Date: | Wed, 22 May 2002 12:12:12 +0100 |
From: | Nicholas Clark <nick [...] ccl4.org> |
To: | Jesse <jesse [...] fsck.com> |
Cc: | perl5-porters [...] perl.org, bug-storable [...] rt.cpan.org |
Subject: | Re: Storable 2.0.0 fails on vendor perl on Mac OS X 10.1 |
On Tue, May 21, 2002 at 04:58:17PM -0400, Jesse wrote:
Show quoted text
> (Yeah, I know that 5.6.0 breaks in various wacky ways, but in this
> case, apple's 5.6.0 is the latest released perl version that will
> build on mac os x.)
I didn't know that. It surprises me that 5.6.1 doesn't, but I don't know
Mac OS X (except indirectly via perl5-porters, particularly w.r.t.
LONG_MIN in <limits.h>)
Show quoted text> The old Storable worked great. Storable 2.0.0 fails tests consistently
> like this:
Show quoted text> t/downgrade.....NOK 34# Failed test (t/downgrade.t at line 99)
> # got: 'Ã'
> # expected: '?
> t/downgrade.....NOK 37# Failed test (t/downgrade.t at line 99)
Thanks for the report. I get these failures on every system I've built
Storable with 5.6.0 on, and documented them in the ChangeLog.
Probably I should have done it in the README, or even as a warning in the
Makefile.PL.
To the best of my knowledge Storable 2.00 works just as well on 5.6.0 as
1.0.14. The test is failing because of bug in eq in perl 5.6.0:
$ cat S560crap.t
#!/usr/local/bin/perl5.6.0
use warnings;
use strict;
my $utf8 = "\x{DF}\x{100}";
chop $utf8;
my $byte = "\xDF";
if ($utf8 eq $byte) {
printf "Yes\n";
} else {
print "No\n";
printf "%d byte of %d, %b byte of %d\n", length $utf8, ord $utf8,
length $byte, ord $byte;
}
__END__
$ perl5.7.3 S560crap.t
Yes
$ perl5.6.1 S560crap.t
Yes
$ perl5.6.0 S560crap.t
No
1 byte of 223, 1 byte of 223
Nicholas Clark