[Intel-gfx] IGT contributions and reviews

[Daniel Vetter]

On Wed, Oct 19, 2016 at 3:06 PM, Paulo Zanoni <paulo.r.zanoni at intel.com> wrote:
>> > > At the very least I would like to see all commits have a visit to
>> > > the
>> > > mailing list before pushing, as the current docs already ask for.
>> > > The
>> > > "right after" part would be changed to a $period of quarantine,
>> > > maybe
>> > > 24 hours?
>> >
>> > Sounds good to me.
>>
>> We've already had this, and people stopped bothering. What will be
>> different this time around? I feel a bit like we do need to be a bit
>> more
>> formal here, to really make this stick ...
>
> I think the problem is that if only one single person doesn't follow
> the rules, a few others will notice and start thinking: "If this guy
> doesn't need to follow the rules, why do I need to? Why is Daniel
> asking me to submit patches to the mailing list if that guy doesn't
> seem to need to follow the same rule? Is this some sort of double-
> standard?". Then person #2 stops following the rule, which makes more
> people realize the same thing, which makes person #3 and #4 stop, etc.
> So please, whatever rule we decide, let's make sure *everybody has to
> follow it*, no exception for special cases, no allowing-people-to-get-
> away-with-it. No double standards. </rant>
>
> As a consequence of what I said, we do need to think: what if someone
> doesn't follow the rule we decide? What will we *actually* do? Will it
> really work if all we do is to politely ask them on IRC? Maybe the lack
> of consequences is that degraded our previous rule?

Yeah I think that was kinda what happened. At first I did gently prod
people who bent the rules a bit into what I think would be the better
direction. But pretty much from the start I didn't want to be too much
of an annoyance, because when I put my maintainer dictator hammer down
and said I want to see testcases I think a few people were very
sceptical. Hence I was really happy with every engineer who discovered
that igts are useful and started hacking on them like mad, even if the
process went out of the window a bit.

But I do think that now igt and tests as a primary deliverable is
well-established, and we seem to have a few issues from the lack of
any kind of review or otherwise coordination:
- Duplicated tests, and sometimes duplicated test infrastructure that
probably should be better extracted into helper libraries.
- Much harder to ramp up new folks since they can't go watch the pros
do their work on the m-l like on the kernel side.
- Defacto double-standards pissing off contributors.
- Lack of sharing of useful testing tricks. E.g. I only learned by
accident of Chris' very clever delay-batch trick, and I think that's
something which should be used in a lot more places.
- And finally I think we have a bit a quality problem with tests
itself ("works on my machine, it must be perfect") which is causing
some serious hiccups and fun with CI.

Would love to here the thoughts of others here.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch