Re: Making it easier on translators

[Pablo Saratxaga <pablo mandrakesoft com>]

Kaixo!

On Mon, Nov 20, 2000 at 09:23:11PM +0100, Keld Jørn Simonsen wrote:

> This could also be made a lot simpler. Pablo from MandrakeSoft has done
> some work along what you have outlined above. Maybe you could use
> some of his setups. It includes:
> 
> - updating all the .po files, so you do not have to do this yourself

Yes, each time I get a new file (or simply, regularly) I launch some scripts
that:
- check that the pot file is up-to-date, rebuild it otherwise
- check that all the po files are up-to-date regarding the pot file;
  call msgmerge otherwise (in fact I do it in rather poor way: I do the
  msgmerge to a tmp file; do a diff, and if the diff is not null keep the
  new one, otherwise delete it)
- checks all po files (msgfmt -v -c) for possible erorrs

Maybe some translators don't like that their po file could chnage on the
sources while they are still working on it.
But I decided that for all sources I have the responsability of the
coordination of the translations it would be better to keep always the files
up to date. If a translator finds that the po file on CVS changed when doing
a cvs commit, that means he has taken too long, and that a re-reading would be
needed anyway.

> - this also removes the need to download a lot form the CVS to build the
>   updated po files. And you do not need a lot of HD to store the cvs.

Well, all the above I do it for my own needs.

What helps the translators is that they can get the files from the status
page; and a script I launch that for all po fiels needed an update it sends
it to them (or send simply a mail telling about it).

> - I think he also merges  new stuff with a translation database per language.

No, I usually don't do that. When I do it it is "by hand".
I did it on some cases when I thought that it could help to retrieve several
strings from po files of other modules.

> > * Create an automated system to generate the .po files for every
> > language, that send a letter to translator when a certain number of
> > fuzzy or undone translation are found in the file (you should be able to
> > conf this)
> 
> yes (Pablo has something there)

But it is a rather crude bash script; for my needs it is enough; but maybe
more flexibility would be needed for a more general usage.

>> * The system should be able to mange branches and make it transparent to
>> translators.
> 
> That would be good.

A way to achieve that would be to *mandate* the use of specified strings
in the Project-Id-Version: field; so, when storing the po files ona web page
or when sending to the user, the name can be get from there;
and then we will have "gtk-es.po" and "gtk-branch-es.po" (or whatever naming
is used; the importance is that it keeps consistent) 

That is how I handle it for the translators for whom I act as an intermediary
for Gnome sttuf, so they get two po files with different names: that doesn't
trouble them, and there is no risk of overwritting files or getting them
inversed.

-- 
Ki ça vos våye bén,
Pablo Saratxaga

http://www.srtxg.easynet.be/		PGP Key available, key ID: 0x8F0E4975