SVK is a client for SVN built using perl. It makes a number of improvements over the standard svn client, while retaining much of the same feel. It works with the standard Subversion server and works perfectly in an environment with some users using svn and some using svk on the client side. It provides a number of sizable advantages over the standard svn client and is a must have for any development project.
UPDATE: It’s been a while since this article was written and while still accurate, do yourself a favor and go use Git.
There are three things SVK provides that are particularly noteworthy.
- Tracks all branching making the process of pulling and pushing changes between trunk and branch very easy.
Optionally adds another layer to the mix. While Subversion has Repo and Working Copy, SVK has Repo (svn server) Depot (local mirror) and Working Copy (where you work). This provides two main features.
- Works offline
- You can commit changes to the depot without committing them to the repo.
The working copy doesn’t have dot files, rather they are stored in ~/.svk (by default). Makes for much cleaner working copies. This also makes it so that you can use svk as a quick and dirty deployment tool as the working copies are clean.
The 2.5 different ways of using SVK.
For practical use the terms depot and mirror are often the same. While they aren’t completely synonymous they are somewhat used interchangeably. A depot can contain many mirrors.
Standard Subversion Approach
You can use it like Subversion with the additional benefit of tracking all branching and merges. In this case the mirror becomes transparent.
You can use it as a distributed version control. Using this approach the mirror is the key. It can be a bit hard to get used to, especially if you are used to SVN or CVS. However stick with it, it has a lot of benefits. Not only does it provide offline functionality, but it also provides local versioning.
Distributed with local branch approach.
When using SVK in mirror mode you can also optionally branch locally without branching on the server (this is the ½). This is especially useful when you are working on a project where you don’t have write access to the repository. This way you can manage your versions (and keep up with the repository) and submit a patch when you have a change ready.
When using this approach the typical routine is “pull” (optionally when in branch), “update”, “modify”, “commit” repeat until ready to “push” (optionally when in branch).
Checkout directly from the repository.
svk co [repository] [local wc path] svk co http://host/svn/project ./code
svk ci , svk add, svk up just like subversion. See the common operations for branching, pulling and pushing.
When using this approach the typical routine is “pull” (optionally when in branch), “sync”, “update”, “modify”, “commit” repeat until ready to “push” (optionally when in branch).
Start by creating your Depot.
svk depotmap --init
Mirror Existing Repository
Next Mirror your existing repository. It is a good practice to mirror the entire repo as it makes things easier down the road and disk space is cheap. Since every operation under this directory is automatically (vs manually) synchronized with the remote SVN repository it is usually a good practice to either work on a branch, either a branch on the server, or a local branch) and not directly on the mirror.
svk mirror [repository url] [local depot url] example: svk mirror svn://HOST/var/svn/repos/Zoop //projects/Zoop
Sync local mirror with the repository
This will sync every commit since you last performed it. If it is the first time you are running the command it may take a while depending on how many revisions are in your project.
svk sync //projects/friends
Check out working copy from local mirror
svk co //projects/friends
Update Working Copy
To get changes from the mirror into your working copy.
Alt: Update Mirror & Working Copy
To get changes from the repo through your mirror into your working copy. Effectively performs a svk sync and then and svk up in one step.
svk up -s
Mirror mode + local branch
When using this approach the typical routine is “sync”, “merge”, “update”, “modify”, “commit” repeat until ready to “submit”.
Your local branch will live in the depot alongside the mirror.
Create your local branch
(Assumes you have already followed the mirror steps above through the sync).
Branching in SVK is the same as in SVN:
svk cp //projects/project-mirror //projects/project-local
Check out working copy from your local branch
Now you can check out code from your local working branch into your working directory. For this, follow a similar procedure as that of SVN:
mkdir project svk co //projects/project-local project or you can checkout just one branch svk co //projects/project-local/trunk project/trunk
You will want to keep your mirror in sync with the repo. The operation is automatic, but requires you to pull the trigger.
svk sync //projects/project-mirror/
If there are any changes you will likely want to merge them into your local branch.
svk smerge //projects/project-mirror/ //projects/project-local/
Lastly update to pull the changes from the depot into your wc.
svk cp [existing_location] [new_location] example: svk cp http://host/svn/trunk http://host/svn/branches/steve or svk cp http://host/svn/branches/steve http://host/svn/branches/steve-branches/cleanup-templates
Push and Pull
Pull will import changes from the branch you branched from.
Push will send your changes to the branch your branched from.
This will happen directly on the repo, unlike smerge which operates similarly, but on the depot only.
svk push svk pull
These are meant to be run from within the wc.
Following the example above.
$ cd branches/steve-branches/cleanup-templates
svk push would merge all changes made in the cleanup-templates branch to the steve branch.
svk pull would take all the changes made in the steve branch and merge them into the cleanup-templates branch.
Directory structure isn’t important here, except to keep things sane.
$ cd branches/steve
svk push would merge all changes made in the steve branch to trunk.
svk pull would take all the changes made in trunk and merge them into the steve branch.
$ cd trunk
can’t push or pull. This is the top level.
Many of the standard svn commands are also available.
svk add svk rm svk mv svk mkdir svk status svk diff
You should NEVER just ‘svk add *’ and ‘svk commit’ without checking
what you are committing. It shouldn’t be a surprise which files go to
The process when ready to commit should be:
close out all of your open files svk up -s # pulls from the repo, not just the mirror svk diff # carefully review your changes
svk commit [ file1 ] [ file2 ] ... [ filen ] or svk commit
without “-m” so you can remove unnecessary files in the commit message screen.