No! I think you should be as free as possible to modify anything. The key is not to be more stringent on your side, but to be more reactive on the translator's side. And it's something difficult because the flow of information on what's new/what sould be translated is not easy.
Could you please describe shortly how you are currently doing the translation. There are already excellent tools available, perhaps you just need to use them. For example, for PO files, I could not live without the emacs PO mode, and the ediff mode. Others prefer gtranslator or kbabyl for that; this is a matter of taste.
* I'd like to see an online database, in which each item to be translated would be linked to all its translation in all "idioms"
What's an "idiom", in this context?
* of course each "idiom" is the metting point of a translators team (free name) and a "language" (iso representation) to allow multiple translations and splits.
I cannot follow. What is a split?
* Each new/modified item in the original idiom would be flagged as "not translated" in all idioms. Each team would then receive by email a list of what's new to do, for example when you insert new strings 7 and 8, idiom3 team would receive:
Shipping po files automatically is certainly a thing that can be done easily. If you want, the Translation Project (TP), at http://www.iro.umontreal.ca/contrib/po/HTML/ can offer maintainance of the PO files. This would result in the following procedure: 1. Barry submits the PO template to the TP. 2. The TP robot merges it with all previous translations, and announces it to all teams; the assigned translator gets a copy of the merge result. 3. the assigned translator updates the PO file, and submits it to the TP robot. This archives the file, and forwards it to Barry for inclusion.
- a tool to browse all items (and search by content in any idiom) - an open comment a translation/give a new translation/ that would result in suggestions being submitted to the commiters.
Are you sure that this kind of collaborative work would actually simplify things? I'd expect that there will be no more than one active translator for each language, so offering collaborative work does not help.
Then we'll get translation info circulating fast, corrections easy to make and overall i18n speed increase, I guess. That's not a project for Mailman, I guess, but something more general. I wonder why/if it does not exist yet?
On the automatic distribution of catalogs, it exists; this is what the Translation Project does, for PO files, or the Debian translation project for HTML files. On the collaborative work part, this does not exist, IMHO, since nobody wants to use it. I'd suggest that you get in contact with your language team within the TP; maybe they can offer their experience. Regards, Martin