Carlos E. R. skreiv 10. mars 2016 18:32:
I think it's safe to say that noone is happy with the current
situation. We need to do something about it and that most likely means some compromise for all involved parties. In the specific case of zypper for example I have a hard time understanding why it's translations are on svn.o.o rather than right in the source repo on github in a po subdirectory next to the code. Ie like pretty much any other upstream project. If translations were there it would be pretty clear which ones are used and where contributions have to go. No matter whether done by paid translators or volunteers. On all the upstream projects where I participate, translators have no access to the source tree. That's only for developers.
I have participated in several projects where the translators (at least some of them) have commit access to the source tree.
For instance, with the "Translation Project" (http://translationproject.org/), developers submit the .pot file in a manner I don't know to a centralized server for all translated projects in the "Translation Project", and from there, a robot mails notices to the translators. These pick the files from that server (we have read access to this one, not write), translate them, and submit them via email robot back to the server. I suppose the robot sends an email to the devs to pick it up.
Translation is only done once the developers submit a release, not for work in progress (would be similar to factory).
IMHO, the way the ‘Translation Project’ works is far from ideal. Your last comment is a case in point; it’s much better if the translators can work on translations continuously, instead of just before a release (and then have to translate a large number of strings at once). And having to send e-mails to update translations is a pain.
Check how KDE team work (upstream). I will not describe it, as I don't work there, but they have directories where all files are stored for translation. Other people here can describe the process (I think somebody did, not very long ago).
I work as a translator for KDE. There the translators have commit access, and fetch and commit translations using SVN. It works extremely well. Translators just have to do a svn up to fetch the files, translate them and then do a svn commit The translations automatically appear in the released applications (it feels almost like magic :) ). And the translations files are automatically updated (with new/changes strings) each night¹. Easy peasy, and no hassle for the translators. (And when an application is renamed, moved, split into an application + library or have other changes that need changes in the translation files (this happens frequently), the nice people at KDE automatically do the needed changes to all the translation files.) In KDE, all applications used to live in the same SVN repository. But a few years ago they were split into multiple Git repositories, stored in several different places (e.g., some at GitHub) (and a few applications stayed in SVN). But it was decided that keeping the central SVN (not Git) repository for translations would be best. So that’s what we use, and it works perfectly. Each language has its own directory (and there’s of course also a templates directory). The other option would be for the translator to manually hunt down each and every application and library (there are many hundreds) repository, and manually fetch the source code and commit the translations to them (or send pull requests), which would of course be untenable. Even having to download the hundred of gigabytes of data from the repositories would be difficult for most translators. But now they just have to type svn up To get all the latest translations files (for one’s language) to translate. It’s even faster than launching a browser. :) Could anything be simpler?
As a translator, I can not go navigating a zillion directories to find files to translate. A coder normally works in the scope of a project, say "zypper". There he sees many *.c files, say, and works in some of them. I suppose. But a translator works in the scope of many different and separate programs, say "KDE", where he will see many *.pot and *.po files, one per package or program typically.
I agree. Footnote: ¹ I said that in KDE the translations files are automatically updated (with new/changes strings) each night. Note that it’s also possible to opt out of this (by placing a ‘magic’ file in one language’s translation directory), and merge the translation files oneselves, if one prefers. Some teams prefer this, and there are some advantages in getting to decide when (and how) to update the translation files. -- Karl Ove Hufthammer -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org