Change in perl blead causes test failures in PDL

[jkeenan]

BBC report:
Perl/perl5#22354

Failure has been bisected to Perl/perl5@42cf752, which was committed to blead on Jun 24 2024.

Sample CPANtesters failure report:
http://www.cpantesters.org/cpan/report/f2acc314-35cb-11ef-989b-e9500d8dafec

Failure is in t/01-pptest.t, line 592.

[Leont]

So in Perl/perl5@2f4409b I aligned the behavior of input maps in ExtUtils::ParseXS with that of output maps. I had no idea PDL would contain a copy of the old input behavior (for both types of maps, which was always wrong for output ones). This patch would bring everything in line again

diff --git Basic/Gen/PP.pm Basic/Gen/PP.pm
index 98f2d5d6..b32a1569 100644
--- Basic/Gen/PP.pm
+++ Basic/Gen/PP.pm
@@ -1032,7 +1032,7 @@ sub typemap_eval { # lifted from ExtUtils::ParseXS::Eval, ignoring eg $ALIAS
   my ($var, $type, $num, $init, $pname, $arg, $ntype, $argoff, $subtype)
     = @$varhash{qw(var type num init pname arg ntype argoff subtype)};
   my $ALIAS;
-  my $rv = eval qq("$code");
+  my $rv = eval qq(qq\a$code\a);
   die $@ if $@;
   $rv;
 }

[mohawk2]

So what does this mean for PDL? Are you asking me to change the code? Will it work on older EU:PXS? Whose life is better as a result of this "alignment"? When is the new EU:PXS going to be released on CPAN?

[jkeenan]

On 6/30/24 17:04, mohawk2 wrote: So what does this mean for PDL? Are you asking me to change the code? Will it work on older EU:PXS? Whose life is better as a result of this "alignment"? When is the new EU:PXS going to be released on CPAN?

At this point I'm just the messenger here; I don't have definitive answers to all your questions. At this point I'd simply advise following the discussion in the perl GH issue to which I linked.

[mohawk2]

At this point I'm just the messenger here; I don't have definitive answers to all your questions. At this point I'd simply advise following the discussion in the perl GH issue to which I linked.

Fair enough! I will do so.