After about a month of bug reports, segmentation faults, patches, and prodding of the core developers, we finally have a new release of everyone's favorite composited dock, AWN.
After some prodding on IRC, I created a branch of the 0.2 release branch, called 0.2-stable-testing. Here's how I described my workflow on this branch in the forums:
- I get patches from IRC/forums/launchpad. If they aren't on launchpad, I ask that they go there so I can put a reference to them in the commit message.
- When the patch is on launchpad, I add the stable branch to the bug with the status message "fix in progress".
- I install/run awn with the patch applied, and check the console. If there are no extra assert/CRITICAL/WARNING messages, and I don't crash within 10 minutes, I commit the patch (usually with whitespace fixes, etc.) to the branch, and push to launchpad.
- I change the status message on the bug to "fix available" and note the revision that the patch was applied on.
In all, there were a total of 13 recorded bugs fixed in my branch. About halfway through, I posted a call for a bugfix release:
So apparently, AWN is now on the front page of launchpad. This inevitably means more users, and more of the same questions about crashes, etc. occurring with the 0.2 release. Many of these crashes have been addressed in my 0.2-stable-testing branch, and I am pleased to report that several people are actively testing this branch and indeed finding it stable. So, I propose that the rest of the patches in my branch be reviewed (you can find them at the bottom of the branch details page), and a point release (0.2.1) be made.
Additionally, I don't think it's in our best interests to have the only available method of retrieving awn-extras be through bzr (even though I am a strong advocate of bzr). We need, at the very least, a snapshot of awn-extras to be released. Preferably, the buggy clock applet should be fixed, moved or removed before this happens.
The result is where we are today. There are only two things that concern me about this release: Two minor features crept in, and I had found a bug the night before, but was too tired to file it. I have been opposed to adding new features (however minor they are) in a point release, because it's convention to leave new stuff to version bumps of one of the more major versions (e.g., 0.2 to 0.3, or 1.0 to 2.0). There's a good reason that most projects do this, too. New features (especially those that are untested)bring new bugs, which is not ideal for a bugfix release.
Thanks are in order for moonbeam (who wrote most of the stability patches), mhr3 (who reviewed said patches), and njpatel (who released it). It sounds like 0.3 is going to be very interesting. Hopefully I can get my desktop-agnostic branch finished and merged.