-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 29/09/15 16:30, Andreas Jaeger wrote:
On 2015-09-29 07:20, Alexandra Settle wrote:
Hi everyone,
I�ve been noticing in the last few months (and I�m sure it�s been a
problem before) that we have contributors submitting bug reports, and
then immediately fixing before triage, or there are patches without
bugs (or blueprints) and it�s becoming quite confusing.
I am finding that this is a problem because some of these bugs are
personal issues they�re finding with their individual builds, or on
occasion it is a bug that needs to be fixed and individuals are offer
ing
the wrong solutions and immediately trying to patch it up in the
documentation.
Personally, I�m unsure how to solve this issue, and wondering if we c
an
get a think tank going on how to solve this.
So far all I�ve been able to do is to remind contributors to wait for
a
second individual to triage, or the patch is eventually abandoned due
to
several negative reviews (which, ultimately encourages said contribut
or
not to come back because we look like a very negative community).
We can do more �bug triaging days�, but not everyone is to give the t
ime
to this exercise.
Thoughts? Suggestions? Comments? Questions?
This is not a simple topic, thanks for raising it!
Some OpenStack projects encourage this behaviour - if you send a patch
without a bug, they often ask for a bug that you create and assign
directly to you...
Normally, this works, and we can trust people to use their best
judgement. I certainly don't want to enforce this for everyone, because
finding a typo or something trivial really doesn't need this level of
process. Recently, though, there's been a flurry of people finding and
'fixing' things that are actually bugs, and I've had a few different
core team members point it out to me that we need to try and remind
people about triage for things like that.
Also, we have Documentation like
http://docs.openstack.org/infra/manual/developers.html#working-on-bug
s
https://wiki.openstack.org/wiki/BugTriage
Neither asks for an independent review.
No, and neither should they, I think. I actually suspect the better way
to handle this is for our core team to be ever-vigilant, and make sure
we're checking for independent review on patches that require it. Most
of the culprits aren't regular committers, anyway, and I don't want to
alienate any of our regular group.
I suspect this is a people problem, which we shouldn't try and fix
through extra processes, but by conversation and education.
I acknowledge the problems we have and I think this might be a broader
topic to discuss - not on this limited list but perhaps together with
developers on openstack-dev? Or in some cross-project meeting?
Let's bring it up at Summit. I've applied for a cross-project session,
so if that gets approved, it might be a good place.
