On Fri, 2003/01/03 at 15:29:27 +0200, Peter Pentchev wrote:
> On Fri, Jan 03, 2003 at 02:31:41PM +0200, Peter Pentchev wrote:
> > FWIW, I can reproduce this on panther.FreeBSD.org, a sparc64 running
> > 5.0-CURRENT as of December 6, 2001. All the other machines in the
> > FreeBSD cluster that I could test - ref5 (i386), pluto1 (IA-64), and
> > beast (alpha) - are able to set both the is_IS.ISO8859-1 and the
> > bg_BG.CP1251 locales correctly for both LC_ALL and LC_TIME. Only the
> > sparc64 machine is having trouble setting LC_ALL for any locale I tried.
>
> Actually, I seem to have found the reason: on panther.FreeBSD.org, there
> are no LC_CTYPE files for any locale, and setlocale(LC_ALL) attempts to
> load, well, *all* locale type definitions, including LC_CTYPE. Thus, it
> would (and does) return NULL with errno set to ENOENT for all locales...
>
> The LC_CTYPE files are not there, because there is an explicit test for
> the build architecture in src/share/Makefile, which excludes the
> mklocale directory for sparc64; it was introduced in rev. 1.27 by David
> O'Brien about 7 months ago. David, what exactly was the 'bad juju'
> mentioned in the commit message, and is it still there? A test run of
> 'cvs up share/mklocale && cd share/mklocale && make' on panther seemed
> to complete OK; I have not yet tested the resulting locale files though.
Yeah, I just noticed that, too. As far as I can tell, this is purely
historical, if I remember correctly mklocale did crash back then. I
will re-enable it.
- Thomas
--
Thomas Moestl <[EMAIL PROTECTED]> http://www.tu-bs.de/~y0015675/
<[EMAIL PROTECTED]> http://people.FreeBSD.org/~tmm/
PGP fingerprint: 1C97 A604 2BD0 E492 51D0 9C0F 1FE6 4F1D 419C 776C
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
