client and server are running Subversion 1.7 (or later). If either client or server is older than version 1.5, things are more complicated: the system won't track changes automatically, forcing you to use painful manual methods to achieve similar results. That is, you'll always need to use the detailed merge syntax to specify specific ranges of revisions to replicate (see the section called “Merge Syntax: Full Disclosure” later in this chapter), and take special care to keep track of what's already been merged and what hasn't. For this reason, we strongly recommend that you make sure your client and server are at least at version 1.5.

Changesets

Before we proceed further, we should warn you that there's a lot of discussion of changes in the pages ahead. A lot of people experienced with version control systems use the terms change and changeset interchangeably, and we should clarify what Subversion understands as a changeset.

Everyone seems to have a slightly different definition of changeset, or at least a different expectation of what it means for a version control system to have one. For our purposes, let's say that a changeset is just a collection of changes with a unique name. The changes might include textual edits to file contents, modifications to tree structure, or tweaks to metadata. In more common speak, a changeset is just a patch with a name you can refer to.

In Subversion, a global revision number N names a tree in the repository: it's the way the repository looked after the Nth commit. It's also the name of an implicit changeset: if you compare tree N with tree N-1, you can derive the exact patch that was committed. For this reason, it's easy to think of revision N as not just a tree, but a changeset as well. If you use an issue tracker to manage bugs, you can use the revision numbers to refer to particular patches that fix bugs—for example, this issue was fixed by r9238. Somebody can then run svn log -r 9238 to read about the exact changeset that fixed the bug, and run svn diff -c 9238 to see the patch itself. And (as you'll see shortly) Subversion's svn merge command is able to use revision numbers. You can merge specific changesets from one branch to another by naming them in the merge arguments: passing -c 9238 to svn merge would merge changeset r9238 into your working copy.

Keeping a Branch in Sync

Continuing with our running example, let's suppose that a week has passed since you started working on your private branch. Your new feature isn't finished yet, but at the same time you know that other people on your team continue to make important changes in the project's /trunk. It's in your best interest to replicate those changes to your own branch, just to make sure they mesh well with your changes. This is done by performing a sync merge—a merge operation designed to bring your branch up to date with any changes made to its ancestral parent branch since your branch was created.

[Tip]Tip

Frequently keeping your branch in sync with the main development line helps prevent surprise conflicts when the time comes for you to fold your changes back into the trunk.

Subversion is aware of the history of your branch and knows when it split away from the trunk. To perform a sync merge, first make sure your working copy of the branch is clean—that it has no local modifications reported by svn status. Then simply run:

$ pwd
/home/user/my-calc-branch

$ svn merge ^/calc/trunk
--- Merging r345 through r356 into '.':
U    button.c
U    integer.c
--- Recording mergeinfo for merge of r345 through r356 into '.':
 U   .
$

This basic syntax—svn merge URL—tells Subversion to merge all changes which have not been previously merged from the URL to the current working directory (which is typically the root of your working copy). Notice that we're using the caret (^) syntax[26] to avoid having to type out the entire /trunk URL. Also note the Recording mergeinfo for merge… notification. This tells you that the merge is updating the svn:mergeinfo property. We'll discuss both this property and these notifications later in this chapter, in the section called “Mergeinfo and Previews”.

[Tip]Tip

In this book and elsewhere (Subversion mailing lists, articles on merge tracking, etc.) you will frequently come across the term mergeinfo. This is simply shorthand for the svn:mergeinfo property.

After running the prior example, your branch working copy now contains new local modifications, and these edits are duplications of all of the changes that have happened on the trunk since you first created your branch:

$ svn status
 M      .
M       button.c
M       integer.c
$

At this point, the wise thing to do is look at the changes carefully with svn diff, and then build and test your branch. Notice that the current working directory (.) has also been modified; svn diff will show that its svn:mergeinfo property has been either created or modified. This is important merge-related metadata that you should not touch, since it is needed by future svn merge commands. (We'll learn more about this metadata later in the chapter.)

After performing the merge, you might also need to resolve some conflicts—just as you do with svn update—or possibly make some small edits to get things working properly. (Remember, just because there are no syntactic conflicts doesn't mean there aren't any semantic conflicts!) If you encounter serious problems, you can always abort the local changes by running svn revert . -R (which will undo all local modifications) and starting a long what's going on? discussion with your collaborators. If things look good, however, you can submit these changes into the repository:

$ svn commit -m "Merged latest trunk changes to my-calc-branch."
Sending        .
Sending        button.c
Sending        integer.c
Transmitting file data ..
Committed revision 357.
$

At this point, your private branch is now in sync with the trunk, so you can rest easier knowing that as you continue to work in isolation, you're not drifting too far away from what everyone else is doing.

Suppose that another week has passed. You've committed more changes to your branch, and your comrades have continued to improve the trunk as well. Once again, you want to replicate the latest trunk changes to your branch and bring yourself in sync. Just run the same merge command again!

$ svn merge ^/calc/trunk
svn: E195020: Cannot merge into mixed-revision working copy [357:378]; try up\
dating first
$

Well that was unexpected! After making changes to your branch over the past week you now find yourself with a working copy that contains a mixture of revisions (see the section called “Mixed-revision working copies”). With the release of Subversion 1.7 the svn merge subcommand disables merges into mixed-revision working copies by default. Without going into too much detail, this is because of limitations in the way merges are tracked by the svn:mergeinfo property (see the section called “Mergeinfo and Previews” for details). These limitations mean that merges into mixed-revision working copies can result in unexpected text and tree conflicts.[27] We don't want any needless conflicts, so we update the working copy and then reattempt the merge.

$ svn up
Updating '.':
At revision 380.

$ svn merge ^/calc/trunk
--- Merging r357 through r380 into '.':
U    integer.c
U    Makefile
A    README
--- Recording mergeinfo for merge of r357 through r380 into '.':
 U   .
$

Subversion knows which trunk changes you previously replicated to your branch, so it carefully replicates only those changes you don't yet have. And once again, you build, test, and svn commit the local modifications to your branch.

[Tip]Tip

Prior to Subversion 1.7, merges unconditionally updated all of the subtree mergeinfo under the target to describe the merge. For users with a lot of subtree mergeinfo this meant that relatively simple merges (e.g. one which applied a diff to only a single file) resulted in changes to every subtree with mergeinfo, even those that were not parents of the effected path(s). This caused some level of confusion and frustration. Subversion 1.7 addresses this problem by only updating the mergeinfo on subtrees which are parents of the paths modified by the merge (i.e. paths changed, added, or deleted by application of the difference, see the section called “Merge Syntax: Full Disclosure”). The one exception to this behavior regards the actual merge target; the merge target's mergeinfo is always updated to describe the merge, even if the applied difference made no changes.

Reintegrating a Branch

What happens when you finally finish your work, though? Your new feature is done, and you're ready to merge your branch changes back to the trunk (so your team can enjoy the bounty of your labor). The process is simple. First, bring your branch into sync with the trunk again, just as you've been doing all along[28]:

$ svn merge ^/calc/trunk
--- Merging r381 through r385 into '.':
U    button.c
U    README
--- Recording mergeinfo for merge of r381 through r385 into '.':
 U   .

$ # build, test, ...

$ svn commit -m "Final merge of trunk changes to my-calc-branch."
Sending        .
Sending        button.c
Sending        README
Transmitting file data ..
Committed revision 390.

Now, use svn merge with the --reintegrate option to replicate your branch changes back into the trunk. You'll need a working copy of /trunk. You can get one by doing an svn checkout, dredging up an old trunk working copy from somewhere on your disk, or using svn switch (see the section called “Traversing Branches”). Your trunk working copy cannot have any local edits or contain a mixture of revisions (see the section called “Mixed-revision working copies”). While these are typically best practices for merging anyway, they are required when using the --reintegrate option.

Once you have a clean working copy of the trunk, you're ready to merge your branch back into it:

$ pwd
/home/user/calc-trunk

$ svn update  # (make sure the working copy is up to date)
Updating '.':
At revision 390.

$ svn merge --reintegrate ^/calc/branches/my-calc-branch
--- Merging differences between repository URLs into '.':
U    button.c
U    integer.c
U    Makefile
--- Recording mergeinfo for merge between repository URLs into '.':
 U   .

$ # build, test, verify, ...

$ svn commit -m "Merge my-calc-branch back into trunk!"
Sending        .
Sending        button.c
Sending        integer.c
Sending        Makefile
Transmitting file data ..
Committed revision 391.

Congratulations, your branch-specific changes have now been merged back into the main line of development. Notice our use of the --reintegrate option this time around. The option is critical for reintegrating changes from a branch back into its original line of development—don't forget it! It's needed because this sort of merge back is a different sort of work than what you've done up until now. Previously, we were asking svn merge to grab the next set of changes from one line of development (the trunk) and duplicate them to another (your branch). This is fairly straightforward, and each time Subversion knows how to pick up where it left off. In our prior examples, you can see that first it merges the ranges 345:356 from trunk to branch; later on, it continues by merging the next contiguously available range, 356:380. When doing the final sync, it merges the range 380:385.

When merging your branch back to the trunk, however, the underlying mathematics are quite different. Your feature branch is now a mishmash of both duplicated trunk changes and private branch changes, so there's no simple contiguous range of revisions to copy over. By specifying the --reintegrate option, you're asking Subversion to carefully replicate only those changes unique to your branch. (And in fact, it does this by comparing the latest trunk tree with the latest branch tree: the resulting difference is exactly your branch changes!)

Keep in mind that the --reintegrate option is quite specialized in contrast to the more general nature of most Subversion subcommand options. It supports the use case described above, but has little applicability outside of that. Because of this narrow focus, in addition to requiring an up-to-date working copy[29] with no mixed-revisions, it will not function in combination with most of the other svn merge options. You'll get an error if you use any non-global options but these: --accept, --dry-run, --diff3-cmd, --extensions, or --quiet.

Now that your private branch is merged to trunk, you may wish to remove it from the repository:

$ svn delete ^/calc/branches/my-calc-branch \
             -m "Remove my-calc-branch, reintegrated with trunk in r391."
Committed revision 392.

But wait! Isn't the history of that branch valuable? What if somebody wants to audit the evolution of your feature someday and look at all of your branch changes? No need to worry. Remember that even though your branch is no longer visible in the /branches directory, its existence is still an immutable part of the repository's history. A simple svn log command on the /branches URL will show the entire history of your branch. Your branch can even be resurrected at some point, should you desire (see the section called “Resurrecting Deleted Items”).

Once a --reintegrate merge is done from branch to trunk, the branch is no longer usable for further work. It's not able to correctly absorb new trunk changes, nor can it be properly reintegrated to trunk again. For this reason, if you want to keep working on your feature branch, we recommend destroying it and then re-creating it from the trunk:

$ svn delete http://svn.example.com/repos/calc/branches/my-calc-branch \
             -m "Remove my-calc-branch, reintegrated with trunk in r391."
Committed revision 392.

$ svn copy http://svn.example.com/repos/calc/trunk \
           http://svn.example.com/repos/calc/branches/my-calc-branch \
           -m "Recreate my-calc-branch from trunk@HEAD."
Committed revision 393.

There is another way of making the branch usable again after reintegration, without deleting the branch. See the section called “Keeping a Reintegrated Branch Alive”.

Mergeinfo and Previews

The basic mechanism Subversion uses to track changesets—that is, which changes have been merged to which branches—is by recording data in versioned properties. Specifically, merge data is tracked in the svn:mergeinfo property attached to files and directories. (If you're not familiar with Subversion properties, see the section called “Properties”.)

You can examine the property, just like any other:

$ cd my-calc-branch
$ svn propget svn:mergeinfo .
/trunk:341-390
$
[Warning]Warning

While it is possible to modify svn:mergeinfo just as you might any other versioned property, we strongly discourage doing so unless you really know what you're doing.

[Tip]Tip

The amount of svn:mergeinfo on a single path can get quite large, as can the output of a svn propget --recursive or svn proplist --recursive when dealing with large amounts of subtree mergeinfo, see Subtree Merges and Subtree Mergeinfo . The formatted output produced by the --verbose option with either of these subcommands is often very helpful in these cases.

The svn:mergeinfo property is automatically maintained by Subversion whenever you run svn merge. Its value indicates which changes made to a given path have been replicated into the directory in question. In our previous example, the path which is the source of the merged changes is /trunk and the directory which has received the changes is /branches/my-calc-branch. Earlier versions of Subversion maintained the svn:mergeinfo property silently. You could still detect the changes, after a merge completed, with the svn diff or svn status subcommands, but the merge itself gave no indication when it changed the svn:mergeinfo property. This is no longer true in Subversion 1.7, which has several new notifications to alert you when a merge updates the svn:mergeinfo property. These notifications all begin with --- Recording mergeinfo for and appear at the end of the merge. Unlike other merge notifications, these don't describe the application of a difference to a working copy (see the section called “Merge Syntax: Full Disclosure”), but instead describe "housekeeping" changes made to keep track of what was merged.

Subversion also provides a subcommand, svn mergeinfo, which is helpful in seeing not only which changesets a directory has absorbed, but also which changesets it's still eligible to receive. This gives a sort of preview of which changes a subsequent svn merge operation would replicate to your branch.

$ cd my-calc-branch

# Which changes have already been merged from trunk to branch?
$ svn mergeinfo ^/calc/trunk
r341
r342
r343
…
r388
r389
r390

# Which changes are still eligible to merge from trunk to branch?
$ svn mergeinfo ^/calc/trunk --show-revs eligible
r391
r392
r393
r394
r395
$

The svn mergeinfo command requires a source URL (where the changes come from), and takes an optional target URL (where the changes merge to). If no target URL is given, it assumes that the current working directory is the target. In the prior example, because we're querying our branch working copy, the command assumes we're interested in receiving changes to /branches/mybranch from the specified trunk URL.

With the release of Subversion 1.7, the svn mergeinfo subcommand can also account for subtree mergeinfo and non-inheritable mergeinfo. It accounts for subtree mergeinfo by use of the --recursive or --depth options, while non-inheritable mergeinfo is considered by default.

Let's say we have a branch with both subtree and non-inheritable mergeinfo:

$ svn propget svn:mergeinfo --recursive -v
# Non-inheritable mergeinfo
Properties on '.':
  svn:mergeinfo
    /trunk:651-652,758*
# Subtree mergeinfo
Properties on 'doc/INSTALL':
  svn:mergeinfo
    /trunk/doc/INSTALL:651-652,958,1060

From the above mergeinfo we see that r758 has only been merged into the root of the branch, but not any of the root's children. We also see that both r958 and r1060 have been merged only to the doc/INSTALL file. When we use svn mergeinfo with the --recursive option to see what has been merged from ^/trunk to this branch, we see two revisions are flagged with the * marker:

$ svn mergeinfo --show-revs=merged ^/trunk . --recursive
651
652
758*
958*
1060

The * indicates revisions that are only partially merged to the target in question (the meaning is the same if we are checking for eligible revisions). What this means in this example is that if we tried to merge r758 or r958 from ^/trunk then more changes would result. Likewise, because r1060 is not flagged with a *, we know that it only affects doc/INSTALL and that trying to merge it would have no result.[30]

Another way to get a more precise preview of a merge operation is to use the --dry-run option:

$ svn merge ^/calc/trunk --dry-run
--- Merging r391 through r395 into 'branch':
U    integer.c

$ svn status
#  nothing printed, working copy is still unchanged.

The --dry-run option doesn't actually apply any local changes to the working copy. It shows only status codes that would be printed in a real merge. It's useful for getting a high-level preview of the potential merge, for those times when running svn diff gives too much detail.

[Tip]Tip

After performing a merge operation, but before committing the results of the merge, you can use svn diff --depth=empty /path/to/merge/target to see only the changes to the immediate target of your merge. If your merge target was a directory, only property differences are displayed. This is a handy way to see the changes to the svn:mergeinfo property recorded by the merge operation, which will remind you about what you've just merged.

Of course, the best way to preview a merge operation is to just do it! Remember, running svn merge isn't an inherently risky thing (unless you've made local modifications to your working copy—but we already stressed that you shouldn't merge into such an environment). If you don't like the results of the merge, simply run svn revert . -R to revert the changes from your working copy and retry the command with different options. The merge isn't final until you actually svn commit the results.

Undoing Changes

An extremely common use for svn merge is to roll back a change that has already been committed. Suppose you're working away happily on a working copy of /calc/trunk, and you discover that the change made way back in revision 303, which changed integer.c, is completely wrong. It never should have been committed. You can use svn merge to undo the change in your working copy, and then commit the local modification to the repository. All you need to do is to specify a reverse difference. (You can do this by specifying --revision 303:302, or by an equivalent --change -303.)

$ svn merge -c -303 ^/calc/trunk
--- Reverse-merging r303 into 'integer.c':
U    integer.c
--- Recording mergeinfo for reverse merge of r303 into 'integer.c':
 U   A-branch

$ svn status
 M      .
M       integer.c

$ svn diff
…
# verify that the change is removed
…

$ svn commit -m "Undoing change committed in r303."
Sending        integer.c
Transmitting file data .
Committed revision 350.

As we mentioned earlier, one way to think about a repository revision is as a specific changeset. By using the -r option, you can ask svn merge to apply a changeset, or a whole range of changesets, to your working copy. In our case of undoing a change, we're asking svn merge to apply changeset r303 to our working copy backward.

Keep in mind that rolling back a change like this is just like any other svn merge operation, so you should use svn status and svn diff to confirm that your work is in the state you want it to be in, and then use svn commit to send the final version to the repository. After committing, this particular changeset is no longer reflected in the HEAD revision.

Again, you may be thinking: well, that really didn't undo the commit, did it? The change still exists in revision 303. If somebody checks out a version of the calc project between revisions 303 and 349, she'll still see the bad change, right?

Yes, that's true. When we talk about removing a change, we're really talking about removing it from the HEAD revision. The original change still exists in the repository's history. For most situations, this is good enough. Most people are only interested in tracking the HEAD of a project anyway. There are special cases, however, where you really might want to destroy all evidence of the commit. (Perhaps somebody accidentally committed a confidential document.) This isn't so easy, it turns out, because Subversion was deliberately designed to never lose information. Revisions are immutable trees that build upon one another. Removing a revision from history would cause a domino effect, creating chaos in all subsequent revisions and possibly invalidating all working copies.[31]

Resurrecting Deleted Items

The great thing about version control systems is that information is never lost. Even when you delete a file or directory, it may be gone from the HEAD revision, but the object still exists in earlier revisions. One of the most common questions new users ask is, How do I get my old file or directory back?

The first step is to define exactly which item you're trying to resurrect. Here's a useful metaphor: you can think of every object in the repository as existing in a sort of two-dimensional coordinate system. The first coordinate is a particular revision tree, and the second coordinate is a path within that tree. So every version of your file or directory is defined by a specific coordinate pair. (Remember the peg revision syntax—foo.c@224—mentioned back in the section called “Peg and Operative Revisions”.)

First, you might need to use svn log to discover the exact coordinate pair you wish to resurrect. A good strategy is to run svn log --verbose in a directory that used to contain your deleted item. The --verbose (-v) option shows a list of all changed items in each revision; all you need to do is find the revision in which you deleted the file or directory. You can do this visually, or by using another tool to examine the log output (via grep, or perhaps via an incremental search in an editor).

$ cd parent-dir
$ svn log -v
…
------------------------------------------------------------------------
r808 | joe | 2003-12-26 14:29:40 -0600 (Fri, 26 Dec 2003) | 3 lines
Changed paths:
   D /calc/trunk/real.c
   M /calc/trunk/integer.c

Added fast fourier transform functions to integer.c.
Removed real.c because code now in double.c.
…

In the example, we're assuming that you're looking for a deleted file real.c. By looking through the logs of a parent directory, you've spotted that this file was deleted in revision 808. Therefore, the last version of the file to exist was in the revision right before that. Conclusion: you want to resurrect the path /calc/trunk/real.c from revision 807.

That was the hard part—the research. Now that you know what you want to restore, you have two different choices.

One option is to use svn merge to apply revision 808 in reverse. (We already discussed how to undo changes in the section called “Undoing Changes”.) This would have the effect of re-adding real.c as a local modification. The file would be scheduled for addition, and after a commit, the file would again exist in HEAD.

In this particular example, however, this is probably not the best strategy. Reverse-applying revision 808 would not only schedule real.c for addition, but the log message indicates that it would also undo certain changes to integer.c, which you don't want. Certainly, you could reverse-merge revision 808 and then svn revert the local modifications to integer.c, but this technique doesn't scale well. What if 90 files were changed in revision 808?

A second, more targeted strategy is not to use svn merge at all, but rather to use the svn copy command. Simply copy the exact revision and path coordinate pair from the repository to your working copy:

$ svn copy ^/calc/trunk/real.c@807 ./real.c

$ svn status
A  +    real.c

$ svn commit -m "Resurrected real.c from revision 807, /calc/trunk/real.c."
Adding         real.c
Transmitting file data .
Committed revision 1390.

The plus sign in the status output indicates that the item isn't merely scheduled for addition, but scheduled for addition with history. Subversion remembers where it was copied from. In the future, running svn log on this file will traverse back through the file's resurrection and through all the history it had prior to revision 807. In other words, this new real.c isn't really new; it's a direct descendant of the original, deleted file. This is usually considered a good and useful thing. If, however, you wanted to resurrect the file without maintaining a historical link to the old file, this technique works just as well:

$ svn cat ^/calc/trunk/real.c@807 > ./real.c

$ svn add real.c
A         real.c

$ svn commit -m "Re-created real.c from revision 807."
Adding         real.c
Transmitting file data .
Committed revision 1390.

Although our example shows us resurrecting a file, note that these same techniques work just as well for resurrecting deleted directories. Also note that a resurrection doesn't have to happen in your working copy—it can happen entirely in the repository:

$ svn copy ^/calc/trunk/real.c@807 ^/calc/trunk/ \
           -m "Resurrect real.c from revision 807."
Committed revision 1390.

$ svn update
Updating '.':
A    real.c
Updated to revision 1390.

Advanced Merging

Here ends the automated magic. Sooner or later, once you get the hang of branching and merging, you're going to have to ask Subversion to merge specific changes from one place to another. To do this, you're going to have to start passing more complicated arguments to svn merge. The next section describes the fully expanded syntax of the command and discusses a number of common scenarios that require it.

Cherrypicking

Just as the term changeset is often used in version control systems, so is the term cherrypicking. This word refers to the act of choosing one specific changeset from a branch and replicating it to another. Cherrypicking may also refer to the act of duplicating a particular set of (not necessarily contiguous!) changesets from one branch to another. This is in contrast to more typical merging scenarios, where the next contiguous range of revisions is duplicated automatically.

Why would people want to replicate just a single change? It comes up more often than you'd think. For example, let's go back in time and imagine that you haven't yet merged your private feature branch back to the trunk. At the water cooler, you get word that Sally made an interesting change to integer.c on the trunk. Looking over the history of commits to the trunk, you see that in revision 355 she fixed a critical bug that directly impacts the feature you're working on. You might not be ready to merge all the trunk changes to your branch just yet, but you certainly need that particular bug fix in order to continue your work.

$ svn diff -c 355 ^/calc/trunk

Index: integer.c
===================================================================
--- integer.c	(revision 354)
+++ integer.c	(revision 355)
@@ -147,7 +147,7 @@
     case 6:  sprintf(info->operating_system, "HPFS (OS/2 or NT)"); break;
     case 7:  sprintf(info->operating_system, "Macintosh"); break;
     case 8:  sprintf(info->operating_system, "Z-System"); break;
-    case 9:  sprintf(info->operating_system, "CP/MM");
+    case 9:  sprintf(info->operating_system, "CP/M"); break;
     case 10:  sprintf(info->operating_system, "TOPS-20"); break;
     case 11:  sprintf(info->operating_system, "NTFS (Windows NT)"); break;
     case 12:  sprintf(info->operating_system, "QDOS"); break;

Just as you used svn diff in the prior example to examine revision 355, you can pass the same option to svn merge:

$ svn merge -c 355 ^/calc/trunk
--- Merging r355 into '.':
U    integer.c
--- Recording mergeinfo for merge of r355 into '.':
 U   .

$ svn status
M       integer.c

You can now go through the usual testing procedures before committing this change to your branch. After the commit, Subversion marks r355 as having been merged to the branch so that future magic merges that synchronize your branch with the trunk know to skip over r355. (Merging the same change to the same branch almost always results in a conflict!)

$ cd my-calc-branch

$ svn propget svn:mergeinfo .
/trunk:341-349,355

# Notice that r355 isn't listed as "eligible" to merge, because
# it's already been merged.
$ svn mergeinfo ^/calc/trunk --show-revs eligible
r350
r351
r352
r353
r354
r356
r357
r358
r359
r360

$ svn merge ^/calc/trunk
--- Merging r350 through r354 into '.':
 U   .
U    integer.c
U    Makefile
--- Merging r356 through r360 into '.':
 U   .
U    integer.c
U    button.c
--- Recording mergeinfo for merge of r350 through r360 into '.':
 U   .

This use case of replicating (or backporting) bug fixes from one branch to another is perhaps the most popular reason for cherrypicking changes; it comes up all the time, for example, when a team is maintaining a release branch of software. (We discuss this pattern in the section called “Release Branches”.)

[Warning]Warning

Did you notice how, in the last example, the merge invocation merged two distinct ranges? The svn merge command applied two independent patches to your working copy to skip over changeset 355, which your branch already contained. There's nothing inherently wrong with this, except that it has the potential to make conflict resolution trickier. If the first range of changes creates conflicts, you must resolve them interactively for the merge process to continue and apply the second range of changes. If you postpone a conflict from the first wave of changes, the whole merge command will bail out with an error message.[32]

A word of warning: while svn diff and svn merge are very similar in concept, they do have different syntax in many cases. Be sure to read about them in Chapter 9, Subversion Complete Reference for details, or ask svn help. For example, svn merge requires a working copy path as a target, that is, a place where it should apply the generated patch. If the target isn't specified, it assumes you are trying to perform one of the following common operations:

  • You want to merge directory changes into your current working directory.

  • You want to merge the changes in a specific file into a file by the same name that exists in your current working directory.

If you are merging a directory and haven't specified a target path, svn merge assumes the first case and tries to apply the changes into your current directory. If you are merging a file, and that file (or a file by the same name) exists in your current working directory, svn merge assumes the second case and tries to apply the changes to a local file with the same name.

Merge Syntax: Full Disclosure

You've now seen some examples of the svn merge command, and you're about to see several more. If you're feeling confused about exactly how merging works, you're not alone. Many users (especially those new to version control) are initially perplexed about the proper syntax of the command and about how and when the feature should be used. But fear not, this command is actually much simpler than you think! There's a very easy technique for understanding exactly how svn merge behaves.

The main source of confusion is the name of the command. The term merge somehow denotes that branches are combined together, or that some sort of mysterious blending of data is going on. That's not the case. A better name for the command might have been svn diff-and-apply, because that's all that happens: two repository trees are compared, and the differences are applied to a working copy.

If you're using svn merge to do basic copying of changes between branches, it will generally do the right thing automatically. For example, a command such as the following:

$ svn merge ^/calc/branches/some-branch

will attempt to duplicate any changes made on some-branch into your current working directory, which is presumably a working copy that shares some historical connection to the branch. The command is smart enough to only duplicate changes that your working copy doesn't yet have. If you repeat this command once a week, it will only duplicate the newest branch changes that happened since you last merged.

If you choose to use the svn merge command in all its full glory by giving it specific revision ranges to duplicate, the command takes three main arguments:

  1. An initial repository tree (often called the left side of the comparison)

  2. A final repository tree (often called the right side of the comparison)

  3. A working copy to accept the differences as local changes (often called the target of the merge)

Once these three arguments are specified, then the two trees are compared and the differences applied to the target working copy as local modifications. When the command is done, the results are no different than if you had hand-edited the files or run various svn add or svn delete commands yourself. If you like the results, you can commit them. If you don't like the results, you can simply svn revert all of the changes.

The syntax of svn merge allows you to specify the three necessary arguments rather flexibly. Here are some examples:

$ svn merge http://svn.example.com/repos/branch1@150 \
            http://svn.example.com/repos/branch2@212 \
            my-working-copy

$ svn merge -r 100:200 http://svn.example.com/repos/trunk my-working-copy

$ svn merge -r 100:200 http://svn.example.com/repos/trunk

The first syntax lays out all three arguments explicitly, naming each tree in the form URL@REV and naming the working copy target. The second syntax is used as a shorthand for situations when you're comparing two different revisions of the same URL. The last syntax shows how the working copy argument is optional; if omitted, it defaults to the current directory.

While the first example shows the full syntax of svn merge, use it very carefully; it can result in merges which do not record any svn:mergeinfo metadata at all. The next section talks a bit more about this.

Merges Without Mergeinfo

Subversion tries to generate merge metadata whenever it can, to make future invocations of svn merge smarter. There are still situations, however, where svn:mergeinfo data is not created or changed. Remember to be a bit wary of these scenarios:

Merging unrelated sources

If you ask svn merge to compare two URLs that aren't related to each other, a patch is still generated and applied to your working copy, but no merging metadata is created. There's no common history between the two sources, and future smart merges depend on that common history.

Merging from foreign repositories

While it's possible to run a command such as svn merge -r 100:200 http://svn.foreignproject.com/repos/trunk, the resultant patch also lacks any historical merge metadata. At the time of this writing, Subversion has no way of representing different repository URLs within the svn:mergeinfo property.

Using --ignore-ancestry

If this option is passed to svn merge, it causes the merging logic to mindlessly generate differences the same way that svn diff does, ignoring any historical relationships. We discuss this later in this chapter in the section called “Noticing or Ignoring Ancestry”.

Applying reverse merges from a target's natural history

Earlier in this chapter (the section called “Undoing Changes”) we discussed how to use svn merge to apply a reverse patch as a way of rolling back changes. If this technique is used to undo a change to an object's personal history (e.g., commit r5 to the trunk, then immediately roll back r5 using svn merge . -c -5), this sort of merge doesn't affect the recorded mergeinfo.[33]

More on Merge Conflicts

Just like the svn update command, svn merge applies changes to your working copy. And therefore it's also capable of creating conflicts. The conflicts produced by svn merge, however, are sometimes different, and this section explains those differences.

To begin with, assume that your working copy has no local edits. When you svn update to a particular revision, the changes sent by the server always apply cleanly to your working copy. The server produces the delta by comparing two trees: a virtual snapshot of your working copy, and the revision tree you're interested in. Because the left hand side of the comparison is exactly equal to what you already have, the delta is guaranteed to correctly convert your working copy into the right hand tree.

But svn merge has no such guarantees and can be much more chaotic: the advanced user can ask the server to compare any two trees at all, even ones that are unrelated to the working copy! This means there's large potential for human error. Users will sometimes compare the wrong two trees, creating a delta that doesn't apply cleanly. The svn merge subcommand does its best to apply as much of the delta as possible, but some parts may be impossible. A common sign that you merged the wrong delta is unexpected tree conflicts:

$ svn merge -r 1288:1351 http://svn.example.com/myrepos/branch
--- Merging r1289 through r1351 into '.':
   C bar.c
   C foo.c
   C docs
--- Recording mergeinfo for merge of r1289 through r1351 into '.':
 U   .
Summary of conflicts:
  Tree conflicts: 3

$ svn st
!     C bar.c
      >   local missing, incoming edit upon merge
!     C foo.c
      >   local missing, incoming edit upon merge
!     C docs
      >   local delete, incoming edit upon merge

In the previous example, it might be the case that bar.c, foo.c, and docs all exist in both snapshots of the branch being compared. The resultant delta wants to change the contents of the corresponding paths in your working copy, but those paths don't exist in the working copy. Whatever the case, the preponderance of tree conflicts most likely means that the user compared the wrong two trees; it's a classic sign of user error. When this happens, it's easy to recursively revert all the changes created by the merge (svn revert . --recursive), delete any unversioned files or directories left behind after the revert, and rerun svn merge with the correct arguments.

Also keep in mind that a merge into a working copy with no local edits can still produce text conflicts.

$ svn merge -c 1701 http://svn.example.com/myrepos/branchX --accept postpone
--- Merging r1701 into '.':
C     glub.c
C     sputter.c
--- Recording mergeinfo for merge of r1701 into '.':
 U   .
Summary of conflicts:
  Text conflicts: 2

C:\SVN\src-branch-1.7.x>svn st
 M      .
?       glub.c.merge-left.r1700
?       glub.c.merge-right.r1701
C       glub.c
?       glub.c.working
?       sputter.c.merge-left.r1700
?       sputter.c.merge-right.r1701
C       sputter.c
?       sputter.c.working
Summary of conflicts:
  Text conflicts: 2

How can a conflict possibly happen? Again, because the user can request svn merge to define and apply any old delta to the working copy, that delta may contain textual changes that don't cleanly apply to a working file, even if the file has no local modifications.

Another small difference between svn update and svn merge is the names of the full-text files created when a conflict happens. In the section called “Resolve Any Conflicts”, we saw that an update produces files named filename.mine, filename.rOLDREV, and filename.rNEWREV. When svn merge produces a conflict, though, it creates three files named filename.working, filename.merge-left.rOLDREV, and filename.merge-right.rNEWREV. In this case, the terms merge-left and merge-right are describing which side of the double-tree comparison the file came from, rOLDREV describes the revision of the left side, and rNEWREV the revision of the right side. In any case, these differing names help you distinguish between conflicts that happened as a result of an update and ones that happened as a result of a merge.

Blocking Changes

Sometimes there's a particular changeset that you don't want automatically merged. For example, perhaps your team's policy is to do new development work on /trunk, but is more conservative about backporting changes to a stable branch you use for releasing to the public. On one extreme, you can manually cherrypick single changesets from the trunk to the branch—just the changes that are stable enough to pass muster. Maybe things aren't quite that strict, though; perhaps most of the time you just let svn merge automatically merge most changes from trunk to branch. In this case, you want a way to mask a few specific changes out, that is, prevent them from ever being automatically merged.

Through Subversion 1.7, the only way to block a changeset is to make the system believe that the change has already been merged. To do this, invoke the merge subcommand with the --record-only option:

$ cd my-calc-branch

$ svn propget svn:mergeinfo .
/trunk:1680-3305

# Let's make the metadata list r3328 as already merged.
$ svn merge -c 3328 --record-only ^/calc/trunk
--- Recording mergeinfo for merge of r3328 into '.':
 U   .

$ svn status
M       .

$ svn propget svn:mergeinfo .
/trunk:1680-3305,3328

$ svn commit -m "Block r3328 from being merged to the branch."
…

Beginning with Subversion 1.7, --record-only merges are transitive. This means that, in addition to recording mergeinfo describing the blocked revision(s), any svn:mergeinfo property differences in the merge source are also applied. For example, let's say we want to block the 'frazzle' feature from ever being merged from ^/trunk to our ^/branches/proj-X branch. We know that all the frazzle work was done on its own branch, which was reintegrated to trunk in revision 1055:

$ svn log -v ^/trunk -r 1055
------------------------------------------------------------------------
r1055 | francesca | 2011-09-22 07:40:06 -0400 (Thu, 22 Sep 2011) | 3 lines
Changed paths:
   M /trunk
   M /trunk/src/frazzle.c

Reintegrate the frazzle-feature-branch to trunk.

Because revision 1055 was a reintegrate merge we know that mergeinfo was recorded describing the merge:

$ svn diff ^/trunk -c 1055 --depth empty
Index: .
===================================================================
--- .   (revision 1054)
+++ .   (revision 1055)

Property changes on: .
___________________________________________________________________
Modified: svn:mergeinfo
   Merged /branches/frazzle-feature-branch:r997-1003

Now simply blocking merges of revision 1055 from ^/trunk isn't foolproof since someone could merge r996:1003 directly from ^/branches/frazzle-feature-branch. Fortunately the transitive nature of --record-only merges in Subversion 1.7 prevents this; the --record-only merge applies the svn:mergeinfo diff from revision 1055, thus blocking merges directly from the frazzle branch and as it has always done prior to Subversion 1.7, it blocks merges of revision 1055 directly from ^/trunk:

$ cd branches/proj-X

$ svn merge ^/trunk . -c 1055 --record-only
--- Merging r1055 into '.':
 G   .
--- Recording mergeinfo for merge of r1055 into '.':
 G   .

$ svn diff --depth empty .
Index: .
===================================================================
--- .   (revision 1070)
+++ .   (working copy)

Property changes on: .
___________________________________________________________________
Modified: svn:mergeinfo
   Merged /trunk:r1055
   Merged /branches/frazzle-feature-branch:r997-1003

Blocking changes with --record-only works, but it's also a little bit dangerous. The main problem is that we're not clearly differentiating between the ideas of I already have this change and I don't have this change, but don't currently want it. We're effectively lying to the system, making it think that the change was previously merged. This puts the responsibility on you—the user—to remember that the change wasn't actually merged, it just wasn't wanted. There's no way to ask Subversion for a list of blocked changelists. If you want to track them (so that you can unblock them someday) you'll need to record them in a text file somewhere, or perhaps in an invented property.

Keeping a Reintegrated Branch Alive

There is an alternative to destroying and re-creating a branch after reintegration. To understand why it works you need to understand why the branch is initially unfit for further use after it has been reintegrated.

Let's assume you created your branch in revision A. While working on your branch, you created one or more revisions which made changes to the branch. Before reintegrating your branch back to trunk, you made a final merge from trunk to your branch, and committed the result of this merge as revision B.

When reintegrating your branch into the trunk, you create a new revision X which changes the trunk. The changes made to trunk in this revision X are semantically equivalent to the changes you made to your branch between revisions A and B.

If you now try to merge outstanding changes from trunk to your branch, Subversion will consider changes made in revision X as eligible for merging into the branch. However, since your branch already contains all the changes made in revision X, merging these changes can result in spurious conflicts! These conflicts are often tree conflicts, especially if renames were made on the branch or the trunk while the branch was in development.

So what can be done about this? We need to make sure that Subversion does not try to merge revision X into the branch. This is done using the --record-only merge option, which was introduced in the section called “Blocking Changes”.

To carry out the record-only merge, get a working copy of the branch which was just reintegrated in revision X, and merge just revision X from trunk into your branch, making sure to use the --record-only option.

This merge uses the cherry-picking merge syntax, which was introduced in the section called “Cherrypicking”. Continuing with the running example from the section called “Reintegrating a Branch”, where revision X was revision 391:

$ cd my-calc-branch
$ svn update
Updating '.':
Updated to revision 393.
$ svn merge --record-only -c 391 ^/calc/trunk
--- Recording mergeinfo for merge of r391 into '.':
 U   .
$ svn commit -m "Block revision 391 from being merged into my-calc-branch."
Sending        .

Committed revision 394.

Now your branch is ready to soak up changes from the trunk again. After another sync of your branch to the trunk, you can even reintegrate the branch a second time. If necessary, you can do another record-only merge to keep the branch alive. Rinse and repeat.

It should now also be apparent why deleting the branch and re-creating it has the same effect as doing the above record-only merge. Because revision X is part of the natural history (see the sidebar Natural History and Implicit Mergeinfo) of the newly created branch, Subversion will never attempt to merge revision X into the branch, avoiding spurious conflicts.

Merge-Sensitive Logs and Annotations

One of the main features of any version control system is to keep track of who changed what, and when they did it. The svn log and svn blame subcommands are just the tools for this: when invoked on individual files, they show not only the history of changesets that affected the file, but also exactly which user wrote which line of code, and when she did it.

When changes start getting replicated between branches, however, things start to get complicated. For example, if you were to ask svn log about the history of your feature branch, it would show exactly every revision that ever affected the branch:

$ cd my-calc-branch
$ svn log -q
------------------------------------------------------------------------
r390 | user | 2002-11-22 11:01:57 -0600 (Fri, 22 Nov 2002)
------------------------------------------------------------------------
r388 | user | 2002-11-21 05:20:00 -0600 (Thu, 21 Nov 2002)
------------------------------------------------------------------------
r381 | user | 2002-11-20 15:07:06 -0600 (Wed, 20 Nov 2002)
------------------------------------------------------------------------
r359 | user | 2002-11-19 19:19:20 -0600 (Tue, 19 Nov 2002)
------------------------------------------------------------------------
r357 | user | 2002-11-15 14:29:52 -0600 (Fri, 15 Nov 2002)
------------------------------------------------------------------------
r343 | user | 2002-11-07 13:50:10 -0600 (Thu, 07 Nov 2002)
------------------------------------------------------------------------
r341 | user | 2002-11-03 07:17:16 -0600 (Sun, 03 Nov 2002)
------------------------------------------------------------------------
r303 | sally | 2002-10-29 21:14:35 -0600 (Tue, 29 Oct 2002)
------------------------------------------------------------------------
r98 | sally | 2002-02-22 15:35:29 -0600 (Fri, 22 Feb 2002)
------------------------------------------------------------------------

But is this really an accurate picture of all the changes that happened on the branch? What's left out here is the fact that revisions 390, 381, and 357 were actually the results of merging changes from the trunk. If you look at one of these logs in detail, the multiple trunk changesets that comprised the branch change are nowhere to be seen:

$ svn log -v -r 390
------------------------------------------------------------------------
r390 | user | 2002-11-22 11:01:57 -0600 (Fri, 22 Nov 2002) | 1 line
Changed paths:
   M /branches/my-calc-branch/button.c
   M /branches/my-calc-branch/README

Final merge of trunk changes to my-calc-branch.

We happen to know that this merge to the branch was nothing but a merge of trunk changes. How can we see those trunk changes as well? The answer is to use the --use-merge-history (-g) option. This option expands those child changes that were part of the merge.

$ svn log -v -r 390 -g
------------------------------------------------------------------------
r390 | user | 2002-11-22 11:01:57 -0600 (Fri, 22 Nov 2002) | 1 line
Changed paths:
   M /branches/my-calc-branch/button.c
   M /branches/my-calc-branch/README

Final merge of trunk changes to my-calc-branch.
------------------------------------------------------------------------
r383 | sally | 2002-11-21 03:19:00 -0600 (Thu, 21 Nov 2002) | 2 lines
Changed paths:
   M /branches/my-calc-branch/button.c
Merged via: r390

Fix inverse graphic error on button.
------------------------------------------------------------------------
r382 | sally | 2002-11-20 16:57:06 -0600 (Wed, 20 Nov 2002) | 2 lines
Changed paths:
   M /branches/my-calc-branch/README
Merged via: r390

Document my last fix in README.

By making the log operation use merge history, we see not just the revision we queried (r390), but also the two revisions that came along on the ride with it—a couple of changes made by Sally to the trunk. This is a much more complete picture of history!

The svn blame command also takes the --use-merge-history (-g) option. If this option is neglected, somebody looking at a line-by-line annotation of button.c may get the mistaken impression that you were responsible for the lines that fixed a certain error:

$ svn blame button.c
…
   390    user    retval = inverse_func(button, path);
   390    user    return retval;
   390    user    }
…

And while it's true that you did actually commit those three lines in revision 390, two of them were actually written by Sally back in revision 383:

$ svn blame button.c -g
…
G    383    sally   retval = inverse_func(button, path);
G    383    sally   return retval;
     390    user    }
…

Now we know who to really blame for those two lines of code!

Noticing or Ignoring Ancestry

When conversing with a Subversion developer, you might very likely hear reference to the term ancestry. This word is used to describe the relationship between two objects in a repository: if they're related to each other, one object is said to be an ancestor of the other.

For example, suppose you commit revision 100, which includes a change to a file foo.c. Then foo.c@99 is an ancestor of foo.c@100. On the other hand, suppose you commit the deletion of foo.c in revision 101, and then add a new file by the same name in revision 102. In this case, foo.c@99 and foo.c@102 may appear to be related (they have the same path), but in fact are completely different objects in the repository. They share no history or ancestry.

The reason for bringing this up is to point out an important difference between svn diff and svn merge. The former command ignores ancestry, while the latter command is quite sensitive to it. For example, if you asked svn diff to compare revisions 99 and 102 of foo.c, you would see line-based diffs; the diff command is blindly comparing two paths. But if you asked svn merge to compare the same two objects, it would notice that they're unrelated and first attempt to delete the old file, then add the new file; the output would indicate a deletion followed by an add:

D    foo.c
A    foo.c

Most merges involve comparing trees that are ancestrally related to one another; therefore, svn merge defaults to this behavior. Occasionally, however, you may want the merge command to compare two unrelated trees. For example, you may have imported two source-code trees representing different vendor releases of a software project (see the section called “Vendor Branches”). If you ask svn merge to compare the two trees, you'd see the entire first tree being deleted, followed by an add of the entire second tree! In these situations, you'll want svn merge to do a path-based comparison only, ignoring any relations between files and directories. Add the --ignore-ancestry option to your merge command, and it will behave just like svn diff. (And conversely, the --notice-ancestry option will cause svn diff to behave like the svn merge command.)

[Tip]Tip

The --ignore-ancestry option also disables Merge Tracking. This means that svn:mergeinfo is not considered when svn merge is determining what revisions to merge, nor is svn:mergeinfo recorded to describe the merge.

Merges and Moves

A common desire is to refactor source code, especially in Java-based software projects. Files and directories are shuffled around and renamed, often causing great disruption to everyone working on the project. Sounds like a perfect case to use a branch, doesn't it? Just create a branch, shuffle things around, and then merge the branch back to the trunk, right?

Alas, this scenario doesn't work so well right now and is considered one of Subversion's current weak spots. The problem is that Subversion's svn update command isn't as robust as it should be, particularly when dealing with copy and move operations.

When you use svn copy to duplicate a file, the repository remembers where the new file came from, but it fails to transmit that information to the client which is running svn update or svn merge. Instead of telling the client, Copy that file you already have to this new location, it sends down an entirely new file. This can lead to problems, especially because the same thing happens with renamed files. A lesser-known fact about Subversion is that it lacks true renames—the svn move command is nothing more than an aggregation of svn copy and svn delete.

For example, suppose that while working on your private branch, you rename integer.c to whole.c. Effectively you've created a new file in your branch that is a copy of the original file, and deleted the original file. Meanwhile, back on trunk, Sally has committed some improvements to integer.c. Now you decide to merge your branch to the trunk:

$ cd calc/trunk

$ svn merge --reintegrate ^/calc/branches/my-calc-branch
--- Merging differences between repository URLs into '.':
D    integer.c
A    whole.c
U    .
--- Recording mergeinfo for merge between repository URLs into '.':
 U   .

This doesn't look so bad at first glance, but it's also probably not what you or Sally expected. The merge operation has deleted the latest version of the integer.c file (the one containing Sally's latest changes), and blindly added your new whole.c file—which is a duplicate of the older version of integer.c. The net effect is that merging your rename to the trunk has removed Sally's recent changes from the latest revision!

This isn't true data loss. Sally's changes are still in the repository's history, but it may not be immediately obvious that this has happened. The moral of this story is that until Subversion improves, be very careful about merging copies and renames from one branch to another.

Blocking Merge-Unaware Clients

If you've just upgraded your server to Subversion 1.5 or later, there's a risk that pre-1.5 Subversion clients can cause problems with Merge Tracking. This is because pre-1.5 clients don't support this feature; when one of these older clients performs svn merge, it doesn't modify the value of the svn:mergeinfo property at all. So the subsequent commit, despite being the result of a merge, doesn't tell the repository about the duplicated changes—that information is lost. Later on, when merge-aware clients attempt automatic merging, they're likely to run into all sorts of conflicts resulting from repeated merges.

If you and your team are relying on the merge-tracking features of Subversion, you may want to configure your repository to prevent older clients from committing changes. The easy way to do this is by inspecting the capabilities parameter in the start-commit hook script. If the client reports itself as having mergeinfo capabilities, the hook script can allow the commit to start. If the client doesn't report that capability, have the hook deny the commit. Example 4.1, “Merge-tracking gatekeeper start-commit hook script” gives an example of such a hook script:

Example 4.1. Merge-tracking gatekeeper start-commit hook script

#!/usr/bin/env python
import sys

# The start-commit hook is invoked before a Subversion txn is created
# in the process of doing a commit.  Subversion runs this hook
# by invoking a program (script, executable, binary, etc.) named
# 'start-commit' (for which this file is a template)
# with the following ordered arguments:
#
#   [1] REPOS-PATH   (the path to this repository)
#   [2] USER         (the authenticated user attempting to commit)
#   [3] CAPABILITIES (a colon-separated list of capabilities reported
#                     by the client; see note below)

capabilities = sys.argv[3].split(':')
if "mergeinfo" not in capabilities:
  sys.stderr.write("Commits from merge-tracking-unaware clients are "
                   "not permitted.  Please upgrade to Subversion 1.5 "
                   "or newer.\n")
  sys.exit(1)
sys.exit(0)


For more information about hook scripts, see the section called “Implementing Repository Hooks”.

The Final Word on Merge Tracking

The bottom line is that Subversion's merge-tracking feature has an extremely complex internal implementation, and the svn:mergeinfo property is the only window the user has into the machinery.

Sometimes mergeinfo will appear on paths that you didn't expect to be touched by an operation. Sometimes mergeinfo won't be generated at all, when you expect it to. Furthermore, the management of mergeinfo metadata has a whole set of taxonomies and behaviors around it, such as explicit versus implicit mergeinfo, operative versus inoperative revisions, specific mechanisms of mergeinfo elision, and even inheritance from parent to child directories.

We've chosen to only briefly cover, if at all, these detailed topics for a couple of reasons. First, the level of detail is absolutely overwhelming for a typical user. Second, and more importantly, the typical user shouldn't have to understand these concepts; they should typically remain in the background as pesky implementation details. All that said, if you enjoy this sort of thing, you can get a fantastic overview in a paper posted at CollabNet's website: http://www.collab.net/community/subversion/articles/merge-info.html.

For now, if you want to steer clear of the complexities of merge tracking, we recommend that you follow these simple best practices:

  • For short-term feature branches, follow the simple procedure described throughout the section called “Basic Merging”.

  • Avoid subtree merges and subtree mergeinfo, perform merges only on the root of your branches, not on subdirectories or files (see Subtree Merges and Subtree Mergeinfo) .

  • Don't ever edit the svn:mergeinfo property directly; use svn merge with the --record-only option to effect a desired change to the metadata (as demonstrated in the section called “Blocking Changes”).

  • Your merge target should be a working copy which represents the root of a complete tree representing a single location in the repository at a single point in time:

    • Don't use the --allow-mixed-revisions option to merge into mixed-revision working copies.

    • Don't merge to targets with switched subdirectories (as described next in the section called “Traversing Branches”).

    • Avoid merges to targets with sparse directories. Likewise, don't merge to depths other than --depth=infinity

    • Be sure you have read access to all of the merge source and read/write access to all of the merge target.

Traversing Branches

The svn switch command transforms an existing working copy to reflect a different branch. While this command isn't strictly necessary for working with branches, it provides a nice shortcut. In one of our earlier examples, after creating your private branch, you checked out a fresh working copy of the new repository directory. Instead, you can simply ask Subversion to change your working copy of /calc/trunk to mirror the new branch location:

$ cd calc

$ svn info | grep URL
URL: http://svn.example.com/repos/calc/trunk

$ svn switch ^/calc/branches/my-calc-branch
U    integer.c
U    button.c
U    Makefile
Updated to revision 341.

$ svn info | grep URL
URL: http://svn.example.com/repos/calc/branches/my-calc-branch

Switching a working copy that has no local modifications to a different branch results in the working copy looking just as it would if you'd done a fresh checkout of the directory. It's usually more efficient to use this command, because often branches differ by only a small degree. The server sends only the minimal set of changes necessary to make your working copy reflect the branch directory.

The svn switch command also takes a --revision (-r) option, so you need not always move your working copy to the HEAD of the branch.

Of course, most projects are more complicated than our calc example, and contain multiple subdirectories. Subversion users often follow a specific algorithm when using branches:

  1. Copy the project's entire trunk to a new branch directory.

  2. Switch only part of the trunk working copy to mirror the branch.

In other words, if a user knows that the branch work needs to happen on only a specific subdirectory, she uses svn switch to move only that subdirectory to the branch. (Or sometimes users will switch just a single working file to the branch!) That way, the user can continue to receive normal trunk updates to most of her working copy, but the switched portions will remain immune (unless someone commits a change to her branch). This feature adds a whole new dimension to the concept of a mixed working copy—not only can working copies contain a mixture of working revisions, but they can also contain a mixture of repository locations as well.

[Tip]Tip

Typically switched subdirectories share common ancestry with the location which is switched away from. However svn switch can switch a subdirectory to mirror a repository location which it shares no common ancestry with. To do this you need to use the --ignore-ancestry option.

If your working copy contains a number of switched subtrees from different repository locations, it continues to function as normal. When you update, you'll receive patches to each subtree as appropriate. When you commit, your local changes are still applied as a single, atomic change to the repository.

Note that while it's okay for your working copy to reflect a mixture of repository locations, these locations must all be within the same repository. Subversion repositories aren't yet able to communicate with one another; that feature is planned for the future.[34]

[Tip]Tip

Administrators who need to change the URL of a repository which is accessed via HTTP are encouraged to add to their httpd.conf configuration file a permanent redirect from the old URL location to the new one (via the RedirectPermanent directive). Subversion clients will generally display the new repository URL in error messages generated when the user attempts to use working copies which still reflect the old URL location. In fact, Subversion 1.7 clients will go a step further, automatically relocating the working copy to the new URL.

Because svn switch is essentially a variant of svn update, it shares the same behaviors; any local modifications in your working copy are preserved when new data arrives from the repository.

[Tip]Tip

Have you ever found yourself making some complex edits (in your /trunk working copy) and suddenly realized, Hey, these changes ought to be in their own branch? There is a great two step technique to do this:

$ svn copy http://svn.example.com/repos/calc/trunk \
           http://svn.example.com/repos/calc/branches/newbranch \
           -m "Create branch 'newbranch'."
Committed revision 353.
$ svn switch ^/calc/branches/newbranch
At revision 353.

The svn switch command, like svn update, preserves your local edits. At this point, your working copy is now a reflection of the newly created branch, and your next svn commit invocation will send your changes there.

Tags

Another common version control concept is a tag. A tag is just a snapshot of a project in time. In Subversion, this idea already seems to be everywhere. Each repository revision is exactly that—a snapshot of the filesystem after each commit.

However, people often want to give more human-friendly names to tags, such as release-1.0. And they want to make snapshots of smaller subdirectories of the filesystem. After all, it's not so easy to remember that release 1.0 of a piece of software is a particular subdirectory of revision 4822.

Creating a Simple Tag

Once again, svn copy comes to the rescue. If you want to create a snapshot of /calc/trunk exactly as it looks in the HEAD revision, make a copy of it:

$ svn copy http://svn.example.com/repos/calc/trunk \
           http://svn.example.com/repos/calc/tags/release-1.0 \
           -m "Tagging the 1.0 release of the 'calc' project."

Committed revision 902.

This example assumes that a /calc/tags directory already exists. (If it doesn't, you can create it using svn mkdir.) After the copy completes, the new release-1.0 directory is forever a snapshot of how the /trunk directory looked in the HEAD revision at the time you made the copy. Of course, you might want to be more precise about exactly which revision you copy, in case somebody else may have committed changes to the project when you weren't looking. So if you know that revision 901 of /calc/trunk is exactly the snapshot you want, you can specify it by passing -r 901 to the svn copy command.

But wait a moment: isn't this tag creation procedure the same procedure we used to create a branch? Yes, in fact, it is. In Subversion, there's no difference between a tag and a branch. Both are just ordinary directories that are created by copying. Just as with branches, the only reason a copied directory is a tag is because humans have decided to treat it that way: as long as nobody ever commits to the directory, it forever remains a snapshot. If people start committing to it, it becomes a branch.

If you are administering a repository, there are two approaches you can take to managing tags. The first approach is hands off: as a matter of project policy, decide where your tags will live, and make sure all users know how to treat the directories they copy. (That is, make sure they know not to commit to them.) The second approach is more paranoid: you can use one of the access control scripts provided with Subversion to prevent anyone from doing anything but creating new copies in the tags area (see Chapter 6, Server Configuration). The paranoid approach, however, isn't usually necessary. If a user accidentally commits a change to a tag directory, you can simply undo the change as discussed in the previous section. This is version control, after all!

Creating a Complex Tag

Sometimes you may want a snapshot that is more complicated than a single directory at a single revision.

For example, pretend your project is much larger than our calc example: suppose it contains a number of subdirectories and many more files. In the course of your work, you may decide that you need to create a working copy that is designed to have specific features and bug fixes. You can accomplish this by selectively backdating files or directories to particular revisions (using svn update with the -r option liberally), by switching files and directories to particular branches (making use of svn switch), or even just by making a bunch of local changes. When you're done, your working copy is a hodgepodge of repository locations from different revisions. But after testing, you know it's the precise combination of data you need to tag.

Time to make a snapshot. Copying one URL to another won't work here. In this case, you want to make a snapshot of your exact working copy arrangement and store it in the repository. Luckily, svn copy actually has four different uses (which you can read about in Chapter 9, Subversion Complete Reference), including the ability to copy a working copy tree to the repository:

$ ls
my-working-copy/

$ svn copy my-working-copy \
           http://svn.example.com/repos/calc/tags/mytag \
           -m "Tag my existing working copy state."

Committed revision 940.

Now there is a new directory in the repository, /calc/tags/mytag, which is an exact snapshot of your working copy—mixed revisions, URLs, local changes, and all.

Other users have found interesting uses for this feature. Sometimes there are situations where you have a bunch of local changes made to your working copy, and you'd like a collaborator to see them. Instead of running svn diff and sending a patch file (which won't capture directory or symlink changes), you can use svn copy to upload your working copy to a private area of the repository. Your collaborator can then either check out a verbatim copy of your working copy or use svn merge to receive your exact changes.

While this is a nice method for uploading a quick snapshot of your working copy, note that this is not a good way to initially create a branch. Branch creation should be an event unto itself, and this method conflates the creation of a branch with extra changes to files, all within a single revision. This makes it very difficult (later on) to identify a single revision number as a branch point.

Branch Maintenance

You may have noticed by now that Subversion is extremely flexible. Because it implements branches and tags with the same underlying mechanism (directory copies), and because branches and tags appear in normal filesystem space, many people find Subversion intimidating. It's almost too flexible. In this section, we'll offer some suggestions for arranging and managing your data over time.

Repository Layout

There are some standard, recommended ways to organize the contents of a repository. Most people create a trunk directory to hold the main line of development, a branches directory to contain branch copies, and a tags directory to contain tag copies. If a repository holds only one project, often people create these top-level directories:



/

   trunk/

   branches/

   tags/

If a repository contains multiple projects, admins typically index their layout by project. See the section called “Planning Your Repository Organization” to read more about project roots, but here's an example of such a layout:



/

   paint/

      trunk/

      branches/

      tags/

   calc/

      trunk/

      branches/

      tags/

Of course, you're free to ignore these common layouts. You can create any sort of variation, whatever works best for you or your team. Remember that whatever you choose, it's not a permanent commitment. You can reorganize your repository at any time. Because branches and tags are ordinary directories, the svn move command can move or rename them however you wish. Switching from one layout to another is just a matter of issuing a series of server-side moves; if you don't like the way things are organized in the repository, just juggle the directories around.

Remember, though, that while moving directories is easy to do, you need to be considerate of other users as well. Your juggling can disorient users with existing working copies. If a user has a working copy of a particular repository directory and your svn move subcommand removes the path from the latest revision, then when the user next runs svn update, she is told that her working copy represents a path that no longer exists. She is then forced to svn switch to the new location.

Data Lifetimes

Another nice feature of Subversion's model is that branches and tags can have finite lifetimes, just like any other versioned item. For example, suppose you eventually finish all your work on your personal branch of the calc project. After merging all of your changes back into /calc/trunk, there's no need for your private branch directory to stick around anymore:

$ svn delete http://svn.example.com/repos/calc/branches/my-calc-branch \
             -m "Removing obsolete branch of calc project."

Committed revision 375.

And now your branch is gone. Of course, it's not really gone: the directory is simply missing from the HEAD revision, no longer distracting anyone. If you use svn checkout, svn switch, or svn list to examine an earlier revision, you can still see your old branch.

If browsing your deleted directory isn't enough, you can always bring it back. Resurrecting data is very easy in Subversion. If there's a deleted directory (or file) that you'd like to bring back into HEAD, simply use svn copy to copy it from the old revision:

$ svn copy http://svn.example.com/repos/calc/branches/my-calc-branch@374 \
           http://svn.example.com/repos/calc/branches/my-calc-branch \
           -m "Restore my-calc-branch."

Committed revision 376.

In our example, your personal branch had a relatively short lifetime: you may have created it to fix a bug or implement a new feature. When your task is done, so is the branch. In software development, though, it's also common to have two main branches running side by side for very long periods. For example, suppose it's time to release a stable version of the calc project to the public, and you know it's going to take a couple of months to shake bugs out of the software. You don't want people to add new features to the project, but you don't want to tell all developers to stop programming either. So instead, you create a stable branch of the software that won't change much:

$ svn copy http://svn.example.com/repos/calc/trunk \
           http://svn.example.com/repos/calc/branches/stable-1.0 \
           -m "Creating stable branch of calc project."

Committed revision 377.

And now developers are free to continue adding cutting-edge (or experimental) features to /calc/trunk, and you can declare a project policy that only bug fixes are to be committed to /calc/branches/stable-1.0. That is, as people continue to work on the trunk, a human selectively ports bug fixes over to the stable branch. Even after the stable branch has shipped, you'll probably continue to maintain the branch for a long time—that is, as long as you continue to support that release for customers. We'll discuss this more in the next section.

Common Branching Patterns

There are many different uses for branching and svn merge, and this section describes the most common.

Version control is most often used for software development, so here's a quick peek at two of the most common branching/merging patterns used by teams of programmers. If you're not using Subversion for software development, feel free to skip this section. If you're a software developer using version control for the first time, pay close attention, as these patterns are often considered best practices by experienced folk. These processes aren't specific to Subversion; they're applicable to any version control system. Still, it may help to see them described in Subversion terms.

Release Branches

Most software has a typical life cycle: code, test, release, repeat. There are two problems with this process. First, developers need to keep writing new features while quality assurance teams take time to test supposedly stable versions of the software. New work cannot halt while the software is tested. Second, the team almost always needs to support older, released versions of software; if a bug is discovered in the latest code, it most likely exists in released versions as well, and customers will want to get that bug fix without having to wait for a major new release.

Here's where version control can help. The typical procedure looks like this:

  1. Developers commit all new work to the trunk. Day-to-day changes are committed to /trunk: new features, bug fixes, and so on.

  2. The trunk is copied to a release branch. When the team thinks the software is ready for release (say, a 1.0 release), /trunk might be copied to /branches/1.0.

  3. Teams continue to work in parallel. One team begins rigorous testing of the release branch, while another team continues new work (say, for version 2.0) on /trunk. If bugs are discovered in either location, fixes are ported back and forth as necessary. At some point, however, even that process stops. The branch is frozen for final testing right before a release.

  4. The branch is tagged and released. When testing is complete, /branches/1.0 is copied to /tags/1.0.0 as a reference snapshot. The tag is packaged and released to customers.

  5. The branch is maintained over time. While work continues on /trunk for version 2.0, bug fixes continue to be ported from /trunk to /branches/1.0. When enough bug fixes have accumulated, management may decide to do a 1.0.1 release: /branches/1.0 is copied to /tags/1.0.1, and the tag is packaged and released.

This entire process repeats as the software matures: when the 2.0 work is complete, a new 2.0 release branch is created, tested, tagged, and eventually released. After some years, the repository ends up with a number of release branches in maintenance mode, and a number of tags representing final shipped versions.

Feature Branches

A feature branch is the sort of branch that's been the dominant example in this chapter (the one you've been working on while Sally continues to work on /trunk). It's a temporary branch created to work on a complex change without interfering with the stability of /trunk. Unlike release branches (which may need to be supported forever), feature branches are born, used for a while, merged back to the trunk, and then ultimately deleted. They have a finite span of usefulness.

Again, project policies vary widely concerning exactly when it's appropriate to create a feature branch. Some projects never use feature branches at all: commits to /trunk are a free-for-all. The advantage to this system is that it's simple—nobody needs to learn about branching or merging. The disadvantage is that the trunk code is often unstable or unusable. Other projects use branches to an extreme: no change is ever committed to the trunk directly. Even the most trivial changes are created on a short-lived branch, carefully reviewed, and merged to the trunk. Then the branch is deleted. This system guarantees an exceptionally stable and usable trunk at all times, but at the cost of tremendous process overhead.

Most projects take a middle-of-the-road approach. They commonly insist that /trunk compile and pass regression tests at all times. A feature branch is required only when a change requires a large number of destabilizing commits. A good rule of thumb is to ask this question: if the developer worked for days in isolation and then committed the large change all at once (so that /trunk were never destabilized), would it be too large a change to review? If the answer to that question is yes, the change should be developed on a feature branch. As the developer commits incremental changes to the branch, they can be easily reviewed by peers.

Finally, there's the issue of how to best keep a feature branch in sync with the trunk as work progresses. As we mentioned earlier, there's a great risk to working on a branch for weeks or months; trunk changes may continue to pour in, to the point where the two lines of development differ so greatly that it may become a nightmare trying to merge the branch back to the trunk.

This situation is best avoided by regularly merging trunk changes to the branch. Make up a policy: once a week, merge the last week's worth of trunk changes to the branch.

When you are eventually ready to merge the synchronized feature branch back to the trunk, begin by doing a final merge of the latest trunk changes to the branch. When that's done, the latest versions of branch and trunk are absolutely identical except for your branch changes. You then merge back with the --reintegrate option:

$ cd trunk-working-copy

$ svn update
Updating '.':
At revision 1910.

$ svn merge --reintegrate ^/calc/branches/mybranch
--- Merging differences between repository URLs into '.':
U    real.c
U    integer.c
A    newdirectory
A    newdirectory/newfile
 U   .
…

Another way of thinking about this pattern is that your weekly sync of trunk to branch is analogous to running svn update in a working copy, while the final merge step is analogous to running svn commit from a working copy. After all, what else is a working copy but a very shallow private branch? It's a branch that's capable of storing only one change at a time.

Vendor Branches

As is especially the case when developing software, the data that you maintain under version control is often closely related to, or perhaps dependent upon, someone else's data. Generally, the needs of your project will dictate that you stay as up to date as possible with the data provided by that external entity without sacrificing the stability of your own project. This scenario plays itself out all the time—anywhere that the information generated by one group of people has a direct effect on that which is generated by another group.

For example, software developers might be working on an application that makes use of a third-party library. Subversion has just such a relationship with the Apache Portable Runtime (APR) library (see the section called “The Apache Portable Runtime Library”). The Subversion source code depends on the APR library for all its portability needs. In earlier stages of Subversion's development, the project closely tracked APR's changing API, always sticking to the bleeding edge of the library's code churn. Now that both APR and Subversion have matured, Subversion attempts to synchronize with APR's library API only at well-tested, stable release points.

Now, if your project depends on someone else's information, you could attempt to synchronize that information with your own in several ways. Most painfully, you could issue oral or written instructions to all the contributors of your project, telling them to make sure they have the specific versions of that third-party information that your project needs. If the third-party information is maintained in a Subversion repository, you could also use Subversion's externals definitions to effectively pin down specific versions of that information to some location in your own working copy (see the section called “Externals Definitions”).

But sometimes you want to maintain custom modifications to third-party code in your own version control system. Returning to the software development example, programmers might need to make modifications to that third-party library for their own purposes. These modifications might include new functionality or bug fixes, maintained internally only until they become part of an official release of the third-party library. Or the changes might never be relayed back to the library maintainers, existing solely as custom tweaks to make the library further suit the needs of the software developers.

Now you face an interesting situation. Your project could house its custom modifications to the third-party data in some disjointed fashion, such as using patch files or full-fledged alternative versions of files and directories. But these quickly become maintenance headaches, requiring some mechanism by which to apply your custom changes to the third-party code and necessitating regeneration of those changes with each successive version of the third-party code that you track.

The solution to this problem is to use vendor branches. A vendor branch is a directory tree in your own version control system that contains information provided by a third-party entity, or vendor. Each version of the vendor's data that you decide to absorb into your project is called a vendor drop.

Vendor branches provide two benefits. First, by storing the currently supported vendor drop in your own version control system, you ensure that the members of your project never need to question whether they have the right version of the vendor's data. They simply receive that correct version as part of their regular working copy updates. Second, because the data lives in your own Subversion repository, you can store your custom changes to it in-place—you have no more need of an automated (or worse, manual) method for swapping in your customizations.

General Vendor Branch Management Procedure

Managing vendor branches generally works like this: first, you create a top-level directory (such as /vendor) to hold the vendor branches. Then you import the third-party code into a subdirectory of that top-level directory. You then copy that subdirectory into your main development branch (e.g., /trunk) at the appropriate location. You always make your local changes in the main development branch. With each new release of the code you are tracking, you bring it into the vendor branch and merge the changes into /trunk, resolving whatever conflicts occur between your local changes and the upstream changes.

An example will help to clarify this algorithm. We'll use a scenario where your development team is creating a calculator program that links against a third-party complex number arithmetic library, libcomplex. We'll begin with the initial creation of the vendor branch and the import of the first vendor drop. We'll call our vendor branch directory libcomplex, and our code drops will go into a subdirectory of our vendor branch called current. And since svn import creates all the intermediate parent directories it needs, we can actually accomplish both of these steps with a single command:

$ svn import /path/to/libcomplex-1.0 \
             http://svn.example.com/repos/vendor/libcomplex/current \
             -m "importing initial 1.0 vendor drop"
…

We now have the current version of the libcomplex source code in /vendor/libcomplex/current. Now, we tag that version (see the section called “Tags”) and then copy it into the main development branch. Our copy will create a new directory called libcomplex in our existing calc project directory. It is in this copied version of the vendor data that we will make our customizations:

$ svn copy http://svn.example.com/repos/vendor/libcomplex/current  \
           http://svn.example.com/repos/vendor/libcomplex/1.0      \
           -m "tagging libcomplex-1.0"
…
$ svn copy http://svn.example.com/repos/vendor/libcomplex/1.0  \
           http://svn.example.com/repos/calc/libcomplex        \
           -m "bringing libcomplex-1.0 into the main branch"
…

We check out our project's main branch—which now includes a copy of the first vendor drop—and we get to work customizing the libcomplex code. Before we know it, our modified version of libcomplex is now completely integrated into our calculator program.[35]

A few weeks later, the developers of libcomplex release a new version of their library—version 1.1—which contains some features and functionality that we really want. We'd like to upgrade to this new version, but without losing the customizations we made to the existing version. What we essentially would like to do is to replace our current baseline version of libcomplex 1.0 with a copy of libcomplex 1.1, and then re-apply the custom modifications we previously made to that library to the new version. But we actually approach the problem from the other direction, applying the changes made to libcomplex between versions 1.0 and 1.1 to our modified copy of it.

To perform this upgrade, we check out a copy of our vendor branch and replace the code in the current directory with the new libcomplex 1.1 source code. We quite literally copy new files on top of existing files, perhaps exploding the libcomplex 1.1 release tarball atop our existing files and directories. The goal here is to make our current directory contain only the libcomplex 1.1 code and to ensure that all that code is under version control. Oh, and we want to do this with as little version control history disturbance as possible.

After replacing the 1.0 code with 1.1 code, svn status will show files with local modifications as well as, perhaps, some unversioned files. If we did what we were supposed to do, the unversioned files are only those new files introduced in the 1.1 release of libcomplex—we run svn add on those to get them under version control. If the 1.1 code no longer has certain files that were in the 1.0 tree, it may be hard to notice them; you'd have to compare the two trees with some external tool and then svn delete any files present in 1.0 but not in 1.1. (Although it might also be just fine to let these same files live on in unused obscurity!) Finally, once our current working copy contains only the libcomplex 1.1 code, we commit the changes we made to get it looking that way.

Our current branch now contains the new vendor drop. We tag the new version as 1.1 (in the same way we previously tagged the version 1.0 vendor drop), and then merge the differences between the tag of the previous version and the new current version into our main development branch:

$ cd working-copies/calc
$ svn merge ^/vendor/libcomplex/1.0      \
            ^/vendor/libcomplex/current  \
            libcomplex
… # resolve all the conflicts between their changes and our changes
$ svn commit -m "merging libcomplex-1.1 into the main branch"
…

In the trivial use case, the new version of our third-party tool would look, from a files-and-directories point of view, just like the previous version. None of the libcomplex source files would have been deleted, renamed, or moved to different locations—the new version would contain only textual modifications against the previous one. In a perfect world, our modifications would apply cleanly to the new version of the library, with absolutely no complications or conflicts.

But things aren't always that simple, and in fact it is quite common for source files to get moved around between releases of software. This complicates the process of ensuring that our modifications are still valid for the new version of code, and things can quickly degrade into a situation where we have to manually re-create our customizations in the new version. Once Subversion knows about the history of a given source file—including all its previous locations—the process of merging in the new version of the library is pretty simple. But we are responsible for telling Subversion how the source file layout changed from vendor drop to vendor drop.

svn_load_dirs.pl

Vendor drops that contain more than a few deletes, additions, and moves complicate the process of upgrading to each successive version of the third-party data. So Subversion supplies the svn_load_dirs.pl script to assist with this process. This script automates the importing steps we mentioned in the general vendor branch management procedure to make sure mistakes are minimized. You will still be responsible for using the merge commands to merge the new versions of the third-party data into your main development branch, but svn_load_dirs.pl can help you more quickly and easily arrive at that stage.

In short, svn_load_dirs.pl is an enhancement to svn import that has several important characteristics:

  • It can be run at any point in time to bring an existing directory in the repository to exactly match an external directory, performing all the necessary adds and deletes, and optionally performing moves, too.

  • It takes care of complicated series of operations between which Subversion requires an intermediate commit—such as before renaming a file or directory twice.

  • It will optionally tag the newly imported directory.

  • It will optionally add arbitrary properties to files and directories that match a regular expression.

svn_load_dirs.pl takes three mandatory arguments. The first argument is the URL to the base Subversion directory to work in. This argument is followed by the URL—relative to the first argument—into which the current vendor drop will be imported. Finally, the third argument is the local directory to import. Using our previous example, a typical run of svn_load_dirs.pl might look like this:

$ svn_load_dirs.pl http://svn.example.com/repos/vendor/libcomplex \
                   current                                        \
                   /path/to/libcomplex-1.1
…

You can indicate that you'd like svn_load_dirs.pl to tag the new vendor drop by passing the -t command-line option and specifying a tag name. This tag is another URL relative to the first program argument.

$ svn_load_dirs.pl -t libcomplex-1.1                              \
                   http://svn.example.com/repos/vendor/libcomplex \
                   current                                        \
                   /path/to/libcomplex-1.1
…

When you run svn_load_dirs.pl, it examines the contents of your existing current vendor drop and compares them with the proposed new vendor drop. In the trivial case, no files will be in one version and not the other, and the script will perform the new import without incident. If, however, there are discrepancies in the file layouts between versions, svn_load_dirs.pl will ask you how to resolve those differences. For example, you will have the opportunity to tell the script that you know that the file math.c in version 1.0 of libcomplex was renamed to arithmetic.c in libcomplex 1.1. Any discrepancies not explained by moves are treated as regular additions and deletions.

The script also accepts a separate configuration file for setting properties on added files and directories which match a regular expression. This configuration file is specified to svn_load_dirs.pl using the -p command-line option. Each line of the configuration file is a whitespace-delimited set of two or four values: a Perl-style regular expression against which to match the added path, a control keyword (either break or cont), and then optionally a property name and value.

\.png$              break   svn:mime-type   image/png
\.jpe?g$            break   svn:mime-type   image/jpeg
\.m3u$              cont    svn:mime-type   audio/x-mpegurl
\.m3u$              break   svn:eol-style   LF
.*                  break   svn:eol-style   native

For each added path, the configured property changes whose regular expression matches the path are applied in order, unless the control specification is break (which means that no more property changes should be applied to that path). If the control specification is cont—an abbreviation for continue—matching will continue with the next line of the configuration file.

Any whitespace in the regular expression, property name, or property value must be surrounded by either single or double quotes. You can escape quotes that are not used for wrapping whitespace by preceding them with a backslash (\) character. The backslash escapes only quotes when parsing the configuration file, so do not protect any other characters beyond what is necessary for the regular expression.

To Branch or Not to Branch?

To branch or not to branch—that is an interesting question. This chapter has provided thus far a pretty deep dive into the waters of branching and merging, topics which have historically been the premier source of Subversion user confusion. As if the rote actions involved in branching and branch management aren't sometimes tricky enough, some users get hung up on deciding whether they need to branch at all. As you've learned, Subversion can handle common branching and branch management scenarios. So, the decision of whether or not to branch a project's history is rarely a technical one. Rather, the social impact of the decision often carries more weight. Let's examine some of the benefits and costs of using branches in a software project.

The most obvious benefit of working on a branch is isolation. Changes made to the branch don't affect the other lines of development in the project; changes made to those other lines don't affect the branch. In this way, a branch can serve as a great place to experiment with new features, complex bug fixes, major code rewrites, and so on. No matter how much stuff Sally breaks on her branch, Harry and the rest of the team can continue with their work unhindered outside the branch.

Branches also provide a great way to organize related changes into readily identifiable collections. For example, the changes which comprise the complete solution to a particular bug might be a list of non-sequential revision numbers. You might describe them in human language as revisions 1534, 1543, 1587 and 1588. You'd probably reproduce those numbers manually (or otherwise) in the issue tracker artifact which tracks the bug. When porting the bug fix to other product versions, you'd need to make sure to port all those revisions. But had those changes been made on a unique branch, you'd find yourself referring only to that branch by its name in conversation, in issue tracker comments, and when porting changes.

The unfortunate downside of branches, though, is that the very isolation that makes them so useful can be at odds with the collaborative needs of the project team. Depending on the work habits of your project peers, changes made to branches might not get the kind of constructive review, criticism, and testing that changes made to the main line of development do. The isolation of a branch can encourage users to forsake certain version control best practices, leading to version history which is difficult to review post facto. Developers on long-lived branches sometimes need to work extra hard to ensure that the evolutionary direction of their isolated copy of the codebase is in harmony with the direction their peers are steering the main code lines. Now, these drawbacks might be less of an issue for true exploratory branches aimed at experimenting with the future of a codebase with no expectation of reintegrating the results back into the main development lines—mere policy needn't be a vision-killer! But the simple fact remains that projects generally benefit from an orderly approach to version control where code and code changes enjoy the review and comprehension of more than one team member.

That's not to say that there are no technical penalties to branching. Pardon us while we go meta for a bit here. If you think about it, every time you checkout a Subversion working copy, you're creating a branch of sorts of your project. It's a special sort of branch. It lives only on your client machine; not in the repository. You synchronize this branch with changes made in the repository using svn update—which acts almost like a special-cased, simplified form of an svn merge command.[36] You effectively reintegrate your branch each time you run svn commit. So, in that special sense, Subversion users deal with branches and merges all the time. Given the similarities between updating and merging, it's no surprise, then, that the areas in which Subversion seems to have the most shortcomings—namely, handling file and directory renames and dealing with tree conflicts in general—are problematic for both the svn update and svn merge operations. Unfortunately, svn merge has a harder time of it precisely because of the fact that, for every way in which svn update is a special-cased, simplified kind of generic merge operation, a true Subversion merge is neither special-cased nor simplified. For this reason, merges perform much more slowly than updates, require explicit tracking (via the svn:mergeinfo property we've discussed in this chapter) and history-crunching arithmetic, and generally offer more opportunities for something to go awry.

To branch or not to branch? Ultimately, that depends on what your team needs in order to find that sweet balance of collaboration and isolation.

Summary

We covered a lot of ground in this chapter. We discussed the concepts of tags and branches and demonstrated how Subversion implements these concepts by copying directories with the svn copy command. We showed how to use svn merge to copy changes from one branch to another or roll back bad changes. We went over the use of svn switch to create mixed-location working copies. And we talked about how one might manage the organization and lifetimes of branches in a repository.

Remember the Subversion mantra: branches and tags are cheap. So don't be afraid to use them when needed!

As a helpful reminder of all the operations we discussed, here is handy reference table you can consult as you begin to make use of branches.

Table 4.1. Branching and merging commands

ActionCommand
Create a branch or tagsvn copy URL1 URL2
Switch a working copy to a branch or tagsvn switch URL
Synchronize a branch with trunksvn merge trunkURL; svn commit
See merge history or eligible changesetssvn mergeinfo SOURCE TARGET
Merge a branch back into trunksvn merge --reintegrate branchURL; svn commit
Merge one specific changesvn merge -c REV URL; svn commit
Merge a range of changessvn merge -r REV1:REV2 URL; svn commit
Block a change from automatic mergingsvn merge -c REV --record-only URL; svn commit
Preview a mergesvn merge URL --dry-run
Abandon merge resultssvn revert -R .
Resurrect something from historysvn copy URL@REV localPATH
Undo a committed changesvn merge -c -REV URL; svn commit
Examine merge-sensitive historysvn log -g; svn blame -g
Create a tag from a working copysvn copy . tagURL
Rearrange a branch or tagsvn move URL1 URL2
Remove a branch or tagsvn delete URL





[25] Subversion does not support copying between different repositories. When using URLs with svn copy or svn move, you can only copy items within the same repository.

[26] This was introduced in svn 1.6.

[27] The svn merge subcommand option --allow-mixed-revisions allows you to override this prohibition, but you should only do so if you understand the ramifications and have a good reason for it.

[28] With Subversion 1.7 you don't absolutely have to do all your sync merges to the root of your branch as we do in this example. If your branch is effectively synced via a series of subtree merges then the reintegrate will work, but ask yourself, if the branch is effectively synced, then why are you doing subtree merges? Doing so is almost always needlessly complex.

[29] Reintegrate merges are allowed if the target is a shallow checkout (see the section called “Sparse Directories”) but any paths affected by the diff which are missing due to the sparse working copy will be skipped, probably not what you intended!

[30] This is often termed an inoperative merge. Though in this example the merge of r1060 would do something: It would update the mergeinfo on the root of the branch, but it would be inoperative in the sense that no diff would be applied.

[31] The Subversion project has plans, however, to someday implement a command that would accomplish the task of permanently deleting information. In the meantime, see the section called “svndumpfilter” for a possible workaround.

[32] At least, this is true in Subversion 1.7 at the time of this writing. This behavior may improve in future versions of Subversion.

[33] Interestingly, after rolling back a revision like this, we wouldn't be able to reapply the revision using svn merge . -c 5, since the mergeinfo would already list r5 as being applied. We would have to use the --ignore-ancestry option to make the merge command ignore the existing mergeinfo!

[34] You can, however, use svn relocate if the URL of your server changes and you don't want to abandon an existing working copy. See svn relocate in Chapter 9, Subversion Complete Reference for more information and an example.

[35] And is entirely bug-free, of course!

[36] Actually, you could use svn merge -rLAST_UPDATED_REV:HEAD . in your working copy to quite literally merge in all the repository changes since your last update if really wanted to!

Chapter 5. Repository Administration

The Subversion repository is the central storehouse of all your versioned data. As such, it becomes an obvious candidate for all the love and attention an administrator can offer. While the repository is generally a low-maintenance item, it is important to understand how to properly configure and care for it so that potential problems are avoided, and so actual problems are safely resolved.

In this chapter, we'll discuss how to create and configure a Subversion repository. We'll also talk about repository maintenance, providing examples of how and when to use various related tools provided with Subversion. We'll address some common questions and mistakes and give some suggestions on how to arrange the data in the repository.

If you plan to access a Subversion repository only in the role of a user whose data is under version control (i.e., via a Subversion client), you can skip this chapter altogether. However, if you are, or wish to become, a Subversion repository administrator,[37] this chapter is for you.

The Subversion Repository, Defined

Before jumping into the broader topic of repository administration, let's further define what a repository is. How does it look? How does it feel? Does it take its tea hot or iced, sweetened, and with lemon? As an administrator, you'll be expected to understand the composition of a repository both from a literal, OS-level perspective—how a repository looks and acts with respect to non-Subversion tools—and from a logical perspective—dealing with how data is represented inside the repository.

Seen through the eyes of a typical file browser application (such as Windows Explorer) or command-line based filesystem navigation tools, the Subversion repository is just another directory full of stuff. There are some subdirectories with human-readable configuration files in them, some subdirectories with some not-so-human-readable data files, and so on. As in other areas of the Subversion design, modularity is given high regard, and hierarchical organization is preferred to cluttered chaos. So a shallow glance into a typical repository from a nuts-and-bolts perspective is sufficient to reveal the basic components of the repository:

$ ls repos
conf/  db/  format  hooks/  locks/  README.txt

Here's a quick fly-by overview of what exactly you're seeing in this directory listing. (Don't get bogged down in the terminology—detailed coverage of these components exists elsewhere in this and other chapters.)

conf/

This directory is a container for configuration files.

db/

This directory contains the data store for all of your versioned data.[38]

format

This file describes the repository's internal organizational scheme. (As it turns out, the db/ subdirectory sometimes also contains a format file which describes only the contents of that subdirectory and which is not to be confused with this file.)

hooks/

This directory contains hook script templates and hook scripts, if any have been installed.

locks/

Subversion uses this directory to house repository lock files, used for managing concurrent access to the repository.

README.txt

This is a brief text file containing merely a notice to readers that the directory they are looking in is a Subversion repository.

[Note]Note

Prior to Subversion 1.5, the on-disk repository structure also always contained a dav subdirectory. mod_dav_svn used this directory to store information about WebDAV activities—mappings of high-level WebDAV protocol concepts to Subversion commit transactions. Subversion 1.5 changed that behavior, moving ownership of the activities directory, and the ability to configure its location, into mod_dav_svn itself. Now, new repositories will not necessarily have a dav subdirectory unless mod_dav_svn is in use and hasn't been configured to store its activities database elsewhere. See the section called “Directives” in Chapter 9, Subversion Complete Reference for more information.

Of course, when accessed via the Subversion libraries, this otherwise unremarkable collection of files and directories suddenly becomes an implementation of a virtual, versioned filesystem, complete with customizable event triggers. This filesystem has its own notions of directories and files, very similar to the notions of such things held by real filesystems (such as NTFS, FAT32, ext3, etc.). But this is a special filesystem—it hangs these directories and files from revisions, keeping all the changes you've ever made to them safely stored and forever accessible. This is where the entirety of your versioned data lives.

Strategies for Repository Deployment

Due largely to the simplicity of the overall design of the Subversion repository and the technologies on which it relies, creating and configuring a repository are fairly straightforward tasks. There are a few preliminary decisions you'll want to make, but the actual work involved in any given setup of a Subversion repository is pretty basic, tending toward mindless repetition if you find yourself setting up multiples of these things.

Some things you'll want to consider beforehand, though, are:

  • What data do you expect to live in your repository (or repositories), and how will that data be organized?

  • Where will your repository live, and how will it be accessed?

  • What types of access control and repository event reporting do you need?

  • Which of the available types of data store do you want to use?

In this section, we'll try to help you answer those questions.

Planning Your Repository Organization

While Subversion allows you to move around versioned files and directories without any loss of information, and even provides ways of moving whole sets of versioned history from one repository to another, doing so can greatly disrupt the workflow of those who access the repository often and come to expect things to be at certain locations. So before creating a new repository, try to peer into the future a bit; plan ahead before placing your data under version control. By conscientiously laying out your repository or repositories and their versioned contents ahead of time, you can prevent many future headaches.

Let's assume that as repository administrator, you will be responsible for supporting the version control system for several projects. Your first decision is whether to use a single repository for multiple projects, or to give each project its own repository, or some compromise of these two.

There are benefits to using a single repository for multiple projects, most obviously the lack of duplicated maintenance. A single repository means that there is one set of hook programs, one thing to routinely back up, one thing to dump and load if Subversion releases an incompatible new version, and so on. Also, you can move data between projects easily, without losing any historical versioning information.

The downside of using a single repository is that different projects may have different requirements in terms of the repository event triggers, such as needing to send commit notification emails to different mailing lists, or having different definitions about what does and does not constitute a legitimate commit. These aren't insurmountable problems, of course—it just means that all of your hook scripts have to be sensitive to the layout of your repository rather than assuming that the whole repository is associated with a single group of people. Also, remember that Subversion uses repository-global revision numbers. While those numbers don't have any particular magical powers, some folks still don't like the fact that even though no changes have been made to their project lately, the youngest revision number for the repository keeps climbing because other projects are actively adding new revisions.[39]

A middle-ground approach can be taken, too. For example, projects can be grouped by how well they relate to each other. You might have a few repositories with a handful of projects in each repository. That way, projects that are likely to want to share data can do so easily, and as new revisions are added to the repository, at least the developers know that those new revisions are at least remotely related to everyone who uses that repository.

After deciding how to organize your projects with respect to repositories, you'll probably want to think about directory hierarchies within the repositories themselves. Because Subversion uses regular directory copies for branching and tagging (see Chapter 4, Branching and Merging), the Subversion community recommends that you choose a repository location for each project root—the topmost directory that contains data related to that project—and then create three subdirectories beneath that root: trunk, meaning the directory under which the main project development occurs; branches, which is a directory in which to create various named branches of the main development line; and tags, which is a collection of tree snapshots that are created, and perhaps destroyed, but never changed.[40]

For example, your repository might look like this:



/

   calc/

      trunk/

      tags/

      branches/

   calendar/

      trunk/

      tags/

      branches/

   spreadsheet/

      trunk/

      tags/

      branches/

   …

Note that it doesn't matter where in your repository each project root is. If you have only one project per repository, the logical place to put each project root is at the root of that project's respective repository. If you have multiple projects, you might want to arrange them in groups inside the repository, perhaps putting projects with similar goals or shared code in the same subdirectory, or maybe just grouping them alphabetically. Such an arrangement might look like this:



/

   utils/

      calc/

         trunk/

         tags/

         branches/

      calendar/

         trunk/

         tags/

         branches/

      …

   office/

      spreadsheet/

         trunk/

         tags/

         branches/

      …

Lay out your repository in whatever way you see fit. Subversion does not expect or enforce a particular layout—in its eyes, a directory is a directory is a directory. Ultimately, you should choose the repository arrangement that meets the needs of the people who work on the projects that live there.

In the name of full disclosure, though, we'll mention another very common layout. In this layout, the trunk, tags, and branches directories live in the root directory of your repository, and your projects are in subdirectories beneath those, like so:



/

   trunk/

      calc/

      calendar/

      spreadsheet/

      …

   tags/

      calc/

      calendar/

      spreadsheet/

      …

   branches/

      calc/

      calendar/

      spreadsheet/

      …

There's nothing particularly incorrect about such a layout, but it may or may not seem as intuitive for your users. Especially in large, multiproject situations with many users, those users may tend to be familiar with only one or two of the projects in the repository. But the projects-as-branch-siblings approach tends to deemphasize project individuality and focus on the entire set of projects as a single entity. That's a social issue, though. We like our originally suggested arrangement for purely practical reasons—it's easier to ask about (or modify, or migrate elsewhere) the entire history of a single project when there's a single repository path that holds the entire history—past, present, tagged, and branched—for that project and that project alone.

Deciding Where and How to Host Your Repository

Before creating your Subversion repository, an obvious question you'll need to answer is where the thing is going to live. This is strongly connected to myriad other questions involving how the repository will be accessed (via a Subversion server or directly), by whom (users behind your corporate firewall or the whole world out on the open Internet), what other services you'll be providing around Subversion (repository browsing interfaces, email-based commit notification, etc.), your data backup strategy, and so on.

We cover server choice and configuration in Chapter 6, Server Configuration, but the point we'd like to briefly make here is simply that the answers to some of these other questions might have implications that force your hand when deciding where your repository will live. For example, certain deployment scenarios might require accessing the repository via a remote filesystem from multiple computers, in which case (as you'll read in the next section) your choice of a repository backend data store turns out not to be a choice at all because only one of the available backends will work in this scenario.

Addressing each possible way to deploy Subversion is both impossible and outside the scope of this book. We simply encourage you to evaluate your options using these pages and other sources as your reference material and to plan ahead.

Choosing a Data Store

Subversion provides two options for the type of underlying data store—often referred to as the backend or, somewhat confusingly, the (versioned) filesystem—that each repository uses. One type of data store keeps everything in a Berkeley DB (or BDB) database environment; repositories that use this type are often referred to as being BDB-backed. The other type stores data in ordinary flat files, using a custom format. Subversion developers have adopted the habit of referring to this latter data storage mechanism as FSFS[41]—a versioned filesystem implementation that uses the native OS filesystem directly—rather than via a database library or some other abstraction layer—to store data.

Table 5.1, “Repository data store comparison” gives a comparative overview of Berkeley DB and FSFS repositories.

Table 5.1. Repository data store comparison

CategoryFeatureBerkeley DBFSFS
ReliabilityData integrityWhen properly deployed, extremely reliable; Berkeley DB 4.4 brings auto-recoveryOlder versions had some rarely demonstrated, but data-destroying bugs
Sensitivity to interruptionsVery; crashes and permission problems can leave the database wedged, requiring journaled recovery proceduresQuite insensitive
AccessibilityUsable from a read-only mountNoYes
Platform-independent storageNoYes
Usable over network filesystemsGenerally, noYes
Group permissions handlingSensitive to user umask problems; best if accessed by only one userWorks around umask problems
ScalabilityRepository disk usageLarger (especially if logfiles aren't purged)Smaller
Number of revision treesDatabase; no problemsSome older native filesystems don't scale well with thousands of entries in a single directory
Directories with many filesSlowerFaster
PerformanceChecking out latest revisionNo meaningful differenceNo meaningful difference
Large commitsSlower overall, but cost is amortized across the lifetime of the commitFaster overall, but finalization delay may cause client timeouts


There are advantages and disadvantages to each of these two backend types. Neither of them is more official than the other, though the newer FSFS is the default data store as of Subversion 1.2. Both are reliable enough to trust with your versioned data. But as you can see in Table 5.1, “Repository data store comparison”, the FSFS backend provides quite a bit more flexibility in terms of its supported deployment scenarios. More flexibility means you have to work a little harder to find ways to deploy it incorrectly. Those reasons—plus the fact that not using Berkeley DB means there's one fewer component in the system—largely explain why today almost everyone uses the FSFS backend when creating new repositories.

Fortunately, most programs that access Subversion repositories are blissfully ignorant of which backend data store is in use. And you aren't even necessarily stuck with your first choice of a data store—in the event that you change your mind later, Subversion provides ways of migrating your repository's data into another repository that uses a different backend data store. We talk more about that later in this chapter.

The following subsections provide a more detailed look at the available backend data store types.

Berkeley DB

When the initial design phase of Subversion was in progress, the developers decided to use Berkeley DB for a variety of reasons, including its open source license, transaction support, reliability, performance, API simplicity, thread safety, support for cursors, and so on.

Berkeley DB provides real transaction support—perhaps its most powerful feature. Multiple processes accessing your Subversion repositories don't have to worry about accidentally clobbering each other's data. The isolation provided by the transaction system is such that for any given operation, the Subversion repository code sees a static view of the database—not a database that is constantly changing at the hand of some other process—and can make decisions based on that view. If the decision made happens to conflict with what another process is doing, the entire operation is rolled back as though it never happened, and Subversion gracefully retries the operation against a new, updated (and yet still static) view of the database.

Another great feature of Berkeley DB is hot backups—the ability to back up the database environment without taking it offline. We'll discuss how to back up your repository later in this chapter (in the section called “Repository Backup”), but the benefits of being able to make fully functional copies of your repositories without any downtime should be obvious.

Berkeley DB is also a very reliable database system when properly used. Subversion uses Berkeley DB's logging facilities, which means that the database first writes to on-disk logfiles a description of any modifications it is about to make, and then makes the modification itself. This is to ensure that if anything goes wrong, the database system can back up to a previous checkpoint—a location in the logfiles known not to be corrupt—and replay transactions until the data is restored to a usable state. See the section called “Managing Disk Space” later in this chapter for more about Berkeley DB logfiles.

But every rose has its thorn, and so we must note some known limitations of Berkeley DB. First, Berkeley DB environments are not portable. You cannot simply copy a Subversion repository that was created on a Unix system onto a Windows system and expect it to work. While much of the Berkeley DB database format is architecture-independent, other aspects of the environment are not. Second, Subversion uses Berkeley DB in a way that will not operate on Windows 95/98 systems—if you need to house a BDB-backed repository on a Windows machine, stick with Windows 2000 or later.

While Berkeley DB promises to behave correctly on network shares that meet a particular set of specifications,[42] most networked filesystem types and appliances do not actually meet those requirements. And in no case can you allow a BDB-backed repository that resides on a network share to be accessed by multiple clients of that share at once (which quite often is the whole point of having the repository live on a network share in the first place).

[Warning]Warning

If you attempt to use Berkeley DB on a noncompliant remote filesystem, the results are unpredictable—you may see mysterious errors right away, or it may be months before you discover that your repository database is subtly corrupted. You should strongly consider using the FSFS data store for repositories that need to live on a network share.

Finally, because Berkeley DB is a library linked directly into Subversion, it's more sensitive to interruptions than a typical relational database system. Most SQL systems, for example, have a dedicated server process that mediates all access to tables. If a program accessing the database crashes for some reason, the database daemon notices the lost connection and cleans up any mess left behind. And because the database daemon is the only process accessing the tables, applications don't need to worry about permission conflicts. These things are not the case with Berkeley DB, however. Subversion (and programs using Subversion libraries) access the database tables directly, which means that a program crash can leave the database in a temporarily inconsistent, inaccessible state. When this happens, an administrator needs to ask Berkeley DB to restore to a checkpoint, which is a bit of an annoyance. Other things can cause a repository to wedge besides crashed processes, such as programs conflicting over ownership and permissions on the database files.

[Note]Note

Berkeley DB 4.4 brings (to Subversion 1.4 and later) the ability for Subversion to automatically and transparently recover Berkeley DB environments in need of such recovery. When a Subversion process attaches to a repository's Berkeley DB environment, it uses some process accounting mechanisms to detect any unclean disconnections by previous processes, performs any necessary recovery, and then continues on as though nothing happened. This doesn't completely eliminate instances of repository wedging, but it does drastically reduce the amount of human interaction required to recover from them.

So while a Berkeley DB repository is quite fast and scalable, it's best used by a single server process running as one user—such as Apache's httpd or svnserve (see Chapter 6, Server Configuration)—rather than accessing it as many different users via file:// or svn+ssh:// URLs. If you're accessing a Berkeley DB repository directly as multiple users, be sure to read the section called “Supporting Multiple Repository Access Methods” later in this chapter.

FSFS

In mid-2004, a second type of repository storage system—one that doesn't use a database at all—came into being. An FSFS repository stores the changes associated with a revision in a single file, and so all of a repository's revisions can be found in a single subdirectory full of numbered files. Transactions are created in separate subdirectories as individual files. When complete, the transaction file is renamed and moved into the revisions directory, thus guaranteeing that commits are atomic. And because a revision file is permanent and unchanging, the repository also can be backed up while hot, just like a BDB-backed repository.

The FSFS revision files describe a revision's directory structure, file contents, and deltas against files in other revision trees. Unlike a Berkeley DB database, this storage format is portable across different operating systems and isn't sensitive to CPU architecture. Because no journaling or shared-memory files are being used, the repository can be safely accessed over a network filesystem and examined in a read-only environment. The lack of database overhead also means the overall repository size is a bit smaller.

FSFS has different performance characteristics, too. When committing a directory with a huge number of files, FSFS is able to more quickly append directory entries. On the other hand, FSFS has a longer delay when finalizing a commit while it performs tasks that the BDB backend amortizes across the lifetime of the commit, which could in extreme cases cause clients to time out while waiting for a response.

The most important distinction, however, is FSFS's imperviousness to wedging when something goes wrong. If a process using a Berkeley DB database runs into a permissions problem or suddenly crashes, the database can be left in an unusable state until an administrator recovers it. If the same scenarios happen to a process using an FSFS repository, the repository isn't affected at all. At worst, some transaction data is left behind.

Creating and Configuring Your Repository

Earlier in this chapter (in the section called “Strategies for Repository Deployment”), we looked at some of the important decisions that should be made before creating and configuring your Subversion repository. Now, we finally get to get our hands dirty! In this section, we'll see how to actually create a Subversion repository and configure it to perform custom actions when special repository events occur.

Creating the Repository

Subversion repository creation is an incredibly simple task. The svnadmin utility that comes with Subversion provides a subcommand (svnadmin create) for doing just that.

$ # Create a repository
$ svnadmin create /var/svn/repos
$

Assuming that the parent directory /var/svn exists and that you have sufficient permissions to modify that directory, the previous command creates a new repository in the directory /var/svn/repos, and with the default filesystem data store (FSFS). You can explicitly choose the filesystem type using the --fs-type argument, which accepts as a parameter either fsfs or bdb.

$ # Create an FSFS-backed repository
$ svnadmin create --fs-type fsfs /var/svn/repos
$
# Create a Berkeley-DB-backed repository
$ svnadmin create --fs-type bdb /var/svn/repos
$

After running this simple command, you have a Subversion repository. Depending on how users will access this new repository, you might need to fiddle with its filesystem permissions. But since basic system administration is rather outside the scope of this text, we'll leave further exploration of that topic as an exercise to the reader.

[Tip]Tip

The path argument to svnadmin is just a regular filesystem path and not a URL like the svn client program uses when referring to repositories. Both svnadmin and svnlook are considered server-side utilities—they are used on the machine where the repository resides to examine or modify aspects of the repository, and are in fact unable to perform tasks across a network. A common mistake made by Subversion newcomers is trying to pass URLs (even local file:// ones) to these two programs.

Present in the db/ subdirectory of your repository is the implementation of the versioned filesystem. Your new repository's versioned filesystem begins life at revision 0, which is defined to consist of nothing but the top-level root (/) directory. Initially, revision 0 also has a single revision property, svn:date, set to the time at which the repository was created.

Now that you have a repository, it's time to customize it.

[Warning]Warning

While some parts of a Subversion repository—such as the configuration files and hook scripts—are meant to be examined and modified manually, you shouldn't (and shouldn't need to) tamper with the other parts of the repository by hand. The svnadmin tool should be sufficient for any changes necessary to your repository, or you can look to third-party tools (such as Berkeley DB's tool suite) for tweaking relevant subsections of the repository. Do not attempt manual manipulation of your version control history by poking and prodding around in your repository's data store files!

Implementing Repository Hooks

A hook is a program triggered by some repository event, such as the creation of a new revision or the modification of an unversioned property. Some hooks (the so-called pre hooks) run in advance of a repository operation and provide a means by which to both report what is about to happen and prevent it from happening at all. Other hooks (the post hooks) run after the completion of a repository event and are useful for performing tasks that examine—but don't modify—the repository. Each hook is handed enough information to tell what that event is (or was), the specific repository changes proposed (or completed), and the username of the person who triggered the event.

The hooks subdirectory is, by default, filled with templates for various repository hooks:

$ ls repos/hooks/
post-commit.tmpl          post-unlock.tmpl  pre-revprop-change.tmpl
post-lock.tmpl            pre-commit.tmpl   pre-unlock.tmpl
post-revprop-change.tmpl  pre-lock.tmpl     start-commit.tmpl
$

There is one template for each hook that the Subversion repository supports; by examining the contents of those template scripts, you can see what triggers each script to run and what data is passed to that script. Also present in many of these templates are examples of how one might use that script, in conjunction with other Subversion-supplied programs, to perform common useful tasks. To actually install a working hook, you need only place some executable program or script into the repos/hooks directory, which can be executed as the name (such as start-commit or post-commit) of the hook.

On Unix platforms, this means supplying a script or program (which could be a shell script, a Python program, a compiled C binary, or any number of other things) named exactly like the name of the hook. Of course, the template files are present for more than just informational purposes—the easiest way to install a hook on Unix platforms is to simply copy the appropriate template file to a new file that lacks the .tmpl extension, customize the hook's contents, and ensure that the script is executable. Windows, however, uses file extensions to determine whether a program is executable, so you would need to supply a program whose basename is the name of the hook and whose extension is one of the special extensions recognized by Windows for executable programs, such as .exe for programs and .bat for batch files.

[Tip]Tip

For security reasons, the Subversion repository executes hook programs with an empty environment—that is, no environment variables are set at all, not even $PATH (or %PATH%, under Windows). Because of this, many administrators are baffled when their hook program runs fine by hand, but doesn't work when run by Subversion. Be sure to explicitly set any necessary environment variables in your hook program and/or use absolute paths to programs.

Subversion executes hooks as the same user who owns the process that is accessing the Subversion repository. In most cases, the repository is being accessed via a Subversion server, so this user is the same user as whom the server runs on the system. The hooks themselves will need to be configured with OS-level permissions that allow that user to execute them. Also, this means that any programs or files (including the Subversion repository) accessed directly or indirectly by the hook will be accessed as the same user. In other words, be alert to potential permission-related problems that could prevent the hook from performing the tasks it is designed to perform.

There are several hooks implemented by the Subversion repository, and you can get details about each of them in the section called “Repository Hooks” in Chapter 9, Subversion Complete Reference. As a repository administrator, you'll need to decide which hooks you wish to implement (by way of providing an appropriately named and permissioned hook program), and how. When you make this decision, keep in mind the big picture of how your repository is deployed. For example, if you are using server configuration to determine which users are permitted to commit changes to your repository, you don't need to do this sort of access control via the hook system.

There is no shortage of Subversion hook programs and scripts that are freely available either from the Subversion community itself or elsewhere. These scripts cover a wide range of utility—basic access control, policy adherence checking, issue tracker integration, email- or syndication-based commit notification, and beyond. Or, if you wish to write your own, see Chapter 8, Embedding Subversion.

[Warning]Warning

While hook scripts can do almost anything, there is one dimension in which hook script authors should show restraint: do not modify a commit transaction using hook scripts. While it might be tempting to use hook scripts to automatically correct errors, shortcomings, or policy violations present in the files being committed, doing so can cause problems. Subversion keeps client-side caches of certain bits of repository data, and if you change a commit transaction in this way, those caches become indetectably stale. This inconsistency can lead to surprising and unexpected behavior. Instead of modifying the transaction, you should simply validate the transaction in the pre-commit hook and reject the commit if it does not meet the desired requirements. As a bonus, your users will learn the value of careful, compliance-minded work habits.

Berkeley DB Configuration

A Berkeley DB environment is an encapsulation of one or more databases, logfiles, region files, and configuration files. The Berkeley DB environment has its own set of default configuration values for things such as the number of database locks allowed to be taken out at any given time, the maximum size of the journaling logfiles, and so on. Subversion's filesystem logic additionally chooses default values for some of the Berkeley DB configuration options. However, sometimes your particular repository, with its unique collection of data and access patterns, might require a different set of configuration option values.

The producers of Berkeley DB understand that different applications and database environments have different requirements, so they have provided a mechanism for overriding at runtime many of the configuration values for the Berkeley DB environment. BDB checks for the presence of a file named DB_CONFIG in the environment directory (namely, the repository's db subdirectory), and parses the options found in that file. Subversion itself creates this file when it creates the rest of the repository. The file initially contains some default options, as well as pointers to the Berkeley DB online documentation so that you can read about what those options do. Of course, you are free to add any of the supported Berkeley DB options to your DB_CONFIG file. Just be aware that while Subversion never attempts to read or interpret the contents of the file and makes no direct use of the option settings in it, you'll want to avoid any configuration changes that may cause Berkeley DB to behave in a fashion that is at odds with what Subversion might expect. Also, changes made to DB_CONFIG won't take effect until you recover the database environment (using svnadmin recover).

FSFS Configuration

As of Subversion 1.6, FSFS filesystems have several configurable parameters which an administrator can use to fine-tune the performance or disk usage of their repositories. You can find these options—and the documentation for them—in the db/fsfs.conf file in the repository.

Repository Maintenance

Maintaining a Subversion repository can be daunting, mostly due to the complexities inherent in systems that have a database backend. Doing the task well is all about knowing the tools—what they are, when to use them, and how. This section will introduce you to the repository administration tools provided by Subversion and discuss how to wield them to accomplish tasks such as repository data migration, upgrades, backups, and cleanups.

An Administrator's Toolkit

Subversion provides a handful of utilities useful for creating, inspecting, modifying, and repairing your repository. Let's look more closely at each of those tools. Afterward, we'll briefly examine some of the utilities included in the Berkeley DB distribution that provide functionality specific to your repository's database backend not otherwise provided by Subversion's own tools.

svnadmin

The svnadmin program is the repository administrator's best friend. Besides providing the ability to create Subversion repositories, this program allows you to perform several maintenance operations on those repositories. The syntax of svnadmin is similar to that of other Subversion command-line programs:

$ svnadmin help
general usage: svnadmin SUBCOMMAND REPOS_PATH  [ARGS & OPTIONS ...]
Type 'svnadmin help <subcommand>' for help on a specific subcommand.
Type 'svnadmin --version' to see the program version and FS modules.

Available subcommands:
   crashtest
   create
   deltify
…

Previously in this chapter (in the section called “Creating the Repository”), we were introduced to the svnadmin create subcommand. Most of the other svnadmin subcommands we will cover later in this chapter. And you can consult the section called “svnadmin—Subversion Repository Administration” in Chapter 9, Subversion Complete Reference for a full rundown of subcommands and what each of them offers.

svnlook

svnlook is a tool provided by Subversion for examining the various revisions and transactions (which are revisions in the making) in a repository. No part of this program attempts to change the repository. svnlook is typically used by the repository hooks for reporting the changes that are about to be committed (in the case of the pre-commit hook) or that were just committed (in the case of the post-commit hook) to the repository. A repository administrator may use this tool for diagnostic purposes.

svnlook has a straightforward syntax:

$ svnlook help
general usage: svnlook SUBCOMMAND REPOS_PATH [ARGS & OPTIONS ...]
Note: any subcommand which takes the '--revision' and '--transaction'
      options will, if invoked without one of those options, act on
      the repository's youngest revision.
Type 'svnlook help <subcommand>' for help on a specific subcommand.
Type 'svnlook --version' to see the program version and FS modules.
…

Most of svnlook's subcommands can operate on either a revision or a transaction tree, printing information about the tree itself, or how it differs from the previous revision of the repository. You use the --revision (-r) and --transaction (-t) options to specify which revision or transaction, respectively, to examine. In the absence of both the --revision (-r) and --transaction (-t) options, svnlook will examine the youngest (or HEAD) revision in the repository. So the following two commands do exactly the same thing when 19 is the youngest revision in the repository located at /var/svn/repos:

$ svnlook info /var/svn/repos
$ svnlook info /var/svn/repos -r 19

One exception to these rules about subcommands is the svnlook youngest subcommand, which takes no options and simply prints out the repository's youngest revision number:

$ svnlook youngest /var/svn/repos
19
$
[Note]Note

Keep in mind that the only transactions you can browse are uncommitted ones. Most repositories will have no such transactions because transactions are usually either committed (in which case, you should access them as revision with the --revision (-r) option) or aborted and removed.

Output from svnlook is designed to be both human- and machine-parsable. Take, as an example, the output of the svnlook info subcommand:

$ svnlook info /var/svn/repos
sally
2002-11-04 09:29:13 -0600 (Mon, 04 Nov 2002)
27
Added the usual
Greek tree.
$

The output of svnlook info consists of the following, in the order given:

  1. The author, followed by a newline

  2. The date, followed by a newline

  3. The number of characters in the log message, followed by a newline

  4. The log message itself, followed by a newline

This output is human-readable, meaning items such as the datestamp are displayed using a textual representation instead of something more obscure (such as the number of nanoseconds since the Tastee Freez guy drove by). But the output is also machine-parsable—because the log message can contain multiple lines and be unbounded in length, svnlook provides the length of that message before the message itself. This allows scripts and other wrappers around this command to make intelligent decisions about the log message, such as how much memory to allocate for the message, or at least how many bytes to skip in the event that this output is not the last bit of data in the stream.

svnlook can perform a variety of other queries: displaying subsets of bits of information we've mentioned previously, recursively listing versioned directory trees, reporting which paths were modified in a given revision or transaction, showing textual and property differences made to files and directories, and so on. See the section called “svnlook—Subversion Repository Examination” in Chapter 9, Subversion Complete Reference for a full reference of svnlook's features.

svndumpfilter

While it won't be the most commonly used tool at the administrator's disposal, svndumpfilter provides a very particular brand of useful functionality—the ability to quickly and easily modify streams of Subversion repository history data by acting as a path-based filter.

The syntax of svndumpfilter is as follows:

$ svndumpfilter help
general usage: svndumpfilter SUBCOMMAND [ARGS & OPTIONS ...]
Type 'svndumpfilter help <subcommand>' for help on a specific subcommand.
Type 'svndumpfilter --version' to see the program version.
  
Available subcommands:
   exclude
   include
   help (?, h)

There are only two interesting subcommands: svndumpfilter exclude and svndumpfilter include. They allow you to make the choice between implicit or explicit inclusion of paths in the stream. You can learn more about these subcommands and svndumpfilter's unique purpose later in this chapter, in the section called “Filtering Repository History”.

svnrdump

The svnrdump program is, to put it simply, essentially just network-aware flavors of the svnadmin dump and svnadmin load subcommands, rolled up into a separate program.

$ svnrdump help
general usage: svnrdump SUBCOMMAND URL [-r LOWER[:UPPER]]
Type 'svnrdump help <subcommand>' for help on a specific subcommand.
Type 'svnrdump --version' to see the program version and RA modules.

Available subcommands:
   dump
   load
   help (?, h)

$

We discuss the use of svnrdump and the aforementioned svnadmin commands later in this chapter (see the section called “Migrating Repository Data Elsewhere”).

svnsync

The svnsync program provides all the functionality required for maintaining a read-only mirror of a Subversion repository. The program really has one job—to transfer one repository's versioned history into another repository. And while there are few ways to do that, its primary strength is that it can operate remotely—the source and sink[43] repositories may be on different computers from each other and from svnsync itself.

As you might expect, svnsync has a syntax that looks very much like every other program we've mentioned in this chapter:

$ svnsync help
general usage: svnsync SUBCOMMAND DEST_URL  [ARGS & OPTIONS ...]
Type 'svnsync help <subcommand>' for help on a specific subcommand.
Type 'svnsync --version' to see the program version and RA modules.

Available subcommands:
   initialize (init)
   synchronize (sync)
   copy-revprops
   info
   help (?, h)
$

We talk more about replicating repositories with svnsync later in this chapter (see the section called “Repository Replication”).

fsfs-reshard.py

While not an official member of the Subversion toolchain, the fsfs-reshard.py script (found in the tools/server-side directory of the Subversion source distribution) is a useful performance tuning tool for administrators of FSFS-backed Subversion repositories. As described in the sidebar Revision files and shards, FSFS repositories use individual files to house information about each revision. Sometimes these files all live in a single directory; sometimes they are sharded across many directories. But the neat thing is that the number of directories used to house these files is configurable. That's where fsfs-reshard.py comes in.

fsfs-reshard.py reshuffles the repository's file structure into a new arrangement that reflects the requested number of sharding subdirectories and updates the repository configuration to preserve this change. When used in conjunction with the svnadmin upgrade command, this is especially useful for upgrading a pre-1.5 Subversion (unsharded) repository to the latest filesystem format and sharding its data files (which Subversion will not automatically do for you). This script can also be used for fine-tuning an already sharded repository.

Berkeley DB utilities

If you're using a Berkeley DB repository, all of your versioned filesystem's structure and data live in a set of database tables within the db/ subdirectory of your repository. This subdirectory is a regular Berkeley DB environment directory and can therefore be used in conjunction with any of the Berkeley database tools, typically provided as part of the Berkeley DB distribution.

For day-to-day Subversion use, these tools are unnecessary. Most of the functionality typically needed for Subversion repositories has been duplicated in the svnadmin tool. For example, svnadmin list-unused-dblogs and svnadmin list-dblogs perform a subset of what is provided by the Berkeley db_archive utility, and svnadmin recover reflects the common use cases of the db_recover utility.

However, there are still a few Berkeley DB utilities that you might find useful. The db_dump and db_load programs write and read, respectively, a custom file format that describes the keys and values in a Berkeley DB database. Since Berkeley databases are not portable across machine architectures, this format is a useful way to transfer those databases from machine to machine, irrespective of architecture or operating system. As we describe later in this chapter, you can also use svnadmin dump and svnadmin load for similar purposes, but db_dump and db_load can do certain jobs just as well and much faster. They can also be useful if the experienced Berkeley DB hacker needs to do in-place tweaking of the data in a BDB-backed repository for some reason, which is something Subversion's utilities won't allow. Also, the db_stat utility can provide useful information about the status of your Berkeley DB environment, including detailed statistics about the locking and storage subsystems.

For more information on the Berkeley DB tool chain, visit the documentation section of the Berkeley DB section of Oracle's web site, located at http://www.oracle.com/technology/documentation/berkeley-db/db/.

Commit Log Message Correction

Sometimes a user will have an error in her log message (a misspelling or some misinformation, perhaps). If the repository is configured (using the pre-revprop-change hook; see the section called “Implementing Repository Hooks”) to accept changes to this log message after the commit is finished, the user can fix her log message remotely using svn propset (see svn propset (pset, ps) in Chapter 9, Subversion Complete Reference). However, because of the potential to lose information forever, Subversion repositories are not, by default, configured to allow changes to unversioned properties—except by an administrator.

If a log message needs to be changed by an administrator, this can be done using svnadmin setlog. This command changes the log message (the svn:log property) on a given revision of a repository, reading the new value from a provided file.

$ echo "Here is the new, correct log message" > newlog.txt
$ svnadmin setlog myrepos newlog.txt -r 388

The svnadmin setlog command, by default, is still bound by the same protections against modifying unversioned properties as a remote client is—the pre-revprop-change and post-revprop-change hooks are still triggered, and therefore must be set up to accept changes of this nature. But an administrator can get around these protections by passing the --bypass-hooks option to the svnadmin setlog command.

[Warning]Warning

Remember, though, that by bypassing the hooks, you are likely avoiding such things as email notifications of property changes, backup systems that track unversioned property changes, and so on. In other words, be very careful about what you are changing, and how you change it.

Managing Disk Space

While the cost of storage has dropped incredibly in the past few years, disk usage is still a valid concern for administrators seeking to version large amounts of data. Every bit of version history information stored in the live repository needs to be backed up elsewhere, perhaps multiple times as part of rotating backup schedules. It is useful to know what pieces of Subversion's repository data need to remain on the live site, which need to be backed up, and which can be safely removed.

How Subversion saves disk space

To keep the repository small, Subversion uses deltification (or delta-based storage) within the repository itself. Deltification involves encoding the representation of a chunk of data as a collection of differences against some other chunk of data. If the two pieces of data are very similar, this deltification results in storage savings for the deltified chunk—rather than taking up space equal to the size of the original data, it takes up only enough space to say, I look just like this other piece of data over here, except for the following couple of changes. The result is that most of the repository data that tends to be bulky—namely, the contents of versioned files—is stored at a much smaller size than the original full-text representation of that data.

While deltified storage has been a part of Subversion's design since the very beginning, there have been additional improvements made over the years. Subversion repositories created with Subversion 1.4 or later benefit from compression of the full-text representations of file contents. Repositories created with Subversion 1.6 or later further enjoy the disk space savings afforded by representation sharing, a feature which allows multiple files or file revisions with identical file content to refer to a single shared instance of that data rather than each having their own distinct copy thereof.

[Note]Note

Because all of the data that is subject to deltification in a BDB-backed repository is stored in a single Berkeley DB database file, reducing the size of the stored values will not immediately reduce the size of the database file itself. Berkeley DB will, however, keep internal records of unused areas of the database file and consume those areas first before growing the size of the database file. So while deltification doesn't produce immediate space savings, it can drastically slow future growth of the database.

Removing dead transactions

Though they are uncommon, there are circumstances in which a Subversion commit process might fail, leaving behind in the repository the remnants of the revision-to-be that wasn't—an uncommitted transaction and all the file and directory changes associated with it. This could happen for several reasons: perhaps the client operation was inelegantly terminated by the user, or a network failure occurred in the middle of an operation. Regardless of the reason, dead transactions can happen. They don't do any real harm, other than consuming disk space. A fastidious administrator may nonetheless wish to remove them.

You can use the svnadmin lstxns command to list the names of the currently outstanding transactions:

$ svnadmin lstxns myrepos
19
3a1
a45
$

Each item in the resultant output can then be used with svnlook (and its --transaction (-t) option) to determine who created the transaction, when it was created, what types of changes were made in the transaction—information that is helpful in determining whether the transaction is a safe candidate for removal! If you do indeed want to remove a transaction, its name can be passed to svnadmin rmtxns, which will perform the cleanup of the transaction. In fact, svnadmin rmtxns can take its input directly from the output of svnadmin lstxns!

$ svnadmin rmtxns myrepos `svnadmin lstxns myrepos`
$

If you use these two subcommands like this, you should consider making your repository temporarily inaccessible to clients. That way, no one can begin a legitimate transaction before you start your cleanup. Example 5.1, “txn-info.sh (reporting outstanding transactions)” contains a bit of shell-scripting that can quickly generate information about each outstanding transaction in your repository.

Example 5.1. txn-info.sh (reporting outstanding transactions)

#!/bin/sh

### Generate informational output for all outstanding transactions in
### a Subversion repository.

REPOS="${1}"
if [ "x$REPOS" = x ] ; then
  echo "usage: $0 REPOS_PATH"
  exit
fi

for TXN in `svnadmin lstxns ${REPOS}`; do 
  echo "---[ Transaction ${TXN} ]-------------------------------------------"
  svnlook info "${REPOS}" -t "${TXN}"
done


The output of the script is basically a concatenation of several chunks of svnlook info output (see the section called “svnlook”) and will look something like this:

$ txn-info.sh myrepos
---[ Transaction 19 ]-------------------------------------------
sally
2001-09-04 11:57:19 -0500 (Tue, 04 Sep 2001)
0
---[ Transaction 3a1 ]-------------------------------------------
harry
2001-09-10 16:50:30 -0500 (Mon, 10 Sep 2001)
39
Trying to commit over a faulty network.
---[ Transaction a45 ]-------------------------------------------
sally
2001-09-12 11:09:28 -0500 (Wed, 12 Sep 2001)
0
$

A long-abandoned transaction usually represents some sort of failed or interrupted commit. A transaction's datestamp can provide interesting information—for example, how likely is it that an operation begun nine months ago is still active?

In short, transaction cleanup decisions need not be made unwisely. Various sources of information—including Apache's error and access logs, Subversion's operational logs, Subversion revision history, and so on—can be employed in the decision-making process. And of course, an administrator can often simply communicate with a seemingly dead transaction's owner (via email, e.g.) to verify that the transaction is, in fact, in a zombie state.

Purging unused Berkeley DB logfiles

Until recently, the largest offender of disk space usage with respect to BDB-backed Subversion repositories were the logfiles in which Berkeley DB performs its prewrites before modifying the actual database files. These files capture all the actions taken along the route of changing the database from one state to another—while the database files, at any given time, reflect a particular state, the logfiles contain all of the many changes along the way between states. Thus, they can grow and accumulate quite rapidly.

Fortunately, beginning with the 4.2 release of Berkeley DB, the database environment has the ability to remove its own unused logfiles automatically. Any repositories created using svnadmin when compiled against Berkeley DB version 4.2 or later will be configured for this automatic logfile removal. If you don't want this feature enabled, simply pass the --bdb-log-keep option to the svnadmin create command. If you forget to do this or change your mind at a later time, simply edit the DB_CONFIG file found in your repository's db directory, comment out the line that contains the set_flags DB_LOG_AUTOREMOVE directive, and then run svnadmin recover on your repository to force the configuration changes to take effect. See the section called “Berkeley DB Configuration” for more information about database configuration.

Without some sort of automatic logfile removal in place, logfiles will accumulate as you use your repository. This is actually somewhat of a feature of the database system—you should be able to recreate your entire database using nothing but the logfiles, so these files can be useful for catastrophic database recovery. But typically, you'll want to archive the logfiles that are no longer in use by Berkeley DB, and then remove them from disk to conserve space. Use the svnadmin list-unused-dblogs command to list the unused logfiles:

$ svnadmin list-unused-dblogs /var/svn/repos
/var/svn/repos/log.0000000031
/var/svn/repos/log.0000000032
/var/svn/repos/log.0000000033
…
$ rm `svnadmin list-unused-dblogs /var/svn/repos`
## disk space reclaimed!
[Warning]Warning

BDB-backed repositories whose logfiles are used as part of a backup or disaster recovery plan should not make use of the logfile autoremoval feature. Reconstruction of a repository's data from logfiles can only be accomplished only when all the logfiles are available. If some of the logfiles are removed from disk before the backup system has a chance to copy them elsewhere, the incomplete set of backed-up logfiles is essentially useless.

Packing FSFS filesystems

As described in the sidebar Revision files and shards, FSFS-backed Subversion repositories create, by default, a new on-disk file for each revision added to the repository. Having thousands of these files present on your Subversion server—even when housed in separate shard directories—can lead to inefficiencies.

The first problem is that the operating system has to reference many different files over a short period of time. This leads to inefficient use of disk caches and, as a result, more time spent seeking across large disks. Because of this, Subversion pays a performance penalty when accessing your versioned data.

The second problem is a bit more subtle. Because of the ways that most filesystems allocate disk space, each file claims more space on the disk than it actually uses. The amount of extra space required to house a single file can average anywhere from 2 to 16 kilobytes per file, depending on the underlying filesystem in use. This translates directly into a per-revision disk usage penalty for FSFS-backed repositories. The effect is most pronounced in repositories which have many small revisions, since the overhead involved in storing the revision file quickly outgrows the size of the actual data being stored.

To solve these problems, Subversion 1.6 introduced the svnadmin pack command. By concatenating all the files of a completed shard into a single pack file and then removing the original per-revision files, svnadmin pack reduces the file count within a given shard down to just a single file. In doing so, it aids filesystem caches and reduces (to one) the number of times a file storage overhead penalty is paid.

Subversion can pack existing sharded repositories which have been upgraded to the 1.6 filesystem format or later (see svnadmin upgrade) in Chapter 9, Subversion Complete Reference. To do so, just run svnadmin pack on the repository:

$ svnadmin pack /var/svn/repos
Packing shard 0...done.
Packing shard 1...done.
Packing shard 2...done.
…
Packing shard 34...done.
Packing shard 35...done.
Packing shard 36...done.
$

Because the packing process obtains the required locks before doing its work, you can run it on live repositories, or even as part of a post-commit hook. Repacking packed shards is legal, but will have no effect on the disk usage of the repository.

svnadmin pack has no effect on BDB-backed Subversion repositories.

Berkeley DB Recovery

As mentioned in the section called “Berkeley DB”, a Berkeley DB repository can sometimes be left in a frozen state if not closed properly. When this happens, an administrator needs to rewind the database back into a consistent state. This is unique to BDB-backed repositories, though—if you are using FSFS-backed ones instead, this won't apply to you. And for those of you using Subversion 1.4 with Berkeley DB 4.4 or later, you should find that Subversion has become much more resilient in these types of situations. Still, wedged Berkeley DB repositories do occur, and an administrator needs to know how to safely deal with this circumstance.

To protect the data in your repository, Berkeley DB uses a locking mechanism. This mechanism ensures that portions of the database are not simultaneously modified by multiple database accessors, and that each process sees the data in the correct state when that data is being read from the database. When a process needs to change something in the database, it first checks for the existence of a lock on the target data. If the data is not locked, the process locks the data, makes the change it wants to make, and then unlocks the data. Other processes are forced to wait until that lock is removed before they are permitted to continue accessing that section of the database. (This has nothing to do with the locks that you, as a user, can apply to versioned files within the repository; we try to clear up the confusion caused by this terminology collision in the sidebar The Three Meanings of Lock.)

In the course of using your Subversion repository, fatal errors or interruptions can prevent a process from having the chance to remove the locks it has placed in the database. The result is that the backend database system gets wedged. When this happens, any attempts to access the repository hang indefinitely (since each new accessor is waiting for a lock to go away—which isn't going to happen).

If this happens to your repository, don't panic. The Berkeley DB filesystem takes advantage of database transactions, checkpoints, and prewrite journaling to ensure that only the most catastrophic of events[44] can permanently destroy a database environment. A sufficiently paranoid repository administrator will have made off-site backups of the repository data in some fashion, but don't head off to the tape backup storage closet just yet.

Instead, use the following recipe to attempt to unwedge your repository:

  1. Make sure no processes are accessing (or attempting to access) the repository. For networked repositories, this also means shutting down the Apache HTTP Server or svnserve daemon.

  2. Become the user who owns and manages the repository. This is important, as recovering a repository while running as the wrong user can tweak the permissions of the repository's files in such a way that your repository will still be inaccessible even after it is unwedged.

  3. Run the command svnadmin recover /var/svn/repos. You should see output such as this:

    Repository lock acquired.
    Please wait; recovering the repository may take some time...
    
    Recovery completed.
    The latest repos revision is 19.
    

    This command may take many minutes to complete.

  4. Restart the server process.

This procedure fixes almost every case of repository wedging. Make sure that you run this command as the user that owns and manages the database, not just as root. Part of the recovery process might involve re-creating from scratch various database files (shared memory regions, e.g.). Recovering as root will create those files such that they are owned by root, which means that even after you restore connectivity to your repository, regular users will be unable to access it.

If the previous procedure, for some reason, does not successfully unwedge your repository, you should do two things. First, move your broken repository directory aside (perhaps by renaming it to something like repos.BROKEN) and then restore your latest backup of it. Then, send an email to the Subversion users mailing list (at ) describing your problem in detail. Data integrity is an extremely high priority to the Subversion developers.

Migrating Repository Data Elsewhere

A Subversion filesystem has its data spread throughout files in the repository, in a fashion generally understood by (and of interest to) only the Subversion developers themselves. However, circumstances may arise that call for all, or some subset, of that data to be copied or moved into another repository.

Subversion provides such functionality by way of repository dump streams. A repository dump stream (often referred to as a dump file when stored as a file on disk) is a portable, flat file format that describes the various revisions in your repository—what was changed, by whom, when, and so on. This dump stream is the primary mechanism used to marshal versioned history—in whole or in part, with or without modification—between repositories. And Subversion provides the tools necessary for creating and loading these dump streams: the svnadmin dump and svnadmin load subcommands, respectively, and the svnrdump program.

[Warning]Warning

While the Subversion repository dump format contains human-readable portions and a familiar structure (it resembles an RFC 822 format, the same type of format used for most email), it is not a plain-text file format. It is a binary file format, highly sensitive to meddling. For example, many text editors will corrupt the file by automatically converting line endings.

There are many reasons for dumping and loading Subversion repository data. Early in Subversion's life, the most common reason was due to the evolution of Subversion itself. As Subversion matured, there were times when changes made to the backend database schema caused compatibility issues with previous versions of the repository, so users had to dump their repository data using the previous version of Subversion and load it into a freshly created repository with the new version of Subversion. Now, these types of schema changes haven't occurred since Subversion's 1.0 release, and the Subversion developers promise not to force users to dump and load their repositories when upgrading between minor versions (such as from 1.3 to 1.4) of Subversion. But there are still other reasons for dumping and loading, including re-deploying a Berkeley DB repository on a new OS or CPU architecture, switching between the Berkeley DB and FSFS backends, or (as we'll cover later in this chapter in the section called “Filtering Repository History”) purging versioned data from repository history.

[Note]Note

The Subversion repository dump format describes versioned repository changes only. It will not carry any information about uncommitted transactions, user locks on filesystem paths, repository or server configuration customizations (including hook scripts), and so on.

The Subversion repository dump format also enables conversion from a different storage mechanism or version control system altogether. Because the dump file format is, for the most part, human-readable, it should be relatively easy to describe generic sets of changes—each of which should be treated as a new revision—using this file format. In fact, the cvs2svn utility (see the section called “Converting a Repository from CVS to Subversion”) uses the dump format to represent the contents of a CVS repository so that those contents can be copied into a Subversion repository.

For now, we'll concern ourselves only with migration of repository data between Subversion repositories, which we'll describe in detail in the sections which follow.

Repository data migration using svnadmin

Whatever your reason for migrating repository history, using the svnadmin dump and svnadmin load subcommands is straightforward. svnadmin dump will output a range of repository revisions that are formatted using Subversion's custom filesystem dump format. The dump format is printed to the standard output stream, while informative messages are printed to the standard error stream. This allows you to redirect the output stream to a file while watching the status output in your terminal window. For example:

$ svnlook youngest myrepos
26
$ svnadmin dump myrepos > dumpfile
* Dumped revision 0.
* Dumped revision 1.
* Dumped revision 2.
…
* Dumped revision 25.
* Dumped revision 26.

At the end of the process, you will have a single file (dumpfile in the previous example) that contains all the data stored in your repository in the requested range of revisions. Note that svnadmin dump is reading revision trees from the repository just like any other reader process would (e.g., svn checkout), so it's safe to run this command at any time.

The other subcommand in the pair, svnadmin load, parses the standard input stream as a Subversion repository dump file and effectively replays those dumped revisions into the target repository for that operation. It also gives informative feedback, this time using the standard output stream:

$ svnadmin load newrepos < dumpfile
<<< Started new txn, based on original revision 1
     * adding path : A ... done.
     * adding path : A/B ... done.
     …
------- Committed new rev 1 (loaded from original rev 1) >>>

<<< Started new txn, based on original revision 2
     * editing path : A/mu ... done.
     * editing path : A/D/G/rho ... done.

------- Committed new rev 2 (loaded from original rev 2) >>>

…

<<< Started new txn, based on original revision 25
     * editing path : A/D/gamma ... done.

------- Committed new rev 25 (loaded from original rev 25) >>>

<<< Started new txn, based on original revision 26
     * adding path : A/Z/zeta ... done.
     * editing path : A/mu ... done.

------- Committed new rev 26 (loaded from original rev 26) >>>

The result of a load is new revisions added to a repository—the same thing you get by making commits against that repository from a regular Subversion client. Just as in a commit, you can use hook programs to perform actions before and after each of the commits made during a load process. By passing the --use-pre-commit-hook and --use-post-commit-hook options to svnadmin load, you can instruct Subversion to execute the pre-commit and post-commit hook programs, respectively, for each loaded revision. You might use these, for example, to ensure that loaded revisions pass through the same validation steps that regular commits pass through. Of course, you should use these options with care—if your post-commit hook sends emails to a mailing list for each new commit, you might not want to spew hundreds or thousands of commit emails in rapid succession at that list! You can read more about the use of hook scripts in the section called “Implementing Repository Hooks”.

Note that because svnadmin uses standard input and output streams for the repository dump and load processes, people who are feeling especially saucy can try things such as this (perhaps even using different versions of svnadmin on each side of the pipe):

$ svnadmin create newrepos
$ svnadmin dump oldrepos | svnadmin load newrepos

By default, the dump file will be quite large—much larger than the repository itself. That's because by default every version of every file is expressed as a full text in the dump file. This is the fastest and simplest behavior, and it's nice if you're piping the dump data directly into some other process (such as a compression program, filtering program, or loading process). But if you're creating a dump file for longer-term storage, you'll likely want to save disk space by using the --deltas option. With this option, successive revisions of files will be output as compressed, binary differences—just as file revisions are stored in a repository. This option is slower, but it results in a dump file much closer in size to the original repository.

We mentioned previously that svnadmin dump outputs a range of revisions. Use the --revision (-r) option to specify a single revision, or a range of revisions, to dump. If you omit this option, all the existing repository revisions will be dumped.

$ svnadmin dump myrepos -r 23 > rev-23.dumpfile
$ svnadmin dump myrepos -r 100:200 > revs-100-200.dumpfile

As Subversion dumps each new revision, it outputs only enough information to allow a future loader to re-create that revision based on the previous one. In other words, for any given revision in the dump file, only the items that were changed in that revision will appear in the dump. The only exception to this rule is the first revision that is dumped with the current svnadmin dump command.

By default, Subversion will not express the first dumped revision as merely differences to be applied to the previous revision. For one thing, there is no previous revision in the dump file! And second, Subversion cannot know the state of the repository into which the dump data will be loaded (if it ever is). To ensure that the output of each execution of svnadmin dump is self-sufficient, the first dumped revision is, by default, a full representation of every directory, file, and property in that revision of the repository.

However, you can change this default behavior. If you add the --incremental option when you dump your repository, svnadmin will compare the first dumped revision against the previous revision in the repository—the same way it treats every other revision that gets dumped. It will then output the first revision exactly as it does the rest of the revisions in the dump range—mentioning only the changes that occurred in that revision. The benefit of this is that you can create several small dump files that can be loaded in succession, instead of one large one, like so:

$ svnadmin dump myrepos -r 0:1000 > dumpfile1
$ svnadmin dump myrepos -r 1001:2000 --incremental > dumpfile2
$ svnadmin dump myrepos -r 2001:3000 --incremental > dumpfile3

These dump files could be loaded into a new repository with the following command sequence:

$ svnadmin load newrepos < dumpfile1
$ svnadmin load newrepos < dumpfile2
$ svnadmin load newrepos < dumpfile3

Another neat trick you can perform with this --incremental option involves appending to an existing dump file a new range of dumped revisions. For example, you might have a post-commit hook that simply appends the repository dump of the single revision that triggered the hook. Or you might have a script that runs nightly to append dump file data for all the revisions that were added to the repository since the last time the script ran. Used like this, svnadmin dump can be one way to back up changes to your repository over time in case of a system crash or some other catastrophic event.

The dump format can also be used to merge the contents of several different repositories into a single repository. By using the --parent-dir option of svnadmin load, you can specify a new virtual root directory for the load process. That means if you have dump files for three repositories—say calc-dumpfile, cal-dumpfile, and ss-dumpfile—you can first create a new repository to hold them all:

$ svnadmin create /var/svn/projects
$

Then, make new directories in the repository that will encapsulate the contents of each of the three previous repositories:

$ svn mkdir -m "Initial project roots" \
            file:///var/svn/projects/calc \
            file:///var/svn/projects/calendar \
            file:///var/svn/projects/spreadsheet
Committed revision 1.
$ 

Lastly, load the individual dump files into their respective locations in the new repository:

$ svnadmin load /var/svn/projects --parent-dir calc < calc-dumpfile
…
$ svnadmin load /var/svn/projects --parent-dir calendar < cal-dumpfile
…
$ svnadmin load /var/svn/projects --parent-dir spreadsheet < ss-dumpfile
…
$

Repository data migration using svnrdump

In Subversion 1.7, svnrdump joined the set of stock Subversion tools. It offers fairly specialized functionality, essentially as a network-aware version of the svnadmin dump and svnadmin load commands which we discuss in depth in the section called “Repository data migration using svnadmin”. svnrdump dump will generate a dump stream from a remote repository, spewing it to standard output; svnrdump load will read a dump stream from standard input and load it into a remote repository. Using svnrdump, you can generate incremental dumps just as you might with svnadmin dump. You can even dump a subtree of the repository—something that svnadmin dump cannot do.

The primary difference is that instead of requiring direct access to the repository, svnrdump operates remotely, using the very same Repository Access (RA) protocols that the Subversion client does. As such, you might need to provide authentication credentials. Also, your remote interations are subject to any authorization limitations configured on the Subversion server.

[Note]Note

svnrdump dump requires that the remote server be running Subversion 1.4 or newer. It currently generates dump streams only of the sort which are created when you pass the --deltas option to svnadmin dump. This isn't interesting in the typical use-cases, but might impact specific types of custom transformations you might wish to apply to the resulting dump stream.

[Note]Note

Because it modifies revision properties after committing new revisions, svnrdump load requires that the target repository have revision property changes enabled via the pre-revprop-change hook. See pre-revprop-change in Chapter 9, Subversion Complete Reference for details.

As you might expect, you can use svnadmin and svnrdump in concert. You can, for example, use svnrdump dump to generate a dump stream from a remote repository, and pipe the results thereof through svnadmin load to copy all that repository history into a local repository. Or you can do the reverse, copying history from a local repository into a remote one.

[Tip]Tip

By using file:// URLs, svnrdump can also access local repositories, but it will be doing so via Subversion's Repository Access (RA) abstraction layer—you'll get better performance out of svnadmin in such situations.

Filtering Repository History

Since Subversion stores your versioned history using, at the very least, binary differencing algorithms and data compression (optionally in a completely opaque database system), attempting manual tweaks is unwise if not quite difficult, and at any rate strongly discouraged. And once data has been stored in your repository, Subversion generally doesn't provide an easy way to remove that data.[45] But inevitably, there will be times when you would like to manipulate the history of your repository. You might need to strip out all instances of a file that was accidentally added to the repository (and shouldn't be there for whatever reason).[46] Or, perhaps you have multiple projects sharing a single repository, and you decide to split them up into their own repositories. To accomplish tasks such as these, administrators need a more manageable and malleable representation of the data in their repositories—the Subversion repository dump format.

As we described earlier in the section called “Migrating Repository Data Elsewhere”, the Subversion repository dump format is a human-readable representation of the changes that you've made to your versioned data over time. Use the svnadmin dump or svnrdump dump command to generate the dump data, and svnadmin load or svnrdump load to populate a new repository with it. The great thing about the human-readability aspect of the dump format is that, if you aren't careless about it, you can manually inspect and modify it. Of course, the downside is that if you have three years' worth of repository activity encapsulated in what is likely to be a very large dump file, it could take you a long, long time to manually inspect and modify it.

That's where svndumpfilter becomes useful. This program acts as a path-based filter for repository dump streams. Simply give it either a list of paths you wish to keep or a list of paths you wish to not keep, and then pipe your repository dump data through this filter. The result will be a modified stream of dump data that contains only the versioned paths you (explicitly or implicitly) requested.

Let's look at a realistic example of how you might use this program. Earlier in this chapter (see the section called “Planning Your Repository Organization”), we discussed the process of deciding how to choose a layout for the data in your repositories—using one repository per project or combining them, arranging stuff within your repository, and so on. But sometimes after new revisions start flying in, you rethink your layout and would like to make some changes. A common change is the decision to move multiple projects that are sharing a single repository into separate repositories for each project.

Our imaginary repository contains three projects: calc, calendar, and spreadsheet. They have been living side-by-side in a layout like this:



/

   calc/

      trunk/

      branches/

      tags/

   calendar/

      trunk/

      branches/

      tags/

   spreadsheet/

      trunk/

      branches/

      tags/

To get these three projects into their own repositories, we first dump the whole repository:

$ svnadmin dump /var/svn/repos > repos-dumpfile
* Dumped revision 0.
* Dumped revision 1.
* Dumped revision 2.
* Dumped revision 3.
…
$

Next, run that dump file through the filter, each time including only one of our top-level directories. This results in three new dump files:

$ svndumpfilter include calc < repos-dumpfile > calc-dumpfile
…
$ svndumpfilter include calendar < repos-dumpfile > cal-dumpfile
…
$ svndumpfilter include spreadsheet < repos-dumpfile > ss-dumpfile
…
$

At this point, you have to make a decision. Each of your dump files will create a valid repository, but will preserve the paths exactly as they were in the original repository. This means that even though you would have a repository solely for your calc project, that repository would still have a top-level directory named calc. If you want your trunk, tags, and branches directories to live in the root of your repository, you might wish to edit your dump files, tweaking the Node-path and Node-copyfrom-path headers so that they no longer have that first calc/ path component. Also, you'll want to remove the section of dump data that creates the calc directory. It will look something like the following:

Node-path: calc
Node-action: add
Node-kind: dir
Content-length: 0
  
[Warning]Warning

If you do plan on manually editing the dump file to remove a top-level directory, make sure your editor is not set to automatically convert end-of-line characters to the native format (e.g., \r\n to \n), as the content will then not agree with the metadata. This will render the dump file useless.

All that remains now is to create your three new repositories, and load each dump file into the right repository, ignoring the UUID found in the dump stream:

$ svnadmin create calc
$ svnadmin load --ignore-uuid calc < calc-dumpfile
<<< Started new transaction, based on original revision 1
     * adding path : Makefile ... done.
     * adding path : button.c ... done.
…
$ svnadmin create calendar
$ svnadmin load --ignore-uuid calendar < cal-dumpfile
<<< Started new transaction, based on original revision 1
     * adding path : Makefile ... done.
     * adding path : cal.c ... done.
…
$ svnadmin create spreadsheet
$ svnadmin load --ignore-uuid spreadsheet < ss-dumpfile
<<< Started new transaction, based on original revision 1
     * adding path : Makefile ... done.
     * adding path : ss.c ... done.
…
$

Both of svndumpfilter's subcommands accept options for deciding how to deal with empty revisions. If a given revision contains only changes to paths that were filtered out, that now-empty revision could be considered uninteresting or even unwanted. So to give the user control over what to do with those revisions, svndumpfilter provides the following command-line options:

--drop-empty-revs

Do not generate empty revisions at all—just omit them.

--renumber-revs

If empty revisions are dropped (using the --drop-empty-revs option), change the revision numbers of the remaining revisions so that there are no gaps in the numeric sequence.

--preserve-revprops

If empty revisions are not dropped, preserve the revision properties (log message, author, date, custom properties, etc.) for those empty revisions. Otherwise, empty revisions will contain only the original datestamp, and a generated log message that indicates that this revision was emptied by svndumpfilter.

While svndumpfilter can be very useful and a huge timesaver, there are unfortunately a couple of gotchas. First, this utility is overly sensitive to path semantics. Pay attention to whether paths in your dump file are specified with or without leading slashes. You'll want to look at the Node-path and Node-copyfrom-path headers.

…
Node-path: spreadsheet/Makefile
…

If the paths have leading slashes, you should include leading slashes in the paths you pass to svndumpfilter include and svndumpfilter exclude (and if they don't, you shouldn't). Further, if your dump file has an inconsistent usage of leading slashes for some reason,[47] you should probably normalize those paths so that they all have, or all lack, leading slashes.

Also, copied paths can give you some trouble. Subversion supports copy operations in the repository, where a new path is created by copying some already existing path. It is possible that at some point in the lifetime of your repository, you might have copied a file or directory from some location that svndumpfilter is excluding, to a location that it is including. To make the dump data self-sufficient, svndumpfilter needs to still show the addition of the new path—including the contents of any files created by the copy—and not represent that addition as a copy from a source that won't exist in your filtered dump data stream. But because the Subversion repository dump format shows only what was changed in each revision, the contents of the copy source might not be readily available. If you suspect that you have any copies of this sort in your repository, you might want to rethink your set of included/excluded paths, perhaps including the paths that served as sources of your troublesome copy operations, too.

Finally, svndumpfilter takes path filtering quite literally. If you are trying to copy the history of a project rooted at trunk/my-project and move it into a repository of its own, you would, of course, use the svndumpfilter include command to keep all the changes in and under trunk/my-project. But the resultant dump file makes no assumptions about the repository into which you plan to load this data. Specifically, the dump data might begin with the revision that added the trunk/my-project directory, but it will not contain directives that would create the trunk directory itself (because trunk doesn't match the include filter). You'll need to make sure that any directories that the new dump stream expects to exist actually do exist in the target repository before trying to load the stream into that repository.

Repository Replication

There are several scenarios in which it is quite handy to have a Subversion repository whose version history is exactly the same as some other repository's. Perhaps the most obvious one is the maintenance of a simple backup repository, used when the primary repository has become inaccessible due to a hardware failure, network outage, or other such annoyance. Other scenarios include deploying mirror repositories to distribute heavy Subversion load across multiple servers, use as a soft-upgrade mechanism, and so on.

Subversion provides a program for managing scenarios such as these. svnsync works by essentially asking the Subversion server to replay revisions, one at a time. It then uses that revision information to mimic a commit of the same to another repository. Neither repository needs to be locally accessible to the machine on which svnsync is running—its parameters are repository URLs, and it does all its work through Subversion's Repository Access (RA) interfaces. All it requires is read access to the source repository and read/write access to the destination repository.

[Note]Note

When using svnsync against a remote source repository, the Subversion server for that repository must be running Subversion version 1.4 or later.

Replication with svnsync

Assuming you already have a source repository that you'd like to mirror, the next thing you need is a target repository that will actually serve as that mirror. This target repository can use either of the available filesystem data-store backends (see the section called “Choosing a Data Store”)—Subversion's abstraction layers ensure that such details don't matter. But by default, it must not yet have any version history in it. (We'll discuss an exception to this later in this section.)

The protocol that svnsync uses to communicate revision information is highly sensitive to mismatches between the versioned histories contained in the source and target repositories. For this reason, while svnsync cannot demand that the target repository be read-only,[48] allowing the revision history in the target repository to change by any mechanism other than the mirroring process is a recipe for disaster.

[Warning]Warning

Do not modify a mirror repository in such a way as to cause its version history to deviate from that of the repository it mirrors. The only commits and revision property modifications that ever occur on that mirror repository should be those performed by the svnsync tool.

Another requirement of the target repository is that the svnsync process be allowed to modify revision properties. Because svnsync works within the framework of that repository's hook system, the default state of the repository (which is to disallow revision property changes; see pre-revprop-change in Chapter 9, Subversion Complete Reference) is insufficient. You'll need to explicitly implement the pre-revprop-change hook, and your script must allow svnsync to set and change revision properties. With those provisions in place, you are ready to start mirroring repository revisions.

[Tip]Tip

It's a good idea to implement authorization measures that allow your repository replication process to perform its tasks while preventing other users from modifying the contents of your mirror repository at all.

Let's walk through the use of svnsync in a somewhat typical mirroring scenario. We'll pepper this discourse with practical recommendations, which you are free to disregard if they aren't required by or suitable for your environment.

We will be mirroring the public Subversion repository which houses the source code for this very book and exposing that mirror publicly on the Internet, hosted on a different machine than the one on which the original Subversion source code repository lives. This remote host has a global configuration that permits anonymous users to read the contents of repositories on the host, but requires users to authenticate to modify those repositories. (Please forgive us for glossing over the details of Subversion server configuration for the moment—those are covered thoroughly in Chapter 6, Server Configuration.) And for no other reason than that it makes for a more interesting example, we'll be driving the replication process from a third machine—the one that we currently find ourselves using.

First, we'll create the repository which will be our mirror. This and the next couple of steps do require shell access to the machine on which the mirror repository will live. Once the repository is all configured, though, we shouldn't need to touch it directly again.

$ ssh admin@svn.example.com "svnadmin create /var/svn/svn-mirror"
admin@svn.example.com's password: ********
$

At this point, we have our repository, and due to our server's configuration, that repository is now live on the Internet. Now, because we don't want anything modifying the repository except our replication process, we need a way to distinguish that process from other would-be committers. To do so, we use a dedicated username for our process. Only commits and revision property modifications performed by the special username syncuser will be allowed.

We'll use the repository's hook system both to allow the replication process to do what it needs to do and to enforce that only it is doing those things. We accomplish this by implementing two of the repository event hooks—pre-revprop-change and start-commit. Our pre-revprop-change hook script is found in Example 5.2, “Mirror repository's pre-revprop-change hook script”, and basically verifies that the user attempting the property changes is our syncuser user. If so, the change is allowed; otherwise, it is denied.

Example 5.2. Mirror repository's pre-revprop-change hook script

#!/bin/sh 

USER="$3"

if [ "$USER" = "syncuser" ]; then exit 0; fi

echo "Only the syncuser user may change revision properties" >&2
exit 1


That covers revision property changes. Now we need to ensure that only the syncuser user is permitted to commit new revisions to the repository. We do this using a start-commit hook script such as the one in Example 5.3, “Mirror repository's start-commit hook script”.

Example 5.3. Mirror repository's start-commit hook script

#!/bin/sh 

USER="$2"

if [ "$USER" = "syncuser" ]; then exit 0; fi

echo "Only the syncuser user may commit new revisions" >&2
exit 1


After installing our hook scripts and ensuring that they are executable by the Subversion server, we're finished with the setup of the mirror repository. Now, we get to actually do the mirroring.

The first thing we need to do with svnsync is to register in our target repository the fact that it will be a mirror of the source repository. We do this using the svnsync initialize subcommand. The URLs we provide point to the root directories of the target and source repositories, respectively. In Subversion 1.4, this is required—only full mirroring of repositories is permitted. Beginning with Subversion 1.5, though, you can use svnsync to mirror only some subtree of the repository, too.

$ svnsync help init
initialize (init): usage: svnsync initialize DEST_URL SOURCE_URL

Initialize a destination repository for synchronization from
another repository.
…
$ svnsync initialize http://svn.example.com/svn-mirror \
                     http://svnbook.googlecode.com/svn \
                     --sync-username syncuser --sync-password syncpass
Copied properties for revision 0 (svn:sync-* properties skipped).
NOTE: Normalized svn:* properties to LF line endings (1 rev-props, 0 node-props).
$

Our target repository will now remember that it is a mirror of the public Subversion source code repository. Notice that we provided a username and password as arguments to svnsync—that was required by the pre-revprop-change hook on our mirror repository.

[Note]Note

In Subversion 1.4, the values given to svnsync's --username and --password command-line options were used for authentication against both the source and destination repositories. This caused problems when a user's credentials weren't exactly the same for both repositories, especially when running in noninteractive mode (with the --non-interactive option). This was fixed in Subversion 1.5 with the introduction of two new pairs of options. Use --source-username and --source-password to provide authentication credentials for the source repository; use --sync-username and --sync-password to provide credentials for the destination repository. (The old --username and --password options still exist for compatibility, but we advise against using them.)

And now comes the fun part. With a single subcommand, we can tell svnsync to copy all the as-yet-unmirrored revisions from the source repository to the target.[49] The svnsync synchronize subcommand will peek into the special revision properties previously stored on the target repository and determine how much of the source repository has been previously mirrored—in this case, the most recently mirrored revision is r0. Then it will query the source repository and determine what the latest revision in that repository is. Finally, it asks the source repository's server to start replaying all the revisions between 0 and that latest revision. As svnsync gets the resultant response from the source repository's server, it begins forwarding those revisions to the target repository's server as new commits.

$ svnsync help synchronize
synchronize (sync): usage: svnsync synchronize DEST_URL [SOURCE_URL]

Transfer all pending revisions to the destination from the source
with which it was initialized.
…
$ svnsync synchronize http://svn.example.com/svn-mirror \
                      http://svnbook.googlecode.com/svn
Committed revision 1.
Copied properties for revision 1.
Committed revision 2.
Copied properties for revision 2.
Transmitting file data .
Committed revision 3.
Copied properties for revision 3.
…
Transmitting file data .
Committed revision 4063.
Copied properties for revision 4063.
Transmitting file data .
Committed revision 4064.
Copied properties for revision 4064.
Transmitting file data ....
Committed revision 4065.
Copied properties for revision 4065.
$

Of particular interest here is that for each mirrored revision, there is first a commit of that revision to the target repository, and then property changes follow. This two-phase replication is required because the initial commit is performed by (and attributed to) the user syncuser and is datestamped with the time as of that revision's creation. svnsync has to follow up with an immediate series of property modifications that copy into the target repository all the original revision properties found for that revision in the source repository, which also has the effect of fixing the author and datestamp of the revision to match that of the source repository.

Also noteworthy is that svnsync performs careful bookkeeping that allows it to be safely interrupted and restarted without ruining the integrity of the mirrored data. If a network glitch occurs while mirroring a repository, simply repeat the svnsync synchronize command, and it will happily pick up right where it left off. In fact, as new revisions appear in the source repository, this is exactly what you do to keep your mirror up to date.

[Warning]Warning

As part of its bookkeeping, svnsync records in the mirror repository the URL with which the mirror was initialized. Because of this, invocations of svnsync which follow the initialization step do not require that you provide the source URL on the command line again. However, for security purposes, we recommend that you continue to do so. Depending on how it is deployed, it may not be safe for svnsync to trust the source URL which it retrieves from the mirror repository, and from which it pulls versioned data.

There is, however, one bit of inelegance in the process. Because Subversion revision properties can be changed at any time throughout the lifetime of the repository, and because they don't leave an audit trail that indicates when they were changed, replication processes have to pay special attention to them. If you've already mirrored the first 15 revisions of a repository and someone then changes a revision property on revision 12, svnsync won't know to go back and patch up its copy of revision 12. You'll need to tell it to do so manually by using (or with some additional tooling around) the svnsync copy-revprops subcommand, which simply rereplicates all the revision properties for a particular revision or range thereof.

$ svnsync help copy-revprops
copy-revprops: usage:

    1. svnsync copy-revprops DEST_URL [SOURCE_URL]
    2. svnsync copy-revprops DEST_URL REV[:REV2]

…
$ svnsync copy-revprops http://svn.example.com/svn-mirror 12
Copied properties for revision 12.
$

That's repository replication via svnsync in a nutshell. You'll likely want some automation around such a process. For example, while our example was a pull-and-push setup, you might wish to have your primary repository push changes to one or more blessed mirrors as part of its post-commit and post-revprop-change hook implementations. This would enable the mirror to be up to date in as near to real time as is likely possible.

Partial replication with svnsync

svnsync isn't limited to full copies of everything which lives in a repository. It can handle various shades of partial replication, too. For example, while it isn't very commonplace to do so, svnsync does gracefully mirror repositories in which the user as whom it authenticates has only partial read access. It simply copies only the bits of the repository that it is permitted to see. Obviously, such a mirror is not useful as a backup solution.

As of Subversion 1.5, svnsync also has the ability to mirror a subset of a repository rather than the whole thing. The process of setting up and maintaining such a mirror is exactly the same as when mirroring a whole repository, except that instead of specifying the source repository's root URL when running svnsync init, you specify the URL of some subdirectory within that repository. Synchronization to that mirror will now copy only the bits that changed under that source repository subdirectory. There are some limitations to this support, though. First, you can't mirror multiple disjoint subdirectories of the source repository into a single mirror repository—you'd need to instead mirror some parent directory that is common to both. Second, the filtering logic is entirely path-based, so if the subdirectory you are mirroring was renamed at some point in the past, your mirror would contain only the revisions since the directory appeared at the URL you specified. And likewise, if the source subdirectory is renamed in the future, your synchronization processes will stop mirroring data at the point that the source URL you specified is no longer valid.

A quick trick for mirror creation

We mentioned previously the cost of setting up an initial mirror of an existing repository. For many folks, the sheer cost of transmitting thousands—or millions—of revisions of history to a new mirror repository via svnsync is a show-stopper. Fortunately, Subversion 1.7 provides a workaround by way of a new --allow-non-empty option to svnsync initialize. This option allows you to initialize one repository as a mirror of another while bypassing the verification that the to-be-initialized mirror has no version history present in it. Per our previous warnings about the sensitivity of this whole replication process, you should rightly discern that this is an option to be used only with great caution. But it's wonderfully handy when you have administrative access to the source repository, where you can simply make a physical copy of the repository and then initialize that copy as a new mirror:

$ svnadmin hotcopy /path/to/repos /path/to/mirror-repos
$ ### create /path/to/mirror-repos/hooks/pre-revprop-change
$ svnsync initialize file:///path/to/mirror-repos \
                     file:///path/to/repos
svnsync: E000022: Destination repository already contains revision history; co
nsider using --allow-non-empty if the repository's revisions are known to mirr
or their respective revisions in the source repository
$ svnsync initialize --allow-non-empty file:///path/to/mirror-repos \
                                       file:///path/to/repos
Copied properties for revision 32042.
$

Admins who are running a version of Subversion prior to 1.7 (and thus do not have access to svnsync initialize's --allow-non-empty feature) can accomplish effectively the same thing that that feature does through careful manipulation of the r0 revision properties on the copy of the repository which is slated to become a mirror of the original. Use svnadmin setrevprop to create the same bookkeeping properties that svnsync would have created there.

Replication wrap-up

We've discussed a couple of ways to replicate revision history from one repository to another. So let's look now at the user end of these operations. How does replication and the various situations which call for it affect Subversion clients?

As far as user interaction with repositories and mirrors goes, it is possible to have a single working copy that interacts with both, but you'll have to jump through some hoops to make it happen. First, you need to ensure that both the primary and mirror repositories have the same repository UUID (which is not the case by default). See the section called “Managing Repository UUIDs” later in this chapter for more about this.

Once the two repositories have the same UUID, you can use svn relocate to point your working copy to whichever of the repositories you wish to operate against, a process that is described in svn relocate in Chapter 9, Subversion Complete Reference. There is a possible danger here, though, in that if the primary and mirror repositories aren't in close synchronization, a working copy up to date with, and pointing to, the primary repository will, if relocated to point to an out-of-date mirror, become confused about the apparent sudden loss of revisions it fully expects to be present, and it will throw errors to that effect. If this occurs, you can relocate your working copy back to the primary repository and then either wait until the mirror repository is up to date, or backdate your working copy to a revision you know is present in the sync repository, and then retry the relocation.

Finally, be aware that the revision-based replication provided by svnsync is only that—replication of revisions. Only the kinds of information carried by the Subversion repository dump file format are available for replication. As such, tools such as svnsync (and svnrdump, which we discuss in the section called “Repository data migration using svnrdump”) are limited in ways similar to that of the repository dump stream. They do not include in their replicated information such things as the hook implementations, repository or server configuration data, uncommitted transactions, or information about user locks on repository paths.

Repository Backup

Despite numerous advances in technology since the birth of the modern computer, one thing unfortunately rings true with crystalline clarity—sometimes things go very, very awry. Power outages, network connectivity dropouts, corrupt RAM, and crashed hard drives are but a taste of the evil that Fate is poised to unleash on even the most conscientious administrator. And so we arrive at a very important topic—how to make backup copies of your repository data.

There are two types of backup methods available for Subversion repository administrators—full and incremental. A full backup of the repository involves squirreling away in one sweeping action all the information required to fully reconstruct that repository in the event of a catastrophe. Usually, it means, quite literally, the duplication of the entire repository directory (which includes either a Berkeley DB or FSFS environment). Incremental backups are lesser things: backups of only the portion of the repository data that has changed since the previous backup.

As far as full backups go, the naïve approach might seem like a sane one, but unless you temporarily disable all other access to your repository, simply doing a recursive directory copy runs the risk of generating a faulty backup. In the case of Berkeley DB, the documentation describes a certain order in which database files can be copied that will guarantee a valid backup copy. A similar ordering exists for FSFS data. But you don't have to implement these algorithms yourself, because the Subversion development team has already done so. The svnadmin hotcopy command takes care of the minutia involved in making a hot backup of your repository. And its invocation is as trivial as the Unix cp or Windows copy operations:

$ svnadmin hotcopy /var/svn/repos /var/svn/repos-backup

The resultant backup is a fully functional Subversion repository, able to be dropped in as a replacement for your live repository should something go horribly wrong.

When making copies of a Berkeley DB repository, you can even instruct svnadmin hotcopy to purge any unused Berkeley DB logfiles (see the section called “Purging unused Berkeley DB logfiles”) from the original repository upon completion of the copy. Simply provide the --clean-logs option on the command line.

$ svnadmin hotcopy --clean-logs /var/svn/bdb-repos /var/svn/bdb-repos-backup

Additional tooling around this command is available, too. The tools/backup/ directory of the Subversion source distribution holds the hot-backup.py script. This script adds a bit of backup management atop svnadmin hotcopy, allowing you to keep only the most recent configured number of backups of each repository. It will automatically manage the names of the backed-up repository directories to avoid collisions with previous backups and will rotate off older backups, deleting them so that only the most recent ones remain. Even if you also have an incremental backup, you might want to run this program on a regular basis. For example, you might consider using hot-backup.py from a program scheduler (such as cron on Unix systems), which can cause it to run nightly (or at whatever granularity of time you deem safe).

Some administrators use a different backup mechanism built around generating and storing repository dump data. We described in the section called “Migrating Repository Data Elsewhere” how to use svnadmin dump with the --incremental option to perform an incremental backup of a given revision or range of revisions. And of course, you can achieve a full backup variation of this by omitting the --incremental option to that command. There is some value in these methods, in that the format of your backed-up information is flexible—it's not tied to a particular platform, versioned filesystem type, or release of Subversion or Berkeley DB. But that flexibility comes at a cost, namely that restoring that data can take a long time—longer with each new revision committed to your repository. Also, as is the case with so many of the various backup methods, revision property changes that are made to already backed-up revisions won't get picked up by a nonoverlapping, incremental dump generation. For these reasons, we recommend against relying solely on dump-based backup approaches.

As you can see, each of the various backup types and methods has its advantages and disadvantages. The easiest is by far the full hot backup, which will always result in a perfect working replica of your repository. Should something bad happen to your live repository, you can restore from the backup with a simple recursive directory copy. Unfortunately, if you are maintaining multiple backups of your repository, these full copies will each eat up just as much disk space as your live repository. Incremental backups, by contrast, tend to be quicker to generate and smaller to store. But the restoration process can be a pain, often involving applying multiple incremental backups. And other methods have their own peculiarities. Administrators need to find the balance between the cost of making the backup and the cost of restoring it.

The svnsync program (see the section called “Repository Replication”) actually provides a rather handy middle-ground approach. If you are regularly synchronizing a read-only mirror with your main repository, in a pinch your read-only mirror is probably a good candidate for replacing that main repository if it falls over. The primary disadvantage of this method is that only the versioned repository data gets synchronized—repository configuration files, user-specified repository path locks, and other items that might live in the physical repository directory but not inside the repository's virtual versioned filesystem are not handled by svnsync.

In any backup scenario, repository administrators need to be aware of how modifications to unversioned revision properties affect their backups. Since these changes do not themselves generate new revisions, they will not trigger post-commit hooks, and may not even trigger the pre-revprop-change and post-revprop-change hooks.[50] And since you can change revision properties without respect to chronological order—you can change any revision's properties at any time—an incremental backup of the latest few revisions might not catch a property modification to a revision that was included as part of a previous backup.

Generally speaking, only the truly paranoid would need to back up their entire repository, say, every time a commit occurred. However, assuming that a given repository has some other redundancy mechanism in place with relatively fine granularity (such as per-commit emails or incremental dumps), a hot backup of the database might be something that a repository administrator would want to include as part of a system-wide nightly backup. It's your data—protect it as much as you'd like.

Often, the best approach to repository backups is a diversified one that leverages combinations of the methods described here. The Subversion developers, for example, back up the Subversion source code repository nightly using hot-backup.py and an off-site rsync of those full backups; keep multiple archives of all the commit and property change notification emails; and have repository mirrors maintained by various volunteers using svnsync. Your solution might be similar, but should be catered to your needs and that delicate balance of convenience with paranoia. And whatever you do, validate your backups from time to time—what good is a spare tire that has a hole in it? While all of this might not save your hardware from the iron fist of Fate,[51] it should certainly help you recover from those trying times.

Managing Repository UUIDs

Subversion repositories have a universally unique identifier (UUID) associated with them. This is used by Subversion clients to verify the identity of a repository when other forms of verification aren't good enough (such as checking the repository URL, which can change over time). Most Subversion repository administrators rarely, if ever, need to think about repository UUIDs as anything more than a trivial implementation detail of Subversion. Sometimes, however, there is cause for attention to this detail.

As a general rule, you want the UUIDs of your live repositories to be unique. That is, after all, the point of having UUIDs. But there are times when you want the repository UUIDs of two repositories to be exactly the same. For example, if you make a copy of a repository for backup purposes, you want the backup to be a perfect replica of the original so that, in the event that you have to restore that backup and replace the live repository, users don't suddenly see what looks like a different repository. When dumping and loading repository history (as described earlier in the section called “Migrating Repository Data Elsewhere”), you get to decide whether to apply the UUID encapsulated in the data dump stream to the repository in which you are loading the data. The particular circumstance will dictate the correct behavior.

There are a couple of ways to set (or reset) a repository's UUID, should you need to. As of Subversion 1.5, this is as simple as using the svnadmin setuuid command. If you provide this subcommand with an explicit UUID, it will validate that the UUID is well-formed and then set the repository UUID to that value. If you omit the UUID, a brand-new UUID will be generated for your repository.

$ svnlook uuid /var/svn/repos
cf2b9d22-acb5-11dc-bc8c-05e83ce5dbec
$ svnadmin setuuid /var/svn/repos   # generate a new UUID
$ svnlook uuid /var/svn/repos
3c3c38fe-acc0-11dc-acbc-1b37ff1c8e7c
$ svnadmin setuuid /var/svn/repos \
           cf2b9d22-acb5-11dc-bc8c-05e83ce5dbec  # restore the old UUID
$ svnlook uuid /var/svn/repos
cf2b9d22-acb5-11dc-bc8c-05e83ce5dbec
$

For folks using versions of Subversion earlier than 1.5, these tasks are a little more complicated. You can explicitly set a repository's UUID by piping a repository dump file stub that carries the new UUID specification through svnadmin load --force-uuid REPOS-PATH.

$ svnadmin load --force-uuid /var/svn/repos <<EOF
SVN-fs-dump-format-version: 2

UUID: cf2b9d22-acb5-11dc-bc8c-05e83ce5dbec
EOF
$ svnlook uuid /var/svn/repos
cf2b9d22-acb5-11dc-bc8c-05e83ce5dbec
$

Having older versions of Subversion generate a brand-new UUID is not quite as simple to do, though. Your best bet here is to find some other way to generate a UUID, and then explicitly set the repository's UUID to that value.

Moving and Removing Repositories

Subversion repository data is wholly contained within the repository directory. As such, you can move a Subversion repository to some other location on disk, rename a repository, copy a repository, or delete a repository altogether using the tools provided by your operating system for manipulating directories—mv, cp -a, and rm -r on Unix platforms; copy, move, and rmdir /s /q on Windows; vast numbers of mouse and menu gyrations in various graphical file explorer applications, and so on.

Of course, there's often still more to be done when trying to cleanly affect changes such as this. For example, you might need to update your Subversion server configuration to point to the new location of a relocated repository or to remove configuration bits for a now-deleted repository. If you have automated processes that publish information from or about your repositories, they may need to be updated. Hook scripts might need to be reconfigured. Users may need to be notified. The list can go on indefinitely, or at least to the extent that you've built processes and procedures around your Subversion repository.

In the case of a copied repository, you should also consider the fact that Subversion uses repository UUIDs to distinguish repositories. If you copy a Subversion repository using a typical shell recursive copy command, you'll wind up with two repositories that are identical in every way—including their UUIDs. In some circumstances, this might be desirable. But in the instances where it is not, you'll need to generate a new UUID for one of these identical repositories. See the section called “Managing Repository UUIDs” for more about managing repository UUIDs.

Summary

By now you should have a basic understanding of how to create, configure, and maintain Subversion repositories. We introduced you to the various tools that will assist you with this task. Throughout the chapter, we noted common administration pitfalls and offered suggestions for avoiding them.

All that remains is for you to decide what exciting data to store in your repository, and finally, how to make it available over a network. The next chapter is all about networking.




[37] This may sound really prestigious and lofty, but we're just talking about anyone who is interested in that mysterious realm beyond the working copy where everyone's data hangs out.

[38] Strictly speaking, Subversion doesn't dictate that the versioned data live here, and there are known (albeit proprietary) alternative backend storage implementations which do not, in fact, store data in this directory.

[39] Whether founded in ignorance or in poorly considered concepts about how to derive legitimate software development metrics, global revision numbers are a silly thing to fear, and not the kind of thing you should weigh when deciding how to arrange your projects and repositories.

[40] The trunk, tags, and branches trio is sometimes referred to as the TTB directories.

[41] Often pronounced fuzz-fuzz, if Jack Repenning has anything to say about it. (This book, however, assumes that the reader is thinking eff-ess-eff-ess.)

[42] Berkeley DB requires that the underlying filesystem implement strict POSIX locking semantics, and more importantly, the ability to map files directly into process memory.

[43] Or is that, the sync?

[44] For example, hard drive + huge electromagnet = disaster.

[45] That's rather the reason you use version control at all, right?

[46] Conscious, cautious removal of certain bits of versioned data is actually supported by real use cases. That's why an obliterate feature has been one of the most highly requested Subversion features, and one which the Subversion developers hope to soon provide.

[47] While svnadmin dump has a consistent leading slash policy (to not include them), other programs that generate dump data might not be so consistent.

[48] In fact, it can't truly be read-only, or svnsync itself would have a tough time copying revision history into it.

[49] Be forewarned that while it will take only a few seconds for the average reader to parse this paragraph and the sample output that follows it, the actual time required to complete such a mirroring operation is, shall we say, quite a bit longer.

[50] svnadmin setlog can be called in a way that bypasses the hook interface altogether.

[51] You know—the collective term for all of her fickle fingers.

Chapter 6. Server Configuration

A Subversion repository can be accessed simultaneously by clients running on the same machine on which the repository resides using URLs carrying the file:// scheme. But the typical Subversion setup involves a single server machine being accessed from clients on computers all over the office—or, perhaps, all over the world.

This chapter describes how to get your Subversion repository exposed outside its host machine for use by remote clients. We will cover Subversion's currently available server mechanisms, discussing the configuration and use of each. After reading this chapter, you should be able to decide which networking setup is right for your needs, as well as understand how to enable such a setup on your host computer.

Overview

Subversion was designed with an abstract repository access layer. This means that a repository can be programmatically accessed by any sort of server process, and the client repository access API allows programmers to write plug-ins that speak relevant network protocols. In theory, Subversion can use an infinite number of network implementations. In practice, there are only two Subversion servers in widespread use today.

Apache is an extremely popular web server; using the mod_dav_svn module, Apache can access a repository and make it available to clients via the WebDAV/DeltaV protocol, which is an extension of HTTP. Because Apache is an extremely extensible server, it provides a number of features for free, such as encrypted SSL communication, logging, integration with a number of third-party authentication systems, and limited built-in web browsing of repositories.

In the other corner is svnserve: a small, lightweight server program that speaks a custom protocol with clients. Because its protocol is explicitly designed for Subversion and is stateful (unlike HTTP), it provides significantly faster network operations—but at the cost of some features as well. While it can use SASL to provide a variety of authentication and encryption options, it has no logging or built-in web browsing. It is, however, extremely easy to set up and is often the best option for small teams just starting out with Subversion.

The network protocol which svnserve speaks may also be tunneled over an SSH connection. This deployment option for svnserve differs quite a bit in features from a traditional svnserve deployment. SSH is used to encrypt all communication. SSH is also used exclusively to authenticate, so real system accounts are required on the server host (unlike vanilla svnserve, which has its own private user accounts). Finally, because this setup requires that each user spawn a private, temporary svnserve process, it's equivalent (from a permissions point of view) to allowing a group of local users to all access the repository via file:// URLs. Path-based access control has no meaning, since each user is accessing the repository database files directly.

Table 6.1, “Comparison of subversion server options” provides a quick summary of the three typical server deployments.

Table 6.1. Comparison of subversion server options

FeatureApache + mod_dav_svnsvnservesvnserve over SSH
Authentication optionsHTTP Basic or Digest auth, X.509 certificates, LDAP, NTLM, or any other mechanism available to Apache httpdCRAM-MD5 by default; LDAP, NTLM, or any other mechanism available to SASLSSH
User account optionsPrivate users file, or other mechanisms available to Apache httpd (LDAP, SQL, etc.)Private users file, or other mechanisms available to SASL (LDAP, SQL, etc.)System accounts
Authorization optionsRead/write access can be granted over the whole repository, or specified per pathRead/write access can be granted over the whole repository, or specified per pathRead/write access only grantable over the whole repository
EncryptionAvailable via optional SSL (https)Available via optional SASL featuresInherent in SSH connection
LoggingHigh-level operational logging of Subversion operations plus detailed logging at the per-HTTP-request levelHigh-level operational logging onlyHigh-level operational logging only
InteroperabilityAccessible by other WebDAV clientsTalks only to svn clientsTalks only to svn clients
Web viewingLimited built-in support, or via third-party tools such as ViewVCOnly via third-party tools such as ViewVCOnly via third-party tools such as ViewVC
Master-slave server replicationTransparent write-proxying available from slave to masterCan only create read-only slave serversCan only create read-only slave servers
SpeedSomewhat slowerSomewhat fasterSomewhat faster
Initial setupSomewhat complexExtremely simpleModerately simple


Choosing a Server Configuration

So, which server should you use? Which is best?

Obviously, there's no right answer to that question. Every team has different needs, and the different servers all represent different sets of trade-offs. The Subversion project itself doesn't endorse one server or another, or consider either server more official than another.

Here are some reasons why you might choose one deployment over another, as well as reasons you might not choose one.

The svnserve Server

Why you might want to use it:
  • Quick and easy to set up.

  • Network protocol is stateful and noticeably faster than WebDAV.

  • No need to create system accounts on server.

  • Password is not passed over the network.

Why you might want to avoid it:
  • By default, only one authentication method is available, the network protocol is not encrypted, and the server stores clear text passwords. (All these things can be changed by configuring SASL, but it's a bit more work to do.)

  • No advanced logging facilities.

  • No built-in web browsing. (You'd have to install a separate web server and repository browsing software to add this.)

svnserve over SSH

Why you might want to use it:
  • The network protocol is stateful and noticeably faster than WebDAV.

  • You can take advantage of existing SSH accounts and user infrastructure.

  • All network traffic is encrypted.

Why you might want to avoid it:
  • Only one choice of authentication method is available.

  • No advanced logging facilities.

  • It requires users to be in the same system group, or use a shared SSH key.

  • If used improperly, it can lead to file permission problems.

The Apache HTTP Server

Why you might want to use it:
  • It allows Subversion to use any of the numerous authentication systems already integrated with Apache.

  • There is no need to create system accounts on the server.

  • Full Apache logging is available.

  • Network traffic can be encrypted via SSL.

  • HTTP(S) can usually go through corporate firewalls.

  • Built-in repository browsing is available via web browser.

  • The repository can be mounted as a network drive for transparent version control (see the section called “Autoversioning”).

Why you might want to avoid it:
  • Noticeably slower than svnserve, because HTTP is a stateless protocol and requires more network turnarounds.

  • Initial setup can be complex.

Recommendations

In general, the authors of this book recommend a vanilla svnserve installation for small teams just trying to get started with a Subversion server; it's the simplest to set up and has the fewest maintenance issues. You can always switch to a more complex server deployment as your needs change.

Here are some general recommendations and tips, based on years of supporting users:

  • If you're trying to set up the simplest possible server for your group, a vanilla svnserve installation is the easiest, fastest route. Note, however, that your repository data will be transmitted in the clear over the network. If your deployment is entirely within your company's LAN or VPN, this isn't an issue. If the repository is exposed to the wide-open Internet, you might want to make sure that either the repository's contents aren't sensitive (e.g., it contains only open source code), or that you go the extra mile in configuring SASL to encrypt network communications.

  • If you need to integrate with existing legacy identity systems (LDAP, Active Directory, NTLM, X.509, etc.), you must use either the Apache-based server or svnserve configured with SASL.

  • If you've decided to use either Apache or stock svnserve, create a single svn user on your system and run the server process as that user. Be sure to make the repository directory wholly owned by the svn user as well. From a security point of view, this keeps the repository data nicely siloed and protected by operating system filesystem permissions, changeable by only the Subversion server process itself.

  • If you have an existing infrastructure that is heavily based on SSH accounts, and if your users already have system accounts on your server machine, it makes sense to deploy an svnserve-over-SSH solution. Otherwise, we don't widely recommend this option to the public. It's generally considered safer to have your users access the repository via (imaginary) accounts managed by svnserve or Apache, rather than by full-blown system accounts. If your deep desire for encrypted communication still draws you to this option, we recommend using Apache with SSL or svnserve with SASL encryption instead.

  • Do not be seduced by the simple idea of having all of your users access a repository directly via file:// URLs. Even if the repository is readily available to everyone via a network share, this is a bad idea. It removes any layers of protection between the users and the repository: users can accidentally (or intentionally) corrupt the repository database, it becomes hard to take the repository offline for inspection or upgrade, and it can lead to a mess of file permission problems (see the section called “Supporting Multiple Repository Access Methods”). Note that this is also one of the reasons we warn against accessing repositories via svn+ssh:// URLs—from a security standpoint, it's effectively the same as local users accessing via file://, and it can entail all the same problems if the administrator isn't careful.

svnserve, a Custom Server

The svnserve program is a lightweight server, capable of speaking to clients over TCP/IP using a custom, stateful protocol. Clients contact an svnserve server by using URLs that begin with the svn:// or svn+ssh:// scheme. This section will explain the different ways of running svnserve, how clients authenticate themselves to the server, and how to configure appropriate access control to your repositories.

Invoking the Server

There are a few different ways to run the svnserve program:

  • Run svnserve as a standalone daemon, listening for requests.

  • Have the Unix inetd daemon temporarily spawn svnserve whenever a request comes in on a certain port.

  • Have SSH invoke a temporary svnserve over an encrypted tunnel.

  • Run svnserve as a Microsoft Windows service.

  • Run svnserve as a launchd job.

The following sections will cover in detail these various deployment options for svnserve.

svnserve as daemon

The easiest option is to run svnserve as a standalone daemon process. Use the -d option for this:

$ svnserve -d
$               # svnserve is now running, listening on port 3690

When running svnserve in daemon mode, you can use the --listen-port and --listen-host options to customize the exact port and hostname to bind to.

Once we successfully start svnserve as explained previously, it makes every repository on your system available to the network. A client needs to specify an absolute path in the repository URL. For example, if a repository is located at /var/svn/project1, a client would reach it via svn://host.example.com/var/svn/project1. To increase security, you can pass the -r option to svnserve, which restricts it to exporting only repositories below that path. For example:

$ svnserve -d -r /var/svn
…

Using the -r option effectively modifies the location that the program treats as the root of the remote filesystem space. Clients then use URLs that have that path portion removed from them, leaving much shorter (and much less revealing) URLs:

$ svn checkout svn://host.example.com/project1
…

svnserve via inetd

If you want inetd to launch the process, you need to pass the -i (--inetd) option. In the following example, we've shown the output from running svnserve -i at the command line, but note that this isn't how you actually start the daemon; see the paragraphs following the example for how to configure inetd to start svnserve.

$ svnserve -i
( success ( 2 2 ( ) ( edit-pipeline svndiff1 absent-entries commit-revprops d\
epth log-revprops atomic-revprops partial-replay ) ) )

When invoked with the --inetd option, svnserve attempts to speak with a Subversion client via stdin and stdout using a custom protocol. This is the standard behavior for a program being run via inetd. The IANA has reserved port 3690 for the Subversion protocol, so on a Unix-like system you can add lines to /etc/services such as these (if they don't already exist):

svn           3690/tcp   # Subversion
svn           3690/udp   # Subversion

If your system is using a classic Unix-like inetd daemon, you can add this line to /etc/inetd.conf:

svn stream tcp nowait svnowner /usr/bin/svnserve svnserve -i

Make sure svnowner is a user that has appropriate permissions to access your repositories. Now, when a client connection comes into your server on port 3690, inetd will spawn an svnserve process to service it. Of course, you may also want to add -r to the configuration line as well, to restrict which repositories are exported.

svnserve over a tunnel

Another way to invoke svnserve is in tunnel mode, using the -t option. This mode assumes that a remote-service program such as rsh or ssh has successfully authenticated a user and is now invoking a private svnserve process as that user. (Note that you, the user, will rarely, if ever, have reason to invoke svnserve with the -t at the command line; instead, the SSH daemon does so for you.) The svnserve program behaves normally (communicating via stdin and stdout) and assumes that the traffic is being automatically redirected over some sort of tunnel back to the client. When svnserve is invoked by a tunnel agent like this, be sure that the authenticated user has full read and write access to the repository database files. It's essentially the same as a local user accessing the repository via file:// URLs.

This option is described in much more detail later in this chapter in the section called “Tunneling over SSH”.

svnserve as a Windows service

If your Windows system is a descendant of Windows NT (Windows 2000 or newer), you can run svnserve as a standard Windows service. This is typically a much nicer experience than running it as a standalone daemon with the --daemon (-d) option. Using daemon mode requires launching a console, typing a command, and then leaving the console window running indefinitely. A Windows service, however, runs in the background, can start at boot time automatically, and can be started and stopped using the same consistent administration interface as other Windows services.

You'll need to define the new service using the command-line tool SC.EXE. Much like the inetd configuration line, you must specify an exact invocation of svnserve for Windows to run at startup time:

C:\> sc create svn
        binpath= "C:\svn\bin\svnserve.exe --service -r C:\repos"
        displayname= "Subversion Server"
        depend= Tcpip
        start= auto

This defines a new Windows service named svn which executes a particular svnserve.exe command when started (in this case, rooted at C:\repos). There are a number of caveats in the prior example, however.

First, notice that the svnserve.exe program must always be invoked with the --service option. Any other options to svnserve must then be specified on the same line, but you cannot add conflicting options such as --daemon (-d), --tunnel, or --inetd (-i). Options such as -r or --listen-port are fine, though. Second, be careful about spaces when invoking the SC.EXE command: the key= value patterns must have no spaces between key= and must have exactly one space before the value. Lastly, be careful about spaces in your command line to be invoked. If a directory name contains spaces (or other characters that need escaping), place the entire inner value of binpath in double quotes, by escaping them:

C:\> sc create svn
        binpath= "\"C:\program files\svn\bin\svnserve.exe\" --service -r C:\repos"
        displayname= "Subversion Server"
        depend= Tcpip
        start= auto

Also note that the word binpath is misleading—its value is a command line, not the path to an executable. That's why you need to surround it with quotes if it contains embedded spaces.

Once the service is defined, it can be stopped, started, or queried using standard GUI tools (the Services administrative control panel), or at the command line:

C:\> net stop svn
C:\> net start svn

The service can also be uninstalled (i.e., undefined) by deleting its definition: sc delete svn. Just be sure to stop the service first! The SC.EXE program has many other subcommands and options; run sc /? to learn more about it.

svnserve as a launchd job

Mac OS X (10.4 and higher) uses launchd to manage processes (including daemons) both system-wide and per-user. A launchd job is specified by parameters in an XML property list file, and the launchctl command is used to manage the lifecycle of those jobs.

When configured to run as a launchd job, svnserve is automatically launched on demand whenever incoming Subversion svn:// network traffic needs to be handled. This is far more convenient than a configuration which requires you to manually invoke svnserve as a long-running background process.

To configure svnserve as a launchd job, first create a job definition file named /Library/LaunchDaemons/org.apache.subversion.svnserve.plist. Example 6.1, “A sample svnserve launchd job definition” provides an example of such a file.

Example 6.1. A sample svnserve launchd job definition

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"
    "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
    <dict>
        <key>Label</key>
        <string>org.apache.subversion.svnserve</string>
        <key>ServiceDescription</key>
        <string>Host Subversion repositories using svn:// scheme</string>
        <key>ProgramArguments</key>
        <array>
            <string>/usr/bin/svnserve</string>
            <string>--inetd</string>
            <string>--root=/var/svn</string>
        </array>
        <key>UserName</key>
        <string>svn</string>
        <key>GroupName</key>
        <string>svn</string>
        <key>inetdCompatibility</key>
        <dict>
            <key>Wait</key>
            <false/>
        </dict>
        <key>Sockets</key>
        <dict>
            <key>Listeners</key>
            <array>
                <dict>
                    <key>SockServiceName</key>
                    <string>svn</string>
                    <key>Bonjour</key>
                    <true/>
                </dict>
            </array>
        </dict>
    </dict>
</plist>


[Warning]Warning

The launchd system can be somewhat challenging to learn. Fortunately, documentation exists for the commands described in this section. For example, run man launchd from the command line to see the manual page for launchd itself, man launchd.plist to read about the job definition format, etc.

Once your job definition file is created, you can activate the job using launchctl load:

$ sudo launchctl load \
       -w /Library/LaunchDaemons/org.apache.subversion.svnserve.plist

To be clear, this action doesn't actually launch svnserve yet. It simply tells launchd how to fire up svnserve when incoming networking traffic arrives on the svn network port; it will be terminated it after the traffic has been handled.

[Note]Note

Because we want svnserve to be a system-wide daemon process, we need to use sudo to manage this job as an administrator. Note also that the UserName and GroupName keys in the definition file are optional—if omitted, the job will run as the user who loaded the job.

Deactivating the job is just as easy to do—use launchctl unload:

$ sudo launchctl unload \
       -w /Library/LaunchDaemons/org.apache.subversion.svnserve.plist

launchctl also provides a way for you to query the status of jobs. If the job is loaded, there will be line which matches the Label specified in the job definition file:

$ sudo launchctl list | grep org.apache.subversion.svnserve
-       0       org.apache.subversion.svnserve
$

Built-in Authentication and Authorization

When a client connects to an svnserve process, the following things happen:

  • The client selects a specific repository.

  • The server processes the repository's conf/svnserve.conf file and begins to enforce any authentication and authorization policies it describes.

  • Depending on the defined policies, one of the following may occur:

    • The client may be allowed to make requests anonymously, without ever receiving an authentication challenge.

    • The client may be challenged for authentication at any time.

    • If operating in tunnel mode, the client will declare itself to be already externally authenticated (typically by SSH).

The svnserve server, by default, knows only how to send a CRAM-MD5[52] authentication challenge. In essence, the server sends a small amount of data to the client. The client uses the MD5 hash algorithm to create a fingerprint of the data and password combined, and then sends the fingerprint as a response. The server performs the same computation with the stored password to verify that the result is identical. At no point does the actual password travel over the network.

If your svnserve server was built with SASL support, it not only knows how to send CRAM-MD5 challenges, but also likely knows a whole host of other authentication mechanisms. See the section called “Using svnserve with SASL” later in this chapter to learn how to configure SASL authentication and encryption.

It's also possible, of course, for the client to be externally authenticated via a tunnel agent, such as ssh. In that case, the server simply examines the user it's running as, and uses this name as the authenticated username. For more on this, see the later section, the section called “Tunneling over SSH”.

As you've already guessed, a repository's svnserve.conf file is the central mechanism for controlling authentication and authorization policies. The file has the same format as other configuration files (see the section called “Runtime Configuration Area”): section names are marked by square brackets ([ and ]), comments begin with hashes (#), and each section contains specific variables that can be set (variable = value). Let's walk through these files and learn how to use them.

Create a users file and realm

For now, the [general] section of svnserve.conf has all the variables you need. Begin by changing the values of those variables: choose a name for a file that will contain your usernames and passwords and choose an authentication realm:

[general]
password-db = userfile
realm = example realm

The realm is a name that you define. It tells clients which sort of authentication namespace they're connecting to; the Subversion client displays it in the authentication prompt and uses it as a key (along with the server's hostname and port) for caching credentials on disk (see the section called “Caching credentials”). The password-db variable points to a separate file that contains a list of usernames and passwords, using the same familiar format. For example:

[users]
harry = foopassword
sally = barpassword

The value of password-db can be an absolute or relative path to the users file. For many admins, it's easy to keep the file right in the conf/ area of the repository, alongside svnserve.conf. On the other hand, it's possible you may want to have two or more repositories share the same users file; in that case, the file should probably live in a more public place. The repositories sharing the users file should also be configured to have the same realm, since the list of users essentially defines an authentication realm. Wherever the file lives, be sure to set the file's read and write permissions appropriately. If you know which user(s) svnserve will run as, restrict read access to the users file as necessary.

Set access controls

There are two more variables to set in the svnserve.conf file: they determine what unauthenticated (anonymous) and authenticated users are allowed to do. The variables anon-access and auth-access can be set to the value none, read, or write. Setting the value to none prohibits both reading and writing; read allows read-only access to the repository, and write allows complete read/write access to the repository. For example:

[general]
password-db = userfile
realm = example realm

# anonymous users can only read the repository
anon-access = read

# authenticated users can both read and write
auth-access = write

The example settings are, in fact, the default values of the variables, should you forget to define them. If you want to be even more conservative, you can block anonymous access completely:

[general]
password-db = userfile
realm = example realm

# anonymous users aren't allowed
anon-access = none

# authenticated users can both read and write
auth-access = write

The server process understands not only these blanket access controls to the repository, but also finer-grained access restrictions placed on specific files and directories within the repository. To make use of this feature, you need to define a file containing more detailed rules, and then set the authz-db variable to point to it:

[general]
password-db = userfile
realm = example realm

# Specific access rules for specific locations
authz-db = authzfile

We discuss the syntax of the authzfile file in detail later in this chapter, in the section called “Path-Based Authorization”. Note that the authz-db variable isn't mutually exclusive with the anon-access and auth-access variables; if all the variables are defined at once, all of the rules must be satisfied before access is allowed.

Using svnserve with SASL

For many teams, the built-in CRAM-MD5 authentication is all they need from svnserve. However, if your server (and your Subversion clients) were built with the Cyrus Simple Authentication and Security Layer (SASL) library, you have a number of authentication and encryption options available to you.

Normally, when a subversion client connects to svnserve, the server sends a greeting that advertises a list of the capabilities it supports, and the client responds with a similar list of capabilities. If the server is configured to require authentication, it then sends a challenge that lists the authentication mechanisms available; the client responds by choosing one of the mechanisms, and then authentication is carried out in some number of round-trip messages. Even when SASL capabilities aren't present, the client and server inherently know how to use the CRAM-MD5 and ANONYMOUS mechanisms (see the section called “Built-in Authentication and Authorization”). If server and client were linked against SASL, a number of other authentication mechanisms may also be available. However, you'll need to explicitly configure SASL on the server side to advertise them.

Authenticating with SASL

To activate specific SASL mechanisms on the server, you'll need to do two things. First, create a [sasl] section in your repository's svnserve.conf file with an initial key-value pair:

[sasl]
use-sasl = true

Second, create a main SASL configuration file called svn.conf in a place where the SASL library can find it—typically in the directory where SASL plug-ins are located. You'll have to locate the plug-in directory on your particular system, such as /usr/lib/sasl2/ or /etc/sasl2/. (Note that this is not the svnserve.conf file that lives within a repository!)

On a Windows server, you'll also have to edit the system registry (using a tool such as regedit) to tell SASL where to find things. Create a registry key named [HKEY_LOCAL_MACHINE\SOFTWARE\Carnegie Mellon\Project Cyrus\SASL Library], and place two keys inside it: a key called SearchPath (whose value is a path to the directory containing the SASL sasl*.dll plug-in libraries), and a key called ConfFile (whose value is a path to the parent directory containing the svn.conf file you created).

Because SASL provides so many different kinds of authentication mechanisms, it would be foolish (and far beyond the scope of this book) to try to describe every possible server-side configuration. Instead, we recommend that you read the documentation supplied in the doc/ subdirectory of the SASL source code. It goes into great detail about every mechanism and how to configure the server appropriately for each. For the purposes of this discussion, we'll just demonstrate a simple example of configuring the DIGEST-MD5 mechanism. For example, if your subversion.conf (or svn.conf) file contains the following:

pwcheck_method: auxprop
auxprop_plugin: sasldb
sasldb_path: /etc/my_sasldb
mech_list: DIGEST-MD5

you've told SASL to advertise the DIGEST-MD5 mechanism to clients and to check user passwords against a private password database located at /etc/my_sasldb. A system administrator can then use the saslpasswd2 program to add or modify usernames and passwords in the database:

$ saslpasswd2 -c -f /etc/my_sasldb -u realm username

A few words of warning: first, make sure the realm argument to saslpasswd2 matches the same realm you've defined in your repository's svnserve.conf file; if they don't match, authentication will fail. Also, due to a shortcoming in SASL, the common realm must be a string with no space characters. Finally, if you decide to go with the standard SASL password database, make sure the svnserve program has read access to the file (and possibly write access as well, if you're using a mechanism such as OTP).

This is just one simple way of configuring SASL. Many other authentication mechanisms are available, and passwords can be stored in other places such as in LDAP or a SQL database. Consult the full SASL documentation for details.

Remember that if you configure your server to only allow certain SASL authentication mechanisms, this forces all connecting clients to have SASL support as well. Any Subversion client built without SASL support (which includes all pre-1.5 clients) will be unable to authenticate. On the one hand, this sort of restriction may be exactly what you want (My clients must all use Kerberos!). However, if you still want non-SASL clients to be able to authenticate, be sure to advertise the CRAM-MD5 mechanism as an option. All clients are able to use CRAM-MD5, whether they have SASL capabilities or not.

SASL encryption

SASL is also able to perform data encryption if a particular mechanism supports it. The built-in CRAM-MD5 mechanism doesn't support encryption, but DIGEST-MD5 does, and mechanisms such as SRP actually require use of the OpenSSL library. To enable or disable different levels of encryption, you can set two values in your repository's svnserve.conf file:

[sasl]
use-sasl = true
min-encryption = 128
max-encryption = 256

The min-encryption and max-encryption variables control the level of encryption demanded by the server. To disable encryption completely, set both values to 0. To enable simple checksumming of data (i.e., prevent tampering and guarantee data integrity without encryption), set both values to 1. If you wish to allow—but not require—encryption, set the minimum value to 0, and the maximum value to some bit length. To require encryption unconditionally, set both values to numbers greater than 1. In our previous example, we require clients to do at least 128-bit encryption, but no more than 256-bit encryption.

Tunneling over SSH

svnserve's built-in authentication (and SASL support) can be very handy, because it avoids the need to create real system accounts. On the other hand, some administrators already have well-established SSH authentication frameworks in place. In these situations, all of the project's users already have system accounts and the ability to SSH into the server machine.

It's easy to use SSH in conjunction with svnserve. The client simply uses the svn+ssh:// URL scheme to connect:

$ whoami
harry

$ svn list svn+ssh://host.example.com/repos/project
harryssh@host.example.com's password:  *****

foo
bar
baz
…

In this example, the Subversion client is invoking a local ssh process, connecting to host.example.com, authenticating as the user harryssh (according to SSH user configuration), then spawning a private svnserve process on the remote machine running as the user harryssh. The svnserve command is being invoked in tunnel mode (-t), and its network protocol is being tunneled over the encrypted connection by ssh, the tunnel agent. If the client performs a commit, the authenticated username harryssh will be used as the author of the new revision.

The important thing to understand here is that the Subversion client is not connecting to a running svnserve daemon. This method of access doesn't require a daemon, nor does it notice one if present. It relies wholly on the ability of ssh to spawn a temporary svnserve process, which then terminates when the network connection is closed.

When using svn+ssh:// URLs to access a repository, remember that it's the ssh program prompting for authentication, and not the svn client program. That means there's no automatic password-caching going on (see the section called “Caching credentials”). The Subversion client often makes multiple connections to the repository, though users don't normally notice this due to the password caching feature. When using svn+ssh:// URLs, however, users may be annoyed by ssh repeatedly asking for a password for every outbound connection. The solution is to use a separate SSH password-caching tool such as ssh-agent on a Unix-like system, or pageant on Windows.

When running over a tunnel, authorization is primarily controlled by operating system permissions to the repository's database files; it's very much the same as if Harry were accessing the repository directly via a file:// URL. If multiple system users are going to be accessing the repository directly, you may want to place them into a common group, and you'll need to be careful about umasks (be sure to read the section called “Supporting Multiple Repository Access Methods” later in this chapter). But even in the case of tunneling, you can still use the svnserve.conf file to block access, by simply setting auth-access = read or auth-access = none.[53]

You'd think that the story of SSH tunneling would end here, but it doesn't. Subversion allows you to create custom tunnel behaviors in your runtime config file (see the section called “Runtime Configuration Area”). For example, suppose you want to use RSH instead of SSH.[54] In the [tunnels] section of your config file, simply define it like this:

[tunnels]
rsh = rsh

And now, you can use this new tunnel definition by using a URL scheme that matches the name of your new variable: svn+rsh://host/path. When using the new URL scheme, the Subversion client will actually be running the command rsh host svnserve -t behind the scenes. If you include a username in the URL (e.g., svn+rsh://username@host/path), the client will also include that in its command (rsh username@host svnserve -t). But you can define new tunneling schemes to be much more clever than that:

[tunnels]
joessh = $JOESSH /opt/alternate/ssh -p 29934

This example demonstrates a couple of things. First, it shows how to make the Subversion client launch a very specific tunneling binary (the one located at /opt/alternate/ssh) with specific options. In this case, accessing an svn+joessh:// URL would invoke the particular SSH binary with -p 29934 as arguments—useful if you want the tunnel program to connect to a nonstandard port.

Second, it shows how to define a custom environment variable that can override the name of the tunneling program. Setting the SVN_SSH environment variable is a convenient way to override the default SSH tunnel agent. But if you need to have several different overrides for different servers, each perhaps contacting a different port or passing a different set of options to SSH, you can use the mechanism demonstrated in this example. Now if we were to set the JOESSH environment variable, its value would override the entire value of the tunnel variable—$JOESSH would be executed instead of /opt/alternate/ssh -p 29934.

SSH Configuration Tricks

It's possible to control not only the way in which the client invokes ssh, but also to control the behavior of sshd on your server machine. In this section, we'll show how to control the exact svnserve command executed by sshd, as well as how to have multiple users share a single system account.

Initial setup

To begin, locate the home directory of the account you'll be using to launch svnserve. Make sure the account has an SSH public/private keypair installed, and that the user can log in via public-key authentication. Password authentication will not work, since all of the following SSH tricks revolve around using the SSH authorized_keys file.

If it doesn't already exist, create the authorized_keys file (on Unix, typically ~/.ssh/authorized_keys). Each line in this file describes a public key that is allowed to connect. The lines are typically of the form:

  ssh-dsa AAAABtce9euch… user@example.com

The first field describes the type of key, the second field is the base64-encoded key itself, and the third field is a comment. However, it's a lesser known fact that the entire line can be preceded by a command field:

  command="program" ssh-dsa AAAABtce9euch… user@example.com

When the command field is set, the SSH daemon will run the named program instead of the typical tunnel-mode svnserve invocation that the Subversion client asks for. This opens the door to a number of server-side tricks. In the following examples, we abbreviate the lines of the file as:

  command="program" TYPE KEY COMMENT

Controlling the invoked command

Because we can specify the executed server-side command, it's easy to name a specific svnserve binary to run and to pass it extra arguments:

  command="/path/to/svnserve -t -r /virtual/root" TYPE KEY COMMENT

In this example, /path/to/svnserve might be a custom wrapper script around svnserve which sets the umask (see the section called “Supporting Multiple Repository Access Methods”). It also shows how to anchor svnserve in a virtual root directory, just as one often does when running svnserve as a daemon process. This might be done either to restrict access to parts of the system, or simply to relieve the user of having to type an absolute path in the svn+ssh:// URL.

It's also possible to have multiple users share a single account. Instead of creating a separate system account for each user, generate a public/private key pair for each person. Then place each public key into the authorized_keys file, one per line, and use the --tunnel-user option:

  command="svnserve -t --tunnel-user=harry" TYPE1 KEY1 harry@example.com
  command="svnserve -t --tunnel-user=sally" TYPE2 KEY2 sally@example.com

This example allows both Harry and Sally to connect to the same account via public key authentication. Each of them has a custom command that will be executed; the --tunnel-user option tells svnserve to assume that the named argument is the authenticated user. Without --tunnel-user, it would appear as though all commits were coming from the one shared system account.

A final word of caution: giving a user access to the server via public-key in a shared account might still allow other forms of SSH access, even if you've set the command value in authorized_keys. For example, the user may still get shell access through SSH or be able to perform X11 or general port forwarding through your server. To give the user as little permission as possible, you may want to specify a number of restrictive options immediately after the command:

  command="svnserve -t --tunnel-user=harry",no-port-forwarding,no-agent-forw
arding,no-X11-forwarding,no-pty TYPE1 KEY1 harry@example.com

Note that this all must be on one line—truly on one line—since SSH authorized_keys files do not even allow the conventional backslash character (\) for line continuation. The only reason we've shown it with a line break is to fit it on the physical page of a book.

httpd, the Apache HTTP Server

The Apache HTTP Server is a heavy-duty network server that Subversion can leverage. Via a custom module, httpd makes Subversion repositories available to clients via the WebDAV/DeltaV[55] protocol, which is an extension to HTTP 1.1. This protocol takes the ubiquitous HTTP protocol that is the core of the World Wide Web, and adds writing—specifically, versioned writing—capabilities. The result is a standardized, robust system that is conveniently packaged as part of the Apache 2.0 software, supported by numerous operating systems and third-party products, and doesn't require network administrators to open up yet another custom port.[56] While an Apache-Subversion server has more features than svnserve, it's also a bit more difficult to set up. With flexibility often comes more complexity.

Much of the following discussion includes references to Apache configuration directives. While some examples are given of the use of these directives, describing them in full is outside the scope of this chapter. The Apache team maintains excellent documentation, publicly available on their web site at http://httpd.apache.org. For example, a general reference for the configuration directives is located at http://httpd.apache.org/docs-2.0/mod/directives.html.

Also, as you make changes to your Apache setup, it is likely that somewhere along the way a mistake will be made. If you are not already familiar with Apache's logging subsystem, you should become aware of it. In your httpd.conf file are directives that specify the on-disk locations of the access and error logs generated by Apache (the CustomLog and ErrorLog directives, respectively). Subversion's mod_dav_svn uses Apache's error logging interface as well. You can always browse the contents of those files for information that might reveal the source of a problem that is not clearly noticeable otherwise.

Prerequisites

To network your repository over HTTP, you basically need four components, available in two packages. You'll need Apache httpd 2.0 or newer, the mod_dav DAV module that comes with it, Subversion, and the mod_dav_svn filesystem provider module distributed with Subversion. Once you have all of those components, the process of networking your repository is as simple as:

  • Getting httpd up and running with the mod_dav module

  • Installing the mod_dav_svn backend to mod_dav, which uses Subversion's libraries to access the repository

  • Configuring your httpd.conf file to export (or expose) the repository

You can accomplish the first two items either by compiling httpd and Subversion from source code or by installing prebuilt binary packages of them on your system. For the most up-to-date information on how to compile Subversion for use with the Apache HTTP Server, as well as how to compile and configure Apache itself for this purpose, see the INSTALL file in the top level of the Subversion source code tree.

Basic Apache Configuration

Once you have all the necessary components installed on your system, all that remains is the configuration of Apache via its httpd.conf file. Instruct Apache to load the mod_dav_svn module using the LoadModule directive. This directive must precede any other Subversion-related configuration items. If your Apache was installed using the default layout, your mod_dav_svn module should have been installed in the modules subdirectory of the Apache install location (often /usr/local/apache2). The LoadModule directive has a simple syntax, mapping a named module to the location of a shared library on disk:

LoadModule dav_svn_module     modules/mod_dav_svn.so

Apache interprets the LoadModule configuration item's library path as relative to its own server root. If configured as previously shown, Apache will look for the Subversion DAV module shared library in its own modules/ subdirectory. Depending on how Subversion was installed on your system, you might need to specify a different path for this library altogether, perhaps even an absolute path such as in the following example:

LoadModule dav_svn_module     C:/Subversion/lib/mod_dav_svn.so

Note that if mod_dav was compiled as a shared object (instead of statically linked directly to the httpd binary), you'll need a similar LoadModule statement for it, too. Be sure that it comes before the mod_dav_svn line:

LoadModule dav_module         modules/mod_dav.so
LoadModule dav_svn_module     modules/mod_dav_svn.so

At a later location in your configuration file, you now need to tell Apache where you keep your Subversion repository (or repositories). The Location directive has an XML-like notation, starting with an opening tag and ending with a closing tag, with various other configuration directives in the middle. The purpose of the Location directive is to instruct Apache to do something special when handling requests that are directed at a given URL or one of its children. In the case of Subversion, you want Apache to simply hand off support for URLs that point at versioned resources to the DAV layer. You can instruct Apache to delegate the handling of all URLs whose path portions (the part of the URL that follows the server's name and the optional port number) begin with /repos/ to a DAV provider whose repository is located at /var/svn/repository using the following httpd.conf syntax:

<Location /repos>
  DAV svn
  SVNPath /var/svn/repository
</Location>

If you plan to support multiple Subversion repositories that will reside in the same parent directory on your local disk, you can use an alternative directive—SVNParentPath—to indicate that common parent directory. For example, if you know you will be creating multiple Subversion repositories in a directory /var/svn that would be accessed via URLs such as http://my.server.com/svn/repos1, http://my.server.com/svn/repos2, and so on, you could use the httpd.conf configuration syntax in the following example:

<Location /svn>
  DAV svn

  # Automatically map any "/svn/foo" URL to repository /var/svn/foo
  SVNParentPath /var/svn
</Location>

Using this syntax, Apache will delegate the handling of all URLs whose path portions begin with /svn/ to the Subversion DAV provider, which will then assume that any items in the directory specified by the SVNParentPath directive are actually Subversion repositories. This is a particularly convenient syntax in that, unlike the use of the SVNPath directive, you don't have to restart Apache to add or remove hosted repositories.

Be sure that when you define your new Location, it doesn't overlap with other exported locations. For example, if your main DocumentRoot is exported to /www, do not export a Subversion repository in <Location /www/repos>. If a request comes in for the URI /www/repos/foo.c, Apache won't know whether to look for a file repos/foo.c in the DocumentRoot, or whether to delegate mod_dav_svn to return foo.c from the Subversion repository. The result is often an error from the server of the form 301 Moved Permanently.