On Wed, Oct 08, 2003 at 01:15:09AM -0700, Paul Eggert wrote:
> [EMAIL PROTECTED] (Denis Barbier) writes:
>
> > I am not sure what you mean by "user-defined locales", since your
> > patch does not check UID.
>
> I meant locales that are generated by a user. In this particular
> case, it's a POSIX-conformance-test shell script that generates
> the locales.
[...]
Ok, I read other related bugreports, here is a
<summary to="busy glibc maintainers">
* LSB 1.3 conformance is a release goal.
* LSB 1.3 does not require support for multibyte encodings.
* If multibyte encodings seem to be supported, core tools are
checked when in such encodings, and some of these tests currently
fail and thus prevent LSB compliance.
* Patches were sent to the BTS to presumably fix these bugs, but
some of them are too intrusive and cannot be safely integrated.
LSB 1.3 tests are buggy, and there are 2 alternatives:
1. Either deal with LSB folks and see what can be done to skip these
buggy tests
2. Or let LSB testers believe that multibyte encodings are not
supported.
Your patch deals with (2), by letting localedef abort (when in POSIX
mode and generating a multibyte encoding) with an error message.
</summary>
Some users run in POSIX mode (see #185924), but Debian glibc comes
with a locale-gen script which is a frontend to localedef.
In this script, POSIXLY_CORRECT is unset (see #206784), so our
users should not be affected by the requested change.
Denis
