[CRIU] github pull-requests

[Andrei Vagin (C)]

On Tue, Nov 20, 2018 at 05:31:23PM +0000, Pavel Emelianov wrote:
> >>> https://github.com/google/pull-request-mailer
> >>> 6. Do we need a robot which will create github pull-requests from patch series?
> >>> 7. Should we sync comments between the mailing list and github?
> >> I see e-mail notifications coming from github on PRs. I've never tried to
> >> respond on any of it :) Isn't it supposed to work "out of the box"?
> > I also see these email notifications and yes replies to these notifications
> > will be displayed in GitHub:
> > 
> > https://blog.github.com/2011-03-10-reply-to-comments-from-email/
> 
> That's good news, as the item 7 is half-done :D
> 
> >>
> >>> 8. Who wants to implement 5., 6. and 7.?
> >>>
> >>> My answers:
> >>> 1. yes
> >>> 2. no
> >>> 3. yes
> >> Doing both looks like a challenging experiment :)
> > In the virt-manager project most of the discussion and patches are being
> > sent to the (virt-tools) mailing list
> > but pull requests are also accepted via GitHub as well.
> > 
> > The way Cole Robinson handles this is by closing (rather than merging)
> > the GitHub pull requests and pushing
> > the commits manually. It seems to work OK for now and I think that, if
> > we want, we could do the same for CRIU.
> 
> I think that would be an overkill for CRIU, if the PR merges well, it can be just
> merged. Two things bother me in this scheme:
> 
> 1. Will PRs get validated the same way as patches-over-ML are?

yes, they will.

> 2. How to attract the attention of people preferring (and sitting in) the
>    mailing lists to the submitted PRs for review?

It is a good question. It is why I think we need to support both ways
for submitting patches and why we need to send PR-s to the mailing list.

> 
> Apparently doing items 5 through 7 will solve this, but I don't know how
> complex this task will be.
> 
> BTW, even if we decide to stop playing with ML and switching to PRs the validation
> question would still stand. Andrey, can you shed some light on this?

We use only Travis to validate patches, so there is no problem at all.

> 
> -- Pavel