Front page | perl.perl5.porters |
Postings from June 2009
Citeren Rafael Garcia-Suarez <rgarciasuarez@gmail.com>:
> In perl.git, the branch blead has been updated
>
> <http://perl5.git.perl.org/perl.git/commitdiff/e8514a9ea0b3241fdae7b1f70306632bc85a7004?hp=242a51a42272a23c60feb9c1b7cd686bfc28e04b>
>
> - Log -----------------------------------------------------------------
> commit e8514a9ea0b3241fdae7b1f70306632bc85a7004
> Author: Rafael Garcia-Suarez <rgarciasuarez@gmail.com>
> Date: Tue Jun 9 10:25:51 2009 +0200
>
> Fix regression test for RT 63110 when running with PERL_UNICODE set
>
> This is because block_start already does a copy of the hints in
> this case.
> -----------------------------------------------------------------------
>
> Summary of changes:
> t/op/eval.t | 3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/t/op/eval.t b/t/op/eval.t
> index 2fef2aa..071b2fa 100644
> --- a/t/op/eval.t
> +++ b/t/op/eval.t
> @@ -549,10 +549,11 @@ $test++;
> # don't leak memory
> {
> use feature qw(:5.10);
> + my $count_expected = ($^H & 0x20000) ? 2 : 1;
> my $t;
> my $s = "a";
> $s =~ s/a/$t = \%^H; qq( qq() );/ee;
> - print "not " if Internals::SvREFCNT(%$t) != 1;
> + print "not " if Internals::SvREFCNT(%$t) != $count_expected;
> print "ok $test - RT 63110\n";
> $test++;
> }
>
> --
I wondered why '$^H & 0x20000' was used in the above change.
The reason: when the test is run with LC_ALL='en_US.utf8'
PERL_UNICODE="IOESioDALa" ./perl harness -v op/eval.t then the hints
are not copied. (0x20000 comes from: #define HINT_LOCALIZE_HH
0x00020000 /* %^H needs to be copied */)
Since this was not very obvious (to me at least) I looked for another
way to test this: when HINT_LOCALIZE_HH is set then $t is always the
same as \%^H.
Attached is a patch which updates the test from '$^H & 0x20000' to '$t
== \%^H'.
Best regards,
Bram