Re: msgctxt (was Re: String changes in Evolution)

[Petr Kovar <pmkovar gnome org>]

"Christian Rose" <menthos gnome org>, Tue, 5 Aug 2008 03:07:03 +0200:

> (Edited the subject from
> http://mail.gnome.org/archives/gnome-i18n/2008-August/msg00036.html in
> order to make it a little more obvious)

At first, I didn't mean to continue in discussion, but there are still some
open questions however.

(...)

> I'm personally all for suggesting the msgctxt migration as an official
> GNOME goal for GNOME 2.26
> (http://live.gnome.org/GnomeGoals/MsgctxtMigration). As Andre pointed
> out, the behavior of the current Q_ macro in glib is error prone to
> translators and causing a lot of problems and real bugs.

Sure, again I think nobody's been objecting to the need of migration itself
after all. 

> We cannot wait for a particular po editor to have support for msgctxt.

Well honestly, I'd say we actually can, but perhaps we just don't want
to. ;-)

> There are plenty of tools and if one doesn't do the job there are
> plenty of alternatives, including using a bare bones text editor if
> necessary.

Yeah, but using (or suggesting people to use) a plain text editor would be
a last resort really. 

> The more important thing IMHO is the toolkits (glib, gettext,
> intltool, pootle, translate toolkit, Damned Lies, Launchpad, etc). All
> of these seem fine, however.

If not, what is the qualitative difference between impossibility of using
a gettext catalog editor, and impossibility of using a "toolkit" like pootle
or translate toolkit? In both cases, user is forced to make a switch
anyway. When doing so, one has to migrate his personal settings including
appropriate translation memory files, one has to adapt to the new program's
environment, behavior, features, etc. It's just not as simple as saying,
"go and change your tool".

And first and foremost, translators really should be aware of GNOME
modules being migrated to msgctxt. For our Czech translators I'll make sure
that they get notified in time, but who is going to warn people in the "l10n
wild" that Poedit doesn't provide the context information (anymore)?
Especially when everything is still just a goal proposal. Another
error-prone l10n problem on the horizon? I hope not.

(Also note that as far as I know, Poedit is the only GTK+ PO editor with a
usable almost bug-free stable/development release for the time being.)

Regards,
Petr Kovar