closures and //o (was [Fwd: failure notice])

This is new, isn't it?  Oh, well, here it is again.

-------- Original Message --------
Subject: failure notice
Date: 31 Mar 2001 05:09:26 -0000
From: MAILER-DAEMON@tmtowtdi.perl.org
To: rick.delaney@home.com

Hi. This is the qmail-send program at tmtowtdi.perl.org.
I'm afraid I wasn't able to deliver your message to the following addresses.
This is a permanent error; I've given up. Sorry it didn't work out.

<perl5-porters@perl.org>:
This mailinglist does not accept postings crossposted to newsgroups - Contact
list-owner@perl.org for help

--- Below this line is a copy of the message.

Return-Path: <rick.delaney@home.com>
Received: (qmail 23855 invoked from network); 31 Mar 2001 05:09:25 -0000
Received: from femail1.rdc1.on.home.com (24.2.9.88)
  by tmtowtdi.perl.org with SMTP; 31 Mar 2001 05:09:25 -0000
Received: from home.com ([24.114.70.197]) by femail1.rdc1.on.home.com
          (InterMail vM.4.01.03.20 201-229-121-120-20010223) with ESMTP
          id <20010331050750.UGZE20157.femail1.rdc1.on.home.com@home.com>
          for <perl5-porters@perl.org>; Fri, 30 Mar 2001 21:07:50 -0800
Sender: rick
Message-ID: <3AC569BB.49AAC734@home.com>
Date: Sat, 31 Mar 2001 00:23:07 -0500
From: Rick Delaney <rick.delaney@home.com>
X-Mailer: Mozilla 4.72C-CCK-MCD Caldera Systems OpenLinux [en] (X11; U; Linux
2.2.14 i686)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: comp.lang.perl.misc
CC: perl5-porters@perl.org
Subject: Summary: closures and //o (was Re: regex-qr// for search and replace)
References: <985712399.15095.0.nnrp-01.c3ad6974@news.demon.co.uk>
<q3j1ctoh4nqs7loj9lrfh6eo20ksmfbjuf@4ax.com> <3AC1644F.357045C9@home.com>
<9uo8ctk0cn8nv4qcu08vvp8paqu902d87b@4ax.com> <3AC49A83.9D063CD8@home.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

[posted to clpmisc and cc'ed p5p]

This is supposed to be a summary but it's kinda long.

Bart Lateur wrote:
> 
>         sub regexsub {
>             my $re = shift;
>             return sub {
>                 shift =~ /$re/o;
>             }
>         }
>         $sub[0] = regexsub('fo+');
>         $sub[1] = regexsub('ba+r');
>         
>         foreach (qw'foo baaar fooooo bar bbbbbbbbar') {
>            foreach my $i(0, 1) {
>               $sub[$i]->($_) and print "Match for '$_' with regex $i\n";
>            }
>         }
> -->
>         Match for 'foo' with regex 0
>         Match for 'baaar' with regex 1
>         Match for 'fooooo' with regex 0
>         Match for 'bar' with regex 1
>         Match for 'bbbbbbbbar' with regex 1
> 
> In pre-5.6.0, //o and closures don't work too well together. There,
> you'd have to put the closure (the sub that gets returned by regexsub)
> in an eval STRING construct.

I wrote that I get

Match for 'foo' with regex 0
Match for 'foo' with regex 1
Match for 'fooooo' with regex 0
Match for 'fooooo' with regex 1

with my copy of 5.6.0 (and up).

Bart wrote:
> Gee. It works properly in ActivePerl 5.6.0 Build 623 for
> MSWin32-x86-multi-thread,
...
> Could it be in the patches for multi-thread support?

Yes!

I recompiled my copy of perl-5.6.0 to include threads and it now 
exhibits the same behaviour as ActivePerl.  I also downloaded and 
compiled ActivePerl without threads and it only gives me the /fo+/ 
matches.

But...

It appears that multi-thread support has actually broken /o which is
the only reason it "works".

$ perl -le 'for (qw/foo bar baz/) {print if /$_/o }'
foo 
$ perl -v           

This is perl, v5.6.0 built for i686-linux
... 
$ as_perl -le 'for (qw/foo bar baz/) {print if /$_/o }'
foo
bar
baz
$ as_perl -v

This is perl, v5.6.0 built for i686-linux-thread-multi
(with 1 registered patch, see perl -V for more detail) 
...

When run under -Mre=debug you can see that the regexp is compiled
3 times with threads but only once without.

Maybe this is to be expected?

Also see bug ID 20000810.008 and

http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2000-08/msg00770.html

-- 
Rick Delaney
rick.delaney@home.com

• closures and //o (was [Fwd: failure notice]) by Rick Delaney
• Re: closures and //o (was [Fwd: failure notice]) by Mark-Jason Dominus