Yesterday, Adam Kennedy announced his latest CPAN Top 100 list: the “Fail 100”. It ranks CPAN distributions by the product of dependent count and FAIL count from CPAN Testers to give a list of distributions at greatest risk of blocking other distributions.
The two measures are correlated, of course, since distributions that other things depend on get tested a lot and a distribution needs to get tested a lot to rack up a big FAIL count. Still, it’s an interesting measure to try to identify critical paths in the CPAN.
I can only wonder (partly in jest) what might be next:
- Buggy 100 – weighting by number of tickets on RT or other bug tracker (ignoring “unimportant” or “wishlist” tickets)
- Stale 100 – weighting by time since last released
- Workload 100 – weighting by total number of FAILS and bug tickets, but divided by number of maintainers and co-maintainers
We’ll have to wait and see…