As part of the CPAN Report post I did recently, I looked at how recently distributions had their last release, and how that varied with the river position (number of dependent distributions). I've been playing with some other ideas related to this, to see if they'll help identify distributions that could do with some TLC.
This was my first thought: a scatterplot of river position against the number of months since the latest release.
The x axis is log10 of the total number of dependents (direct and indirect).
I excluded core modules (since they're released at least annually, by definition), and modules with no dependents. The plot now only shows dists with at least one bug or CPAN Testers fail.
This is generated with d3, and has a tooltip on the dots, to give information about the distribution.
As you can see, there are some clear outliers. But there are some distributions that are "finished", and don't need regular releases. I'm thinking I'll update this to only display distributions with notable numbers of CPAN Testers fails, or outstanding bugs.
Other thoughts I've had:
In the meantime, it's prompted me to (a) fix some outstanding bugs in distributions that I maintain, and (b) to look for some low-risk adoption candidates. This also identified a bug in my CPAN trawler, flagged by Dave Rolsky.
comments powered by Disqus