[Osmf-talk] Seeking feedback and interest in the OSMF Engineering Working Group

Andy Allan gravitystorm at gmail.com
Thu Jul 15 13:58:20 UTC 2021


On Wed, 14 Jul 2021 at 18:55, Michal Migurski <mike at teczno.com> wrote:
>
>  The bars in blue are composed solely of commits from core committers and a handful of bots vs. PRs in green with any other participants included. We see that community PRs are accepted approximately half the time, while owner-only PRs are merged almost always. We also see a long-standing pattern of community changes taking multiple weeks to merge.

I find these results deeply unsurprising. You've highlighted that PRs
from bots (i.e. dependabot) are quick and easy to merge. What else
would you expect here? You've also highlighted that PRs made by
maintainers are usually merged quickly. So again, is it really
surprising that maintainers are good at writing PRs that pass the
tests, are easy to review, and likely to be merged without need for
changes? Isn't that likely to be one of the abilities expected of a
maintainer?

That community changes can take multiple weeks to merge is also not
surprising. By ignoring the "changes requested" label, any review
status on the PR, and whether PRs need follow-up commits added later
(or rewritten wholesale into a new PR) before being merged, I don't
know what conclusions you can draw from these statistics. There's no
expectation that community contributors must be able to write PRs that
pass the tests, are easy to review and likely to be merged without
need for changes. That would be an unreasonable expectation for an
open community, as opposed to a corporate environment. So naturally
it's expected that people will make PRs and need help to get them
merged, and that PRs will sit in the "waiting for changes" state for a
while until they have some free time to do so. I also don't expect our
volunteers to immediately have time to amend their PR after review,
but your analysis seems to expect that all community PRs are
immediately eligible for merging.

> My conclusion from these graphs is that OSM’s software is not encouraging a healthy level of community involvement.

That's a heck of a conclusion to make from two poorly constructed charts.

Perhaps in future, before attempting to make sweeping conclusions on
such flawed analysis, you could give the maintainers of the project a
chance to discuss with you your methodology? That way we can get some
more useful analysis and ideally some actionable conclusions. Because
I put in a ton of effort, week in and week out, year after year to
make it easier to contribute, and I'm happy to hear how to improve the
situation further, but posts like this just don't help.

Thanks,
Andy



More information about the osmf-talk mailing list