Re: 3.0 branch: POTFILES.in broken

[Dennis Bjorklund <db at zigo dot dhs dot org>]

On Tue, 12 Jun 2001, Joseph S. Myers wrote:

>                 status_warning (status, "%s ignored with %s in %s format",
>                                 _(s->name), _(t->name), fki->name);
>
> The first two names here are such things as "precision" or "length
> modifier" (the third is something like "printf" to which i18n doesn't
> apply).  If the translations of these words get declined differently in
> different places in the sentence or in different warning messages, there
> would be the problem of providing a single translation of "precision" that
> works in all places.

The normal solution to that is to identify these places where the same
english word needs to be translated to different foreign words depending
on the context, and replace that word with something like
2+_("A:precision") and in the other place 2+_("B:precision") and some
comment just above that automatically goes in to the po file so that the
translator knows that the A: part should be left in the translated string,
and what the semantics of the word is in this context, so he knows what to
translate for.

This technique can be difficult since the gcc error function hack works in
a different way.

Translations is also something that needs fine tuning and bug reports.
Even more so in programs like gcc where it is important with a uniform
language and exact concepts, and where there is a lot of technical terms.

People seems to forget that sometimes.

One thing that is needed once there are translations is to have a string
freeze some time before the release, so that the translations have a
chance to get up to date.

-- 
/Dennis