On Tue, May 17, 2022 at 12:19:07PM +0200, Jiri Denemark wrote:
On Tue, May 17, 2022 at 10:46:21 +0100, Daniel P. Berrangé wrote:
> On Tue, May 17, 2022 at 02:34:00AM -0700, Andrea Bolognani wrote:
> > On Tue, May 17, 2022 at 11:11:00AM +0200, Peter Krempa wrote:
> > > Hmm, so is this something based perhaps on a different way how Jirka
> > > generates these? I'm asking because I'm fairly certain that Jirka
has a
> > > script for this so that needs to be fixed too.
> >
> > No, you can just call the standard 'ninja libvirt-pot' target
> > generated by meson and reproduce this - only sometimes, of course ;)
> >
> > I assume Jirka's script uses the existing functionality rather than
> > reimplementing it.
>
> The pot files should just be imported from the pristine potfile
> published by our CI job 'potfile' which is AlmaLinux8 based.
>
> If that's generating something bad, we need to address that.
Right, I just download the generated file from
https://gitlab.com/libvirt/libvirt/-/jobs/artifacts/master/download?job=p...
and push it to git. So updating the file manually without fixing the
generator makes little sense as the file will be overwritten in
preparation for the following release.
You could run 'ninja libvirt-pot' locally and avoid the round-trip to
GitLab, but whatever works for you :)
Alternative approach: we could take the options coming from the glib
preset, minus --add-comments, and pass them ourselves. Bit ugly, but
would guarantee that we get the correct file generated every single
time. And we can go back to using the preset once we are in a
position where we can rely on meson >= 0.60 being used. Sounds good?
--
Andrea Bolognani / Red Hat / Virtualization