Re: utf8 bug ?

[Christian Borup <borup borup com>]

On Sat, 2004-09-04 at 23:48 +0200, Jan Hudec wrote:
> On Sat, Sep 04, 2004 at 21:22:54 +0200, Christian Borup wrote:
> > On Sat, 2004-09-04 at 10:25 -0400, muppet wrote:
> > > On Sep 3, 2004, at 12:15 PM, Quentin wrote:
> > >
> > > > On Fri, 2004-09-03 at 15:23, Christian Borup wrote:
> > > > > This is a bug in perl - which may have been fixed in the latest 
> > > > > releases
> > > > > (5.8.5 had a bunch of regexp vs utf8 fixes).
> > > > >
> > > > > Can you test with perl 5.8.5 and report back? If the problem persists 
> > > > > I
> > > > > will pester one of the perl5-porters to fix it.
> > >
> > > just now while answering email, i compiled a new perl 5.8.5 and Gtk2 
> > > (all into a sandbox, so i'm presuming i got all the paths right), and i 
> > > still see the panic.
> >
> > Me too. In fact I've taken Gtk2 out of the equation. The attached
> > script still crash.
> >
> > ./borup
>
> Content-Description: sv_pos_b2u panic script
> > use Encode qw( _utf8_on );
> > my $string='string';
> > _utf8_on($string);
> > $_='a %1 b %1';
> > s/%1/replace($string)/eg;
> > warn $_;
> >
> > sub replace
> > {       $_=shift;
> >         return $_;
> > }
>
> This is really strange:
>
> The code:
>
> use utf8;
> $s1 = 'string';
> $s2 = 'štring';
> print "\$s1 ($s1) is " . (utf8::is_utf8($s1) ? '' : 'NOT ') . "unicode\n";
> print "\$s2 ($s2) is " . (utf8::is_utf8($s2) ? '' : 'NOT ') . "unicode\n";
>
> Produces:
>
> $s1 (string) is NOT unicode
> $s2 (štring) is unicode

This isn't strange - that is exactly what should happen...

> Upgrading the first string to unicode is what causes the error, though
> the string is ASCII and so is also correct UTF8.

That part is odd. But I'm starting to think that the problem is $_
related if you change the replace function to not use $_, it doesn't
panic.

./borup