Collaborative Translation of XBMC Language Files on Launchpad or similar online tool?

Any news?

Hi,

Is Launchpad translation active? or When activated?

Not yet, will be soon. Just have some minor tweeks to sort out.

I will also set up a translationservice on launchpad for scripts/plugins/skins aswell.

Hello,

Today I looked at how translations are done in XBMC.
Correct me if wrong, but right now we need to:
-Download language files
-Change them with translation tool
-Upload them on track.

Am I the only one who finds this a crappy process?
I mean, there has to be a way to make this easier.

I was thinking of a web interface.
That way you give everybody authority to be working on the same file, maybe with a voting system when there are multiple translation possibilities.
There could be authorized translators who's changes go directly to SVN. New translators need to be reviewed by authorized translators.

It would be totally awesome if this then could be integrated with svn and if add-ons could be translated the same way.


If people like the idea I'm willing to code this myself.

I suggest checking out how other projects do it. There are already interfaces available via launchpad or similar I think, so it makes sense to take advantage of these. Essentially all that is required is to convert our language files into the required form so that the web translation services can be used, and convert back out of it. Coding up this side of it would be most welcome.

We have multiple things to translate:

1. Main strings (strings.xml in the language folder in SVN).
2. Addon strings (addons/resources/language/<lang>/strings.xml) in the addons git.
3. Addon descriptions (descriptions.xml for each addon).

We could perhaps look at making part of 3 autogenerated from 2 should it simplify things further and/or moving the language packs themselves to be addons so that we can consolidate all things french etc.

Cheers,
Jonathan

Great! I'll have a look at different tools from other projects.

1 year later..
not happening after all?

It took a long time, but now I'm working on a platform to make it easier to translate XBMC.

It can be used to translate xbmc & add-ons, all translations are centralized, which makes it easier to work together with multiple translators.

Hi sebak.
Thanks for investing your time on this. I'm the Portuguese translator and really think that a proper method to translations has to be found.
Current method of downloading files, translating and publishing diff for old files or new files for new translations is madness.

I say madness because add-ons resulted on hundreds of files that need translation. I can deal with translations on the hundreds... What I _can't_ deal is remembering all the gits, svns and looking every day to add-ons mailing list for updates and checkout if new strings exist.

We really need a new method for translations. Something along the lines of:

1 - Get all translators (Portuguese in my case) 'official/available' translators registered on the site/tool adopted

2 - Maybe adopting the method of anyone can collaborate on translations but only the language translation responsible can review and approve the correct translation to be committed by devs on svn, git, whatever

3 - Files created/updated in need of translation (usually English strings.xml is the one created/updated first) trigger an automatic 'call on translators' to update/create translation

4 - Get _all_ add-on developers on board such a solution, properly maintaining a 'translate convention' and translatable files database for ease of use and maximum efficiency on translations

Seriously, guys!
It's not funny to translate things, send patches to add-ons mailing list or trac only to realize your translation is outdated because the add-on developer added/remove/changed a string. Not my Friday night first choice, really.

One case that happened a couple days ago: my Portuguese scrapers translation on xbmc trunk were synced with dharma-pre git on sourceforge. Not a problem, right? Wrong! Some of the translations on git were older than the translations present on xbmc trunk.
Result? A few translations were lost on xbmc trunk...
Not funny...

Please, we need to get to a consensus on a solution.

hudo

Instead of reinventing the wheel you should rather focus on:

1) detokenize xbmc
2) removing duplicates

Then the transition to gettext can be made and rosetta (launchpad), which can already do everything hudo mentioned, can be used.