p5pRT
commented
25 years ago From gnat@frii.com
[gnat@localhost]$ make test
PERL_DL_NONLAZY=1 /usr/bin/perl -Iblib/arch -Iblib/lib -I/opt/perl/5.005_58/i586-linux -I/opt/perl/5.005_58 -e 'use Test::Harness qw(&runtests $verbose); $verbose=0; runtests @ARGV;' t/*.t
t/basic.............Use of uninitialized value at /opt/perl/5.005_58/ExtUtils/MM_Unix.pm line 110.
t/basic.............ok 2/5Exiting subroutine via last at blib/lib/CPAN/WAIT.pm line 46.
Exiting subroutine via last at blib/lib/CPAN/WAIT.pm line 46.
Label not found for "last SERVER" at blib/lib/CPAN/WAIT.pm line 46.
t/basic.............dubious
Test returned status 255 (wstat 65280\, 0xff00)
With 5.005_57:
[gnat@localhost]$ make test
PERL_DL_NONLAZY=1 /usr/bin/perl5.00557 -Iblib/arch -Iblib/lib -I/opt/perl/5.005_57/i586-linux -I/opt/perl/5.005_57 -e 'use Test::Harness qw(&runtests $verbose); $verbose=0; runtests @ARGV;' t/*.t
t/basic.............Use of uninitialized value at /opt/perl/5.005_57/ExtUtils/MM_Unix.pm line 111.
t/basic.............ok
All tests successful.
Files=1\, Tests=5\, 5 wallclock secs ( 1.24 cusr + 0.13 csys = 1.37 CPU)
The weird thing being\, of course\, that 5.005_58 can't find the SERVER symbol\, while 5.005_57 can. I've tried looking at the op tree\, and while there's a heap of it\, I couldn't see any obvious differences between the two. I tried looking at the output of B::Terse on both\, and I see that _58 is using the new method op of Chip's\, but otherwise they seem identical.
The troublesome thing is that I can't narrow it down to a single piece of code. I tried perl -e 'SERVER: for (1\,2\,3) { last if 1 }' and perl -e 'sub foo { SERVER: for (1\,2\,3) { last if 1 }} foo()' but neither reproduces the difference between _57 and _58.
If anyone does manage to track this down\, I'd love to know how you did it.
This problem occurs on both FreeBSD and Redhat Linux running _58.
Nat
Migrated from rt.perl.org#1125 (status was 'resolved')
Searchable as RT1125$