[Mesa-dev] GitLab merge requests, discussion summary - Re: [PATCH] docs: Document optional GitLab code review process

[Brian Paul]

On 12/03/2018 06:50 PM, Jordan Justen wrote:
> Lots of discussion here. :) I hoped to just dip our toes into the
> merge request stream, but the consensus is clearly against my idea. Oh
> well. :)
> 
> Here's what I gathered from the discussion:
> 
> == Keep email, allow duplicate MR, my idea ==
> 
> I don't think anyone else was in favor of this. Several were against
> it.
> 
> == NACK merge requests altogether ==
> 
> I did not see this, so that seems promising for merge requests. :) I
> guess no one strongly disagrees with trying merge requests.
> 
> == Allow submitter (or driver) to choose to email or MR, but not both ==
> 
> Several seemed to argue that this could be a good next step. But, in
> all those cases, I think they wanted to use merge requests going
> forward.
> 
> It was mentioned that vc4 has moved to github in order to allow github
> pull requests to be used.
> 
> == Require all email list, or all merge requests ==
> 
> A few argued in favor of this, and seemed to feel pretty strongly
> about it. In both cases I believe they wanted to more to merge
> requests. :)
> 
> == Next step? ==
> 
> So, what do you all think is the best next step after the discussion?
> 
> We could take the smaller step to allow the submitter, or project to
> decide which they prefer. It would allow those who prefer email to
> continue with email, while trying out gitlab merge requests. But, it
> could mean that devs need to check both gitlab and email.
> 
> Or, we could just try to make the jump to MR only. Maybe if I write
> that patch someone will finally stand up to try to NACK moving away
> from email. :)
> 
> Or, are we a little too early for this discussion, and we should just
> drop it for now?
> 
> My opinion would be to try to move a bit slower and allow the
> submitter/project to choose for a trial period. (Dylan and Daniel seem
> to think this could be really frustrating for devs though.) But, if
> everyone seems to think it's reasonable to try to jump straight to
> using merge requests exclusively, I can type that version up...

FWIW, I'd prefer the trial period where both methods are available.

-Brian