Improving the GNU libc locales

[Petter Reinholdtsen <pere at hungry dot com>]

I'm pleased to see so many finding time to join this mailing list.
The topic at hand is GNU libc locales.  My goal is to make sure the
locale support in GNU libc is as good as possible.  The motivation is
to make sure as many people as possible can use it and this way make
sure as many of the remaining issues are detected and fixed.  I
believe this will make it easier for us in Norway to use our locales
as well, and improving the norwegian locale is my primary goal.

Some of you are already aware of my page with info on the locale
system, <URL:http://www.student.uit.no/~pere/linux/glibc/>.  It is
slightly out of date.

The bugs reported against locales using the old gnats based system is
lost, and we need to make sure the remaining issues are reported into
bugzilla with enough info for the GNU libc maintainers to be confident
enough about the quality of the fix to commit it to CVS.

The lack of authorative documentation on the locale format in GNU libc
is problematic.  I hope we can join forces and add some documentation
to the GNU libc info pages that reduces this problem.

I'm not quite sure where to go from here, but I have a list of issues
on my TODO list.  After working on the currency formatting for a
while, I believe that part is working as it should.  Next on my list
was charset conversion.  The issue I planned to look at is now lost in
the gnats database; "libc/4348: Converting charset
'eucjp->sjis->iso2022jp' and 'eucjp->iso2022jp' give different
result.".  Anyone know more about the issue?  I plan to write a
testing framework to test such conversions automatically.