Re: [jdt-dev] what exactly does the PR job on jenkins build?

[Christoph Läubrich <laeubi@xxxxxxxxxxxxxx>]

> What is the advantage of that? I see only disadvantage.

As with the other topic, there might be different views. So saying there 
are "only disadvantages" might be claimed for nay of the solution 
depending on who you ask.

For example you have some advantages/disadvantage for the one strategy 
over the other:

1) The previous strategy triggered a rebuild of PRs on each and every 
change on master, especially if there are a lot of (long standing) PRs 
this considerably can block the build pipeline.

2) If there is a conflict you can't build the PR and need to 
rebase/resolve/... first

3) Maybe more Pro/Cons


So as always it depends specially on the time a PR is lingering around, 
I personally don't see any problem in keeping a PR on the "old" master 
for the time of review (with many comments, commits, ...) and if we are 
done squash and rebase and merge afterwards after the "final" build has 
passed, the comments will still be retained in the PR history (with 
original code).

Am 18.06.24 um 22:03 schrieb Stephan Herrmann via jdt-dev:
> Am 18.06.24 um 21:30 schrieb Jörg Kubitz via jdt-dev:
> > Its a configuration that was changed by Andrey ~1 year ago, because he
> > wanted to see a tests on what exactly was submited.
>
> What is the advantage of that? I see only disadvantage.
>
> I assume every contributor tests locally exactly what he/she is going to 
> submit. Afterwards I want to know how the change will behave in master. 
> What is wrong with that?
>
> > As long as you
> > rebase to head - as we usually do - there is no rebase anyway.
>
> I knew someone would say that :)
>
> I know that people keep preaching this, but such repetition doesn't 
> improve github's support to work with review histories across force 
> push, which is lousy if you ask me.
>
> But let's simply agree to disagree in this regard. I will continue to 
> require no-force-push on a PR while I'm reviewing, and let you do 
> whatever you think good. Let's keep the truce.
>
>
> Perhaps the strategy for PR builds can still be discussed regardless of 
> the issue of force-push?
>
> thanks,
> Stephan
>
> > Am 18.06.2024 um 21:03 schrieb Stephan Herrmann via jdt-dev:
> > > Some time I ago I checked that each PR build actually builds the
> > > result of locally merging the PR with master (or whatever target 
> > > branch).
> > >
> > > Re-checking today, it looks like the HEAD of the PR is built directly.
> > > This would imply that a successful PR build does not inform us about
> > > the result after the PR is eventually merged, which looks risky to me.
> > >
> > > Has any such change been applied intentionally?
> >
> > _______________________________________________
> > jdt-dev mailing list
> > jdt-dev@xxxxxxxxxxx
> > To unsubscribe from this list, visit 
> > https://www.eclipse.org/mailman/listinfo/jdt-dev
>
> _______________________________________________
> jdt-dev mailing list
> jdt-dev@xxxxxxxxxxx
> To unsubscribe from this list, visit 
> https://www.eclipse.org/mailman/listinfo/jdt-dev