Re: Translating schema files (was: Re: Survey results (yay!))

[Kenneth Nielsen <k nielsen81 gmail com>]

Hallo all

I thought I would chime in shortly on this slightly heated debate. I
think we should realise that this is in fact 2 separate discussions.

1) Should we split up the files?
2) On which numbers should the criterion for the golden "supported
language" badge be based?

For part 1 I think there is a way we can solve this, pretty much to
the satisfaction of all, if we wait until damned lies has the
capability to commit to git.

Since people here has already shown that you can split a po-file into
a "schemas/UI[1]"-set with the translation toolkit, then we could make
Damned Lies do it. So instead of having 1 file for download for each
module, there would be 3. There would be total/schemas/UI, where total
= schemas+UI and the schemas and ui files would just be generated by
filtering the total po-files. Then when committing something, that is
only a gconf or UI file, we would merge them into the main file before
commit. Doing it this way means:

A) We have all the statistics that people have asked for.
B) No changes for developers
C) Minimal change to translators that wants to commit via git (they
need to exchange their "copy into po-folder"-command with a
merge-command).[2]

If we focus on solving part 1 first (when we git commit in DL), then
translators can at least have the opportunity to prioritize. Then we
can always fight about what the criterion for "supported" should be
later.

Regards Kenneth

[1] I know that UI is probably not an appropriate name since it will
contains other stuff than UI, but I hope you get the point.
[2]  They would off course need to do it twice as often, which is
another goof reason to wait until we can commit from damned lies. I
think I do quite enough git grunt work as it is.