From @nwc10

Moritz Lenz mailed p5p in 477FACED.4000505@​casella.verplant.org
http​://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-01/msg00167.html

  With my shiny perl 5.10.0 on Linux I observed some bugs with extended
  regexp features. I know they're "experimental", but perhaps the reports
  help you anyway.

Dave notes​:

Bug #1 in the email still exists (bug 2 is fixed)​:

$ perl5.10.0 -Mre=eval -wlE 'my $re = qr/(\w+)/;"foo" =~ m/($re)(?{say +$^N})/'
Can't locate object method "say" via package "foo" (perhaps you forgot to load "foo"?) at (re_eval 1) line 1.

From @schwern

It appears no pragmas propagate into (?{ code }).

$ perl5.10.0 -Mre=eval -wlE 'my $re = qr/(\w+)/; use strict; "foo" =~
m/($re)(?{print +$bar})/'
Use of uninitialized value $bar in print at (re_eval 1) line 1.

$ perl5.10.0 -Mre=eval -wlE 'my $re = qr/(\w+)/; no warnings; "foo" =~
m/($re)(?{print +$bar})/'
Use of uninitialized value $bar in print at (re_eval 1) line 1.

This isn't new to 5.10.

$ perl5.8.8 -Mre=eval -wle 'my $re = qr/(\w+)/; no warnings; "foo" =~
m/($re)(?{print +$bar})/'
Use of uninitialized value in print at (re_eval 1) line 1.

Only the "use feature" pragma is new.

As its not a 5.10 regression, and (?{...}) is experimental, I'd suggest
removing it from the 5.10.1 blockers.

The RT System itself - Status changed from 'new' to 'open'

From @iabyn

On Sat, Jul 11, 2009 at 02​:02​:37AM -0700, Michael G Schwern via RT wrote​:

It appears no pragmas propagate into (?{ code }).

$ perl5.10.0 -Mre=eval -wlE 'my $re = qr/(\w+)/; use strict; "foo" =~
m/($re)(?{print +$bar})/'
Use of uninitialized value $bar in print at (re_eval 1) line 1.

$ perl5.10.0 -Mre=eval -wlE 'my $re = qr/(\w+)/; no warnings; "foo" =~
m/($re)(?{print +$bar})/'
Use of uninitialized value $bar in print at (re_eval 1) line 1.

This isn't new to 5.10.

$ perl5.8.8 -Mre=eval -wle 'my $re = qr/(\w+)/; no warnings; "foo" =~
m/($re)(?{print +$bar})/'
Use of uninitialized value in print at (re_eval 1) line 1.

Only the "use feature" pragma is new.

As its not a 5.10 regression, and (?{...}) is experimental, I'd suggest
removing it from the 5.10.1 blockers.

By the same logic, I don't think this bug needs to be a 5.12 showstopper.

Its also likely to be fixed as a side-effect my intended post-5.12
re-engineering of /(?{...})/ (which I've been promising for years....)

--
My Dad used to say 'always fight fire with fire', which is probably why
he got thrown out of the fire brigade.

From @obra

Only the "use feature" pragma is new.

As its not a 5.10 regression, and (?{...}) is experimental, I'd suggest
removing it from the 5.10.1 blockers.

By the same logic, I don't think this bug needs to be a 5.12 showstopper.

Its also likely to be fixed as a side-effect my intended post-5.12
re-engineering of /(?{...})/ (which I've been promising for years....)

I can buy that. I'm removing this as a blocker.

From @cpansprout

Fixed by 558b442.

@cpansprout - Status changed from 'open' to 'resolved'

From @nwc10

On Tue Dec 07 18​:04​:16 2010, sprout wrote​:

Fixed by 558b442.

That revision doesn't exist. I guess you made a typo. Do you still have
the details of the revision handy?

Nicholas Clark

From @nwc10

On Wed Dec 08 01​:33​:44 2010, nicholas wrote​:

On Tue Dec 07 18​:04​:16 2010, sprout wrote​:

Fixed by 558b442.

That revision doesn't exist. I guess you made a typo. Do you still have
the details of the revision handy?

Aha. It does *now*. PEBKAC, at least when I thought I'd done a git pull.

Nicholas Clark