Em qua., 10 de jun. de 2020 às 06:13, Jonas Hahnfeld <
hahnjo@xxxxxxxxx> escreveu:
Am Dienstag, den 09.06.2020, 23:51 +0200 schrieb Federico Bruni:
> On Mon, 8 Jun, 2020 at 16:16, Caio Barros <caio.barros@xxxxxxxxx> wrote:
> > [...]
> > 5. opening a merge request, changing the destination branch from
> > lilypond/lilypond:master
> > to
> > lilypond/lilypond:translation
> >
> > I found steps 2 and 5 particularly non-intuitive, so if this worflow
> > is correct, I propose we change the contributor's manual to explain
> > them.
>
> [...]
>
> 5. This is a valid point, but I think that Jonas (here in Cc) did not
> take translation into account, as most translators have write access
> and push their work without asking for a review (I think that active
> languages have only one maintainer).
Yes, I've seen only few translation patches go through the usual review
process since I started working on LilyPond last year. IIRC all of them
targeted master and were not pushed to translation first.
In that case, do you think it is useful to change the documentation to add these steps to commit to the translation branch? From what I understand, this would mean to change the workflow of virtually all translators except (maybe) new ones. Perhaps we can just carry on pushing to master and add a label to translation MRs.
Actually, because the project moved to GitLab a lot of sections in the CG are outdated. Is there anyone in charge of updating the CG? Since I'm fairly new in contributing I've been noticing some things that got me confused there, maybe I can suggest some modifications.
> Don't know if GitLab may be able to understand that a MR is created
> from the translation branch and therefore change the target branch to
> translation...
As far as I know, GitLab will automatically target the default branch.
This is 'master', and that choice is probably correct for 99% of the
cases.
Yes, when I open the merge request URL it automatically sets the merge request to master. I have chose to change MR !141 to the translation branch because of the section on CG on git [1].
That said, if we want to use merge request for translations in the
future, we can surely figure out improvements for that workflow. Most
importantly, the merge request !141 cannot currently be merged because
it is missing automated testing. I think having some dummy check
suffices here, the "real" testing takes place when merging translation
to master.
Well, the merge was approved by Werner Lemberg. Should I send a new merge request to master? I don't understand.
Caio