What do "branch", "tag" and "trunk" mean in Subversion repositories?

Hmm, not sure I agree with Nick re tag being similar to a branch. A tag is just a marker

• Trunk would be the main body of development, originating from the start of the project until the present.

• Branch will be a copy of code derived from a certain point in the trunk that is used for applying major changes to the code while preserving the integrity of the code in the trunk. If the major changes work according to plan, they are usually merged back into the trunk.

• Tag will be a point in time on the trunk or a branch that you wish to preserve. The two main reasons for preservation would be that either this is a major release of the software, whether alpha, beta, RC or RTM, or this is the most stable point of the software before major revisions on the trunk were applied.

In open source projects, major branches that are not accepted into the trunk by the project stakeholders can become the bases for forks -- e.g., totally separate projects that share a common origin with other source code.

The branch and tag subtrees are distinguished from the trunk in the following ways:

Subversion allows sysadmins to create hook scripts which are triggered for execution when certain events occur; for instance, committing a change to the repository. It is very common for a typical Subversion repository implementation to treat any path containing "/tag/" to be write-protected after creation; the net result is that tags, once created, are immutable (at least to "ordinary" users). This is done via the hook scripts, which enforce the immutability by preventing further changes if tag is a parent node of the changed object.

Subversion also has added features, since version 1.5, relating to "branch merge tracking" so that changes committed to a branch can be merged back into the trunk with support for incremental, "smart" merging.

First of all, as @AndrewFinnell and @KenLiu point out, in SVN the directory names themselves mean nothing -- "trunk, branches and tags" are simply a common convention that is used by most repositories. Not all projects use all of the directories (it's reasonably common not to use "tags" at all), and in fact, nothing is stopping you from calling them anything you'd like, though breaking convention is often confusing.

I'll describe probably the most common usage scenario of branches and tags, and give an example scenario of how they are used.

• Trunk: The main development area. This is where your next major release of the code lives, and generally has all the newest features.

• Branches: Every time you release a major version, it gets a branch created. This allows you to do bug fixes and make a new release without having to release the newest - possibly unfinished or untested - features.

• Tags: Every time you release a version (final release, release candidates (RC), and betas) you make a tag for it. This gives you a point-in-time copy of the code as it was at that state, allowing you to go back and reproduce any bugs if necessary in a past version, or re-release a past version exactly as it was. Branches and tags in SVN are lightweight - on the server, it does not make a full copy of the files, just a marker saying "these files were copied at this revision" that only takes up a few bytes. With this in mind, you should never be concerned about creating a tag for any released code. As I said earlier, tags are often omitted and instead, a changelog or other document clarifies the revision number when a release is made.

For example, let's say you start a new project. You start working in "trunk", on what will eventually be released as version 1.0.

• trunk/ - development version, soon to be 1.0
• branches/ - empty

Once 1.0.0 is finished, you branch trunk into a new "1.0" branch, and create a "1.0.0" tag. Now work on what will eventually be 1.1 continues in trunk.

• trunk/ - development version, soon to be 1.1
• branches/1.0 - 1.0.0 release version
• tags/1.0.0 - 1.0.0 release version

You come across some bugs in the code, and fix them in trunk, and then merge the fixes over to the 1.0 branch. You can also do the opposite, and fix the bugs in the 1.0 branch and then merge them back to trunk, but commonly projects stick with merging one-way only to lessen the chance of missing something. Sometimes a bug can only be fixed in 1.0 because it is obsolete in 1.1. It doesn't really matter: you only want to make sure that you don't release 1.1 with the same bugs that have been fixed in 1.0.

• trunk/ - development version, soon to be 1.1
• branches/1.0 - upcoming 1.0.1 release
• tags/1.0.0 - 1.0.0 release version

Once you find enough bugs (or maybe one critical bug), you decide to do a 1.0.1 release. So you make a tag "1.0.1" from the 1.0 branch, and release the code. At this point, trunk will contain what will be 1.1, and the "1.0" branch contains 1.0.1 code. The next time you release an update to 1.0, it would be 1.0.2.

• trunk/ - development version, soon to be 1.1
• branches/1.0 - upcoming 1.0.2 release
• tags/1.0.0 - 1.0.0 release version
• tags/1.0.1 - 1.0.1 release version

Eventually you are almost ready to release 1.1, but you want to do a beta first. In this case, you likely do a "1.1" branch, and a "1.1beta1" tag. Now, work on what will be 1.2 (or 2.0 maybe) continues in trunk, but work on 1.1 continues in the "1.1" branch.

• trunk/ - development version, soon to be 1.2
• branches/1.0 - upcoming 1.0.2 release
• branches/1.1 - upcoming 1.1.0 release
• tags/1.0.0 - 1.0.0 release version
• tags/1.0.1 - 1.0.1 release version
• tags/1.1beta1 - 1.1 beta 1 release version

Once you release 1.1 final, you do a "1.1" tag from the "1.1" branch.

You can also continue to maintain 1.0 if you'd like, porting bug fixes between all three branches (1.0, 1.1, and trunk). The important takeaway is that for every main version of the software you are maintaining, you have a branch that contains the latest version of code for that version.

Another use of branches is for features. This is where you branch trunk (or one of your release branches) and work on a new feature in isolation. Once the feature is completed, you merge it back in and remove the branch.

• trunk/ - development version, soon to be 1.2
• branches/1.1 - upcoming 1.1.0 release
• branches/ui-rewrite - experimental feature branch

The idea of this is when you're working on something disruptive (that would hold up or interfere with other people from doing their work), something experimental (that may not even make it in), or possibly just something that takes a long time (and you're afraid if it holding up a 1.2 release when you're ready to branch 1.2 from trunk), you can do it in isolation in branch. Generally you keep it up to date with trunk by merging changes into it all the time, which makes it easier to re-integrate (merge back to trunk) when you're finished.

Also note, the versioning scheme I used here is just one of many. Some teams would do bug fix/maintenance releases as 1.1, 1.2, etc., and major changes as 1.x, 2.x, etc. The usage here is the same, but you may name the branch "1" or "1.x" instead of "1.0" or "1.0.x". (Aside, semantic versioning is a good guide on how to do version numbers).

In addition to what Nick has said you can find out more at Streamed Lines: Branching Patterns for Parallel Software Development

In this figure main is the trunk, rel1-maint is a branch and 1.0 is a tag.

In general (tool agnostic view), a branch is the mechanism used for parallel development. An SCM can have from 0 to n branches. Subversion has 0.

• Trunk is a main branch recommended by Subversion, but you are in no way forced to create it. You could call it 'main' or 'releases', or not have one at all!

• Branch represents a development effort. It should never be named after a resource (like 'vonc_branch') but after:

  • a purpose 'myProject_dev' or 'myProject_Merge'
  • a release perimeter 'myProjetc1.0_dev'or myProject2.3_Merge' or 'myProject6..2_Patch1'...

• Tag is a snapshot of files in order to easily get back to that state. The problem is that tag and branch is the same in Subversion. And I would definitely recommend the paranoid approach:

  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.

A tag is final. Its content should never change. NEVER. Ever. You forgot a line in the release note? Create a new tag. Obsolete or remove the old one.

Now, I read a lot about "merging back such and such in such and such branches, then finally in the trunk branch". That is called merge workflow and there is nothing mandatory here. It is not because you have a trunk branch that you have to merge back anything.

By convention, the trunk branch can represent the current state of your development, but that is for a simple sequential project, that is a project which has:

• no 'in advance' development (for the preparing the next-next version implying such changes that they are not compatible with the current 'trunk' development)
• no massive refactoring (for testing a new technical choice)
• no long-term maintenance of a previous release

Because with one (or all) of those scenario, you get yourself four 'trunks', four 'current developments', and not all you do in those parallel development will necessarily have to be merged back in 'trunk'.