[openstack-dev] [TripleO] Set WIP for stale patches?

Jeremy Stanley fungi at yuggoth.org
Fri Sep 19 16:25:10 UTC 2014


On 2014-09-19 15:27:36 +0000 (+0000), Sullivan, Jon Paul wrote:
[...]
> I think that the abandoning happening from an automated process is
> easier to accept than if it came from a person, and so less likely
> to create a poor and emotional response.

Here we'll just have to agree to disagree. I think core reviewers
hiding behind an automated process so that they don't have to
confront contributors about stalled/inadequate changes is inherently
less friendly. Clearly you feel that contributors are less likely to
be offended if a machine tells them they need to revisit a change
because it's impersonal and therefore without perceived human
judgement.

> If your personal opinion was that it wasn’t useful to your
> project, then perhaps what you are really saying is that the
> implementation of it was not configurable enough to allow
> individual projects to tailor it to their needs.
[...]

Sure. For what it's worth, I haven't said I would push back on
someone writing a reasonable implementation of this feature, but it
definitely is something I wouldn't want imposed on everyone's
workflow just because the majority of core reviewers on some subset
of projects found it easier to have changes abandoned for them.

> So the removal of the auto-abandon, imho, has increased core
> reviewer workload, increased the chance that a good change may get
> ignored for extended periods of time, and has increased the
> possibility of code committers becoming frustrated with core
> reviewers adding a wip or abandon to their patches, so a decrease
> in productivity all around. :(

I think this is an inaccurate representation of the situation. It
wasn't explicitly removed. It was a buggy hack which was effectively
unmaintainable, didn't work with modern versions of Gerrit, and
nobody felt like investing time in a new implementation of it nor
was it deemed a critical feature for which we should hold back
progress and continue to fester on a years-old Gerrit installation.
It broke and has never been fixed.

This is not the same thing as being removed, but since it's been
gone I've come to wish we had removed it rather than just living
with it until it ceased working due to bitrot. In this case, an
automated process determined that feature was no longer suitable and
abandoned functional use of it. I think it would have been more
friendly to our community if the people who were no longer
interested in maintaining that feature had explicitly removed it
instead... but then it sounds like you would assert that having the
machine abandon this feature for us was less likely to offend
anyone. ;)
-- 
Jeremy Stanley



More information about the OpenStack-dev mailing list