As i18n stared during a hackatron it was implemented for rewrite branch only and I ported it to the trunk version. That's the reason why the files are place in the rewrite branch. You should not place new translation files to the trunk but the rewrite release. All new translations should be done at twn only after the file is ported.

Regards
xqt

2012/2/11 Dr. Trigon <dr.trigon@surfeu.ch>

-
http://www.mail-archive.com/pywikipedia-l@lists.wikimedia.org/msg02255.html

and thus

http://lists.wikimedia.org/pipermail/pywikipedia-l/2011-January/006632.html

Hoping this helps... ;)

Thank you, but I still don't really understand.
This is about creating translations in rewrite/i18n. OK.
How do they get to TW?
How do they get back?
This letter tells more about this: http://www.mail-archive.com/pywikipedia-l@lists.wikimedia.org/msg02269.html So some manual work is done, but who does it, and what is the algorithm?
Which is the main field of translating, those "shadow" scripts in /i18n subdirectory, or translatewiki? May translations be done in any of these? Or is the TW just for those who have no access to Pywiki SVN, and committers should translate in scripts?

One more: xqt says: "i18n dicts are placed in the rewrite branch. In trunk release i18n directory is defined as svn:external which copies its content from rewrite."
What happens if I write a script to trunk that is still not implemented in rewrite? Should its "shadow" script also be placed in rewrite/i18n?