Re: gettext's glocale

[Roozbeh Pournader <roozbeh farsiweb info>]

Thanks Bruno,

I believe that we have some answers now and a partial consensus on some
matters. I guess we should put these in a wiki somewhere, to make sure
they are not lost.

On Thu, 2005-08-11 at 16:11 +0200, Bruno Haible wrote:
> Both glibc and CLDR have bugzillas, where users can report differences.

Well, one of the main reasons we started this project was that there was
absolutely no timeline for updates to glibc locales get checked in the
CVS.

While we assume that sane requests will pass the CLDR process (I have
tried it, it usually works), they may not necessarily pass the glibc
locale process. People may object that some data has been that way for a
while, and changing them would have this and that bad effect somewhere.
Or an original locale submitter may wish to impose his own opinion, and
practically block changes, while in CLDR the process is less dependent
on individuals.

I've also been thinking about trying to maintain small difference to
CLDR data, in between the CLDR releases (final 1.3 appeared in June
2005, final 1.4 will appear in April 2006). This is specially useful
with less developed languages. Any opinions for or against this?

> The end user should have to set a single LC_ALL variable and not
> several ones, and its syntax is given by glibc.

I don't like this. I prefer that the user doesn't set LC_ALL at all but
sets a new environment variable, defined like LANGUAGE, containing a
list of CLDR locales to choose from. The glibc locale name does not have
all of CLDR locale name's power, which will limit the library's
functionality.

We need to sort out differences with LANGUAGE, LANG, and LC_* of course,
and say how do things behave when a random setting of those is in
effect.

> Qt wants to be a platform on their own, [...]
> 
> KDE is different: [...]
> 
> Mozilla is different too: [...]

Then I assume LGPL is fine. Any objections?

roozbeh