Bug#601549: perl: sprintf ignores LC_NUMERIC with constants

Niko Tyni ntyni at debian.org
Thu Oct 28 11:36:57 UTC 2010


On Wed, Oct 27, 2010 at 07:36:26PM +0300, Damyan Ivanov wrote:
> -=| Niko Tyni, Wed, Oct 27, 2010 at 11:28:42AM +0300 |=-
> > Package: perl
> > Version: 5.10.1-1
> > Tags: patch
> > Forwarded: http://rt.perl.org/rt3/Ticket/Display.html?id=78632
> > 
> > A regression in 5.10.1 breaks sprintf and LC_NUMERIC with numeric constants:
> > 
> >  perl -Mlocale -MPOSIX=locale_h -e 'setlocale(LC_NUMERIC, "fi_FI") or die; printf("%s should be %g\n", sprintf("%g", 2.4), 2.4)'
> >  2.4 should be 2,4

> > I think this is a possible candidate for squeeze but may not worth an
> > upload by itself.

> Hm, can I convince you that it is worth an upload all by itself? :)
> I remember being bitten by that behaviour, but can't remember details.

My initial thought was that as nobody had reported this upstream in the
14 months since 5.10.1 was released, it can't really be a big issue.

But OK, somebody (you and probably Helge Kreutzmann too) does care and
my fix just got applied upstream, so convinced :)

This should go nicely together with the LC_NUMERIC doc fixes from #379329
that also got applied upstream. The test suite changes in the upstream
commits could be left out, they probably don't apply to the 5.10 branch
as-is.

I'm going to be short on time for the next week or so. Any chance you'd
like to become a comaintainer and handle this yourself? :)
-- 
Niko Tyni   ntyni at debian.org






More information about the Perl-maintainers mailing list