[RFC] locales vs. belocs-locales-data
the locales package has not been updated for months. I do not
want to blame anyone, but please realize that people are very
frustrated because this has an impact on their l10n effort.
I am advocating for months for the locales package being
put into its own source package. There are obvious advantages
(glibc is frozen for a long time before releases) and some
drawbacks (compiled data files need to be generated with the
same format as expected by libc functions).
I packaged belocs-locales-data as a proof of concept, and it
looks like Ubuntu developers are convinced and follow this way.
But my primary goal was to convince Debian glibc maintainers ;)
My proposal is to:
* Exchange priority between locales and belocs-locales-data
* Keep the locales package for now as a failsafe option (and
because some packages Depends/Build-Depends on it).
You can stick to the glibc-2.3 or fedora branches, so that
your maintenance work is very low on localedate files.
This way you have less work with locale data (but there are still
issues like #310635 which needs to be fixed, I will send an update
very soon about this bugreport) and I believe that everyone could
be happy with it.