[python-committers] number of active core devs [was: Comments on moving issues to GitHub]
ncoghlan at gmail.com
Sun Jun 3 01:44:01 EDT 2018
On 3 June 2018 at 11:07, Guido van Rossum <guido at python.org> wrote:
> Sounds to me like these are probably just past committers who are no
> longer active for whatever personal reasons, and took no action when we
> moved to GitHub. We basically never remove the commit bit from anyone
> except by request, and I only recall seeing one such request, ever. Some of
> them probably expect to come back in the future (like Neil Schemenauer
> did). I recall only one person who said they refused to move to GitHub (but
> AFAIK we didn't remove their commit bit from b.p.o), so I don't think that
> we can blame these numbers on the move to GitHub.
OpenHub  shows the average rate of commits declining fairly steadily
since the exceptional ~40-commits-per-day spike in September 2016 down to
our current steady state of ~4 commits per day (we still get spikes up to
10+ commits per day for PyCon US and the core dev sprints, but not of the
magnitude of previous sprints). Those metrics only record the actual commit
rate (not the code churn rate), so some of that may be due to the switch to
a PR based workflow with pre-merge CI reducing the volume of fix-up
commits, and I also don't know how the switch from our
patch-and-merge-forward workflow in Mercurial to the
squash-merge-and-cherry-pick workflow in git affects the accounting.
While the switch to GitHub does show up clearly in the "contributor" stats
on OpenHub, the move to git is also when the VCS metadata started recording
the committer and author information separately in a way that OpenHub can
read (rather than only providing the patch author information in the commit
message and NEWS entry), so someone would need to go back and extract the
real pre-git contributor metrics to make that a valid comparison.
On the issue management & patch review side of things, while
https://bugs.python.org/issue?@template=stats does show the number of open
issues with patches declining slightly post-migration, it's since leveled
off and then started climbing again.
So based on the numbers we're seeing, my own assessment would be that the
move to GitHub didn't hurt, but it also didn't really help address the
review bottleneck problem either (which surprises me as much as it does
anyone else - perhaps now that patch reviews are more pleasant to engage
in, we're also making them more thorough?).
Nick Coghlan | ncoghlan at gmail.com | Brisbane, Australia
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the python-committers