[Libreoffice] l10n based on .po files

On Wed, 2011-03-16 at 16:30 +0100, Thorsten Behrens wrote:
> Regarding gettext, I have a slight reservation - we'd be strongly 
> depending on correct context, otherwise short strings from different
> places will all be mixed up. I gather this will be mostly manual
> work?

        If the translation is ultimately done using .po files, surely that
duplication will already be there ? It is not as if we suggest to put
all translations for all of LibreOffice into many .po files; and there
are always misc. prefixes we can add to strings I suppose to
disambiguate them.

        Of course, git seems to store individual files - that is true; but it
should also do a reasonable job of compressing similar files together -
if that is not going well running 'git gc' can help.

        My understanding of the sdf file problem is that both Andras and
Christian are right - in theory there is no real problem with sdf; and
of course changing a string in an English .po file results in a similar
number of changes trickling all through the other .po files. I was lead
to believe ( can't remember why ) that the .sdf file tended to change /
renumber the whole file resulting in huge changes whenever strings were
added / removed - perhaps that is not so; no idea :-) how else to
explain the vast size of the l10n repo over time ? :-)

        ATB,

                Michael.

-- 
 michael.meeks@novell.com  <><, Pseudo Engineer, itinerant idiot