Re: git status update?

[Owen Taylor <otaylor redhat com>]

On Tue, 2009-03-03 at 20:39 +0100, Claude Paroz wrote:
> Le mardi 03 mars 2009 à 13:42 -0500, Owen Taylor a écrit :
> > On Tue, 2009-03-03 at 13:08 -0500, Owen Taylor wrote:
> > 
> > > > I just wonder about a pre-commit hook which I think checks the po files
> > > > validity (msgfmt check). IMHO It should also be rather easy to rewrite
> > > > for git. 
> > > 
> > > Looks easy to rewrite, yes.
> > 
> > Any advise and insight into the code in that hook.
> > 
> >         # The Persian (fa) team is using localized numeric formats
> >         # in their .po files, but gettext on cvs.gnome.org isn't
> >         # new enough to handle them.
> > #       DASH_C=""
> > #       BASENAME=`basename 
> > #       if [ "$BASENAME" != "fa.po" ]; then
> > #               DASH_C="-c"
> > #       fi
> > 
> > (we used to pass in -c for everything but fa.po, and now we pass it 
> > in for nothing)
>
> Are they other check options passed (--check-format, --check-header,
> --check-domain)?

No.

> > Would be appreciated.
> > 
> > mgfmt on git.gnome.org (RHEL 5) is version 0.14.6
> 
> Well, this is indeed an outdated version. e.g. msgctxt support has been
> introduced in 0.15. Would it be possible to build a more recent version?

When that comment was written, cvs.gnome.org probably was running the
RHEL 4 version, so two years or so older than that. :-)

I'm hesitant to update such a core package as gettext away from the
system version. So, a fix would probably require creating a package of
gettext that installs in a non-system location (we're trying with newer
systems like to git.gnome.org to have all software reproducibly
package.) Definitely possible but outside the scope of the git
migration.

> With a more recent version, the -c check could be reintroduced and
> general po quality improved.

So, the conclusion is that the "localized numeric formats" is no longer
a significant issue? The GNOME translation team is using other gettext
features that prevent the use of -c (as of 0.14.6) in any case?

- Owen