This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Free Translation Project
- To: pthomas at suse dot de
- Subject: Re: Free Translation Project
- From: "Martin v. Loewis" <martin at loewis dot home dot cs dot tu-berlin dot de>
- Date: Sun, 7 May 2000 22:57:50 +0200
- CC: tromey at cygnus dot com, gcc at gcc dot gnu dot org
- References: <87em7ep977.fsf@cygnus.com> <20000507203040.C24972@Jeffreys.suse.de>
> Currently there is no way to do that, as POTFILES.in lists all sources
> that contain translatable messages. And even if this would be changed,
> on what part of the message catalog should translations be based?
How serious is this problem? I.e. would it be acceptable to have the
catalog generated and installed for the complete source code, even if
only part of the compiler gets installed? If not, why not?
> The easiest solution would be to have seperate catalogs for each
> language, but AFAIK, you can't use multiple catalogs simultaneously
> (sp?).
Sure you can. You only need to have multiple domains, and then access
them with dgettext. However, I think this won't be needed.
> - how do we support i18n when cross-compiling?
A plain cross-compiler would operate just as a native compiler in this
respect; it is a native program, after all. In a canadian cross
compiler, one could either use the included libintl unconditionally,
or try to automagically detect support for libintl on the host system,
which probably a cross-msgfmt would be needed. As another alternative,
not supporting i18n if host!=build seems acceptable to me, as well.
> - what version of gcc.pot should translators start from?
Just start with today's version. Translators will need some months to
complete initial translations. When the release date gets closer, send
updated versions more frequently, so the translators can adjust the
messages that have changed.
> - how do we provide a way for the translation robot (or Francois Pinard)
> to fetch the most recent version of gcc.pot?
I'd say that the maintainer for that should decide on the exact
version that gets submitted. IMO, it is important
a) to have a clear identification in each version of the catalog, so
that everybody knows what version of the catalog has been translated
b) to produce a small number of catalogs only, so that translators do
not get overwhelmed with quickly out-dated catalogs. In particular,
there is no point in feeding weekly catalogs to the robot.
As for the specific procedure, I believe the robot will accept
catalogs from an authorized source once configured. I'm sure François
will be helpful in arranging things.
> I volunteer to act as maintainer for the i18n specific stuff and in
> the end get gcc listed, but we have to do our part first, which
> includes the points mentioned above.
I'm not sure there are much points left, so IMO, translations could
start tomorrow.
Regards,
Martin