Re: who is doing patch-review -> patch-countdown

[David Kastrup]

James <address@hidden> writes:

> On 19 February 2013 22:25, David Kastrup <address@hidden> wrote:
>     
>     Our rule basically is "if you bypass the proper review process,
>     you deserve everything you get if there is any problem".
>
> I'm not sure how Colin decided this kind of case

The thing is simple: it's not in your job description as a patch meister
to decide this kind of case, namely make decisions based on technical
expertise or getting into arguments with developers.  In a manner,
Graham is right that this is an administrative job.  Your look is to
take a look at the review and discussion, judge whether the discussion
about possibly unresolved or unaddressed points has settled and/or the
issue has been sitting on review long enough in relation to its
superficial complexity/size/type, and then shift status accordingly.

If a developer feels that the nature of his patch requires a different
treatment, that's outside of your responsibility and he should not try
coaxing you into serving as a scapegoat when something does go wrong.

Your job is not judging the quality of patches but rather ensuring
proper time windows in which other developers may assess patches and
react accordingly.

So there is no point in asking you to make exceptions.  This is clearly
"bypassing the proper review process" area.  If people want to do that,
that's on their own head with the above-mentioned consequences.

> Some patches are easy to see what is 'trivial' (for me) some are not
> (for me). So I defer to any dev that wants to contest my decisions on
> patch countdown in this kind of case, I guess as long as there is a
> comment in the tracker why the patch has been bumped up everyone else
> can see and allow it or not.

Personally, I don't see that people should in general be setting
"Patch-countdown" on their own.  There may be some excusable case when
you stated "this would have been on countdown in case the trivial points
x and y had been addressed" and we are early in the countdown phase.
But maybe it is better in this case if you state "Patch xxx is
tentatively on countdown but will not move to patch-push unless the
trivial points x and y have been addressed".  That way your bookkeeping
reflects the actual flow of state.

Addressing a trivial point can initiate a fast cycle Patch-countdown
(set by patch submission to) Patch-new (set by test-patchy to)
Patch-review (set by patch submitter manually back to) Patch-countdown.

Those fast cycles can be initiated by patch submitters even during a
regular countdown (improving details, but not redesigning parts, should
be permissible before pushing), and if you have set a conditional
countdown, that's the only way to avoid another one.

That's all details in execution of the patch meister job.

But asking you for your blessings for bypassing the system isn't.

-- 
David Kastrup