Am Freitag, 18. April 2014, 21:11:10 schrieben Sie: > 2014-04-18 20:50 GMT+02:00 Burkhard Lück <lu...@hube-lueck.de>: > > Am Freitag, 18. April 2014, 16:25:40 schrieb Matthias Klumpp: > >> These scripts illustrate how translation of the XML can be done in > >> KDE. I would kindly ask the l10n-script developers to take a look at > >> it any maybe integrate it at the proper places, because they know the > >> code way better than I do. > >> > >> The scripts can be found at: > >> http://people.freedesktop.org/~mak/appstream/kde-asmetadata-l10n.tar.gz > >> > >> (I also included a demo XML file) > >> I wanted to post that to Reviewboard, but couldn't find a project for > >> Scripty. > > Please use https://svn.reviewboard.kde.org/groups/l10n/ for your patch
> > Quoting my comment on https://git.reviewboard.kde.org/r/117609/ > > > > Please have a look at the recent changes in > > http://websvn.kde.org/trunk/l10n-kde4/scripts/?sortby=date#dirlist > > > > extract-xml.sh > > update_translations > > fillxmlfrompo.py > > > > These changes are about a similar problem, ectracting strings for > > translation from mimetype xml files and fill the xml files with the > > translations. > Some of these files already existed back in the days. From what I've > seen, the code takes the mime-xml, extracts interesting tag > (<comment/>) performs some hacks so intltool likes it (<comment/> => > <_comment/>) and then does similar stuff my examples do (but with > intltool instead of itstool, which IMHO is mot really nice for this > purpose (I love intltool otherwise!)). > I also don't fully understand the need for XmlMessages.sh in the source > tree... So, because of this and especially because of: > # Note to developers: > # there are not many comments in this code to avoid to slow down the code > # So please look also in the svn log to have comments about the code > I would need help from someone experienced with the l10n workflow and > especially the l10n code to integrate the changes - that stuff is > really hard to understand if you don't know the details and if there > are zero comments explaining things[1]. The UML helps a little bit, > but it still leaves lots of questions. > I think with the scripts above it's trivial for someone who knows the > code to call them from the appropriate places and make the necessary > adjustments ;-) > Cheers, > Matthias > > [1]: Maybe long-term it would make sense to attempt a Python rewrite > of this? (there are currently bits of C++, Python, Perl, Sh and Bash > in this - Esp. calls like "g++ -O2 -march=nocona -o apply > scripts/applycontext.cpp" in the generation code make me wonder...) Thanks. -- Burkhard Lück