SVK 1.0 Beta 2 released.

Yesterday, the release of SVK 1.0 Beta 2 was announced .

The announcements list the following changes since Beta 1:

  • Fix keyword translation for undesired characters. [matthewd]
  • Fix svk switch from a removed branch.
  • Don’t trust LML would return a valid encoding that Encode knows. [Eric Gillespie ]
  • Recognize merge-conflict error, so no stacktrace for it.
  • svk info now skips files not under version control. [#8220] [gugod]
  • Fix locked mirror messages.
  • Various pool usage fixes.

The software can be downloaded from the SVK download area.

svk 1.0beta1 Released

In a message to the SVK dev list, Chia-liang Kao, original author and project maintainer for the SVK version control system, has announced that the software has hit beta stage.

According to the message there will be a three week beta period, followed by a two week release candidate release cycle, if everything goes well. The 1.o release would follow.

Posted in SVK

Explaining the SVK Workflow

As I’m beginning to talk to people about the SVK version control tool more, I’m noticing that the concept of mirroring isn’t as obvious to some people as it was to me when I started messing around with it (or maybe I just choose not to remember how unobvious it was), so I thought I would take some time to try to explain some of the concepts at an extremely high (and hopefully easier to understand) level.

Sometimes the best way to explain a concept is with a picture, so I threw together this small data flow diagram that shows the flow of data through the SVK system and what commands cause this data transfer to happen.

SVK WorkFlow

The basic high level workflow of an SVK session looks like this:

  1. Create A Mirror
    You mirror the repository using the svk mirror command. For me, I find it easiest to create my mirrors at the topmost level (for example, a mirror of the bieberlabs trunk might be at //bieberlabs/trunk).
  2. Synchronize Your Mirror
    This brings down the revisions that you do not have yet in your main (and remote) repository to your newly established mirrored path.
  3. Create a Local Branch To Do Your Work
    Many times I see that people default to checking out from the actual mirror (in the above example, //bieberlabs/trunk). This will work, but you lose the whole beneifit of SVK. If you check out of a mirror, every commit you make will be synchronized back to the main repository. This is not desired behavior most of the time. So, create a local branch first. I like to create a top level directory in the repository called //local that I put my local branches in, just so I know where to find them.
  4. Periodically synchronize your mirror and merge those changes into your local branch
    This is where all of the power is. You now have a local branch that you can keep completely up to date with the repository that others are doing work in. Using the svk smerge command, you can merge changes from //bieberlabs//trunk to //local/mychange as you synchronize your mirror. This allows you to always be working on the current code. You can commit your work and not effect anyone else working in the repository until you are ready to merge into the mirror.
  5. Once you have merged changes to your local branch, update your workarea
    A common misunderstanding when you smerge into your local branch is that the changes appear magically in your workarea. This is not the case. Remember, when you are smerging you are working in the repository. You still have to update your workarea (and possibly resolve conflicts) from the repository in order to reflect your changes.
  6. Merge your branch back into the mirror
    Once you have finished your change, tested and committed (probably multiple times), you can now smerge your branch back into the mirror. When you perform this merge, the changes will be moved to the source repository and subsequently synced to your mirror (the line in the diagram is red to signify this). Once this happens, all of your changes are visible to everyone else, in one commit.

Here’s the nice thing about working like this. When merging back and forth, you are not searching for ranges of revision numbers to merge. SVK remembers each merge done, so you can use your local branch forever, because you can constantly keep it up to date. This gets rid of the “make a branch, change the branch, merge the branch , delete the branch” cycle and allows you to have one place to work all of the time.

I hope that this really high level view gives a better understanding of the workflow surrounding using SVK and illustrates very simply the advantages of the tool.

For a step by step walk through the tool, you can go through the SVK tutorials.

SVK 0.30 Install for Windows Available

An installer for SVK 0.30 for Windows is available here. This is not a standalone .EXE file, but an installer for the SVK tool.

Some caveats from looking at it briefly this morning:

  • Make sure you have Subversion installed. I thought I had installed it already on my Windows XP machine – but I was wrong.
  • I was unable to authenticate with my SSL WebDAV server through SVK. I had to execute an svn ls https://blah.blah.blah in order to authenticate properly. Once I did this, I was able to create a mirror and sync to it.

I’m not sure if this is the “official” version of the Windows install, but it is one that is available for those who want to start running 0.30.

SVK Tutorial Feedback

I’m getting some positive feedback on the set of SVK tutorials that I wrote a few weeks back. Thanks to everyone sending comments and commenting on them on your blogs. It’s really encouraging to know that something you have worked through and written down has been helpful to other people in investigating a really useful tool.

Keep the feedback coming.

As an aside, there is a great article on perl.com about SVK that was written by the tools author. Well worth the read.