Someone recently delicioused Michael Ogawa's code_swarm project. It provides neat little visualisations of all the commits to a source code repository, focussing on interactions between all the different committers.
It turned out to be quite easy to create a JIRA code swarm. The project is open source with some pretty good instructions. It uses the processing framework to generate visualisations. I made some quick modifications to add images for committers (if available) and stop them from overlapping too much. Here's the history of JIRA:
Each of the particles represents a file that was committed. The particules will hover around the author that committed them. Particles are also colour coded:
- Blue particles - JSPs & VM files
- Green particles - Anything under **/test/*
- Red particles - *.java & **/src/*
- Turquoise particles - image files
Charles was kind enough to create a video for Confluence (using the original code swarm project):
2 Comment(s)
It's interesting with stats and visualisations after you've got over the "wow that's cool" stage to ask "what has become visible?"
Some features I noticed were:
* major explosions in activity as large branch work was merged to trunk
* weekends are visible as pronounced gaps in the commit stream. I think Atlassian should be proud of its sustainable work hours.
* week-long and two-week-long support rotations are visible as the developer's name falls out of the main group briefly and then swings back in.
By Chris Mountford at July 7, 2008 5:58 PM
Another interesting thing the videos demonstrate is the lack of "code ownership" in either product. Names will drift a little way out of the centre as people work on particular features, but everyone is very quickly dragged back to the middle.
By Charles Miller at July 11, 2008 7:31 PM