Re: Translating schema files (was: Re: Survey results (yay!))

[Khaled Hosny <khaledhosny eglug org>]

On Mon, Nov 01, 2010 at 08:26:12PM +0100, Ask Hjorth Larsen wrote:
> On Mon, Nov 1, 2010 at 7:32 PM, Gil Forcada <gforcada gnome org> wrote:
> > El dl 01 de 11 de 2010 a les 18:35 +0100, en/na Jorge González González
> > va escriure:
> >> El lun, 01-11-2010 a las 19:29 +0200, Ihar Hrachyshka escribió:
> >>
> >> > You're wrong when speaking about "normal user". Normal user:
> >> > 1) either use Google or NOT for searching problem solutions;
> >> > 2) should be able to fix the problem with no external help (google,
> >> > friends, tech support). If not, then the software is wrong, the error
> >> > messages are wrong, but the errors localisation is ok.
> >> Well, obviously we don't deal with the same "normal users".
> >>
> >> >
> >> > So though I'm for separating schema messages, I'm against separating
> >> > errors from UI messages.
> >> I'm against separating them since it will include more files to push and
> >> handle, but I'm against translating them at all.
> >
> > My first idea was to split to different files, but Simo's approach was
> > about giving D-L the ability to show different filtered files from the
> > same file (and only one po file in git).
> >
> > So you then would be able to either translate everything (errors,
> > schemas and "ui") or just some parts (first only the ui, then the errors
> > and finishing with the schemas).
> >
> > That way keeping only a single file for each module but showing them as
> > a separate files small teams and new contributors would feel that their
> > translations make sense from the very beginning.
> >
> > Then the question is how to create this filters and implementing them in
> > D-L.
> >
> > Anyone is up to this?
> >
> > Cheers,
> 
> But how would you reliably distinguish between error messages and
> non-error messages?  I'm not convinced that this can be done without
> requiring unreasonable work from programmers.

I think a "translators comment" with a certain magic word would suffice,
it need not be done over a day and night, then some tool (can be
translate toolkit based) would extract such strings besides schema
strings from the PO file. This should not be different from asking for
msgctx for ambiguous strings that we do now.

We can even try to make identifying schema strings more robust by making
intltool insert similar comments when extracting translations from
schema files.

Regards,
 Khaled

-- 
 Khaled Hosny
 Arabic localiser and member of Arabeyes.org team
 Free font developer