[perl #104504] Variable "$bar" is not available

On Mon Nov 28 02:23:04 2011, nicholas wrote:
> On Thu Nov 24 07:04:45 2011, dk@tetsuo.karasik.eu.org wrote:
> > -----------------------------------------------------------------
> > [Please describe your issue here]
> > 
> > perl 5.14 shows behavior not present in 5.12. I've used that behavior
> > for
> > conditional debugging.
> > 
> > $ cat a.pl
> > use strict;
> > use warnings;
> > 
> > my $debug = 0;
> > 
> > sub foo
> > {
> > 	my $bar = 0 if $debug;
> > 	return sub { $bar++ if $debug };
> > }
> > 
> > foo();
> > 
> > $ perl a.pl
> > Variable "$bar" is not available at a.pl line 9.
> 
> Assuming from your test program that your intent is to be able to
> generate closures at runtime with optional debugging information, a
> small variation of your test program shows that in the general case your
> approach will *also* warn on 5.12:
> 
> $ cat 104504.pl 
> use strict;
> use warnings;
> 
> my $debug;
> 
> sub foo
> {
>     warn "\$debug = $debug";
>     my $bar = 0 if $debug;
>     return sub { $bar++ if $debug };
> }
> 
> foreach (@ARGV) {
>     $debug = $_;
>     foo();
> }
> 
> __END__
> 
> $ ~/Sandpit/5120/bin/perl 104504.pl 0 1 0 1 0 1
> $debug = 0 at 104504.pl line 8.
> $debug = 1 at 104504.pl line 8.
> $debug = 0 at 104504.pl line 8.
> Variable "$bar" is not available at 104504.pl line 10.
> $debug = 1 at 104504.pl line 8.
> $debug = 0 at 104504.pl line 8.
> Variable "$bar" is not available at 104504.pl line 10.
> $debug = 1 at 104504.pl line 8.
> $ ~/Sandpit/5140/bin/perl 104504.pl 0 1 0 1 0 1
> $debug = 0 at 104504.pl line 8.
> Variable "$bar" is not available at 104504.pl line 10.
> $debug = 1 at 104504.pl line 8.
> $debug = 0 at 104504.pl line 8.
> Variable "$bar" is not available at 104504.pl line 10.
> $debug = 1 at 104504.pl line 8.
> $debug = 0 at 104504.pl line 8.
> Variable "$bar" is not available at 104504.pl line 10.
> $debug = 1 at 104504.pl line 8.
> 
> You only avoid seeing warnings on 5.12 if you only ever call your
> routine with $debug unchanged as zero.
> 
> Given that the general case version of your routine will warn, I assume
> that this is what your are doing.
> 
> In which case, change my $debug = 0; to use constant DEBUG => 0; which
> won't want, and will actually generate better code.
> 
> Alternatively, rewrite you routine as
> 
> sub foo
> {
> 	my $bar = 0;
> 	return sub { $bar++ if $debug };
> }
> 
> 
> ie remove the trailing conditional from the my statement.
> 
> 
> IIRC the change in visible behaviour for the *first* calls to a
> subroutine is a side effect of an internal refactoring, intended to
> reduce the amount of storage pre-allocated in pads.

For the record:

adf8f095c5881bcedf07b8e41072f8125e00b5a6 is the first bad commit
commit adf8f095c5881bcedf07b8e41072f8125e00b5a6
Author: Nicholas Clark <nick@ccl4.org>
Date:   Fri Feb 26 09:18:44 2010 +0000

    Set PADSTALE on all lexicals at the end of sub creation.
    
    The PADSTALEness of lexicals between the 0th and 1st call to a
subroutine is now
    consistent with the state between the nth and (n + 1)th call.
    
    This permits a work around in Perl_padlist_dup() to avoid leaking
active pad
    data into a new thread, whilst still correctly bodging the external
references
    needed by the current ?{} implementation. Fix that, and this can be
removed.

:100644 100644 cc2ade250e1c91cac2d5cca8e3dfdfdc3fab4ca5
8015154bf3041f4288bc881c39ae98eecd93e568 M	pad.c
:040000 040000 7af0e765a69f9b8942f86515956ea59024d0fb5f
611c812667cce814b844658b5146a74523465066 M	t
bisect run success
That took 343 seconds, believe it or not.

-- 

Father Chrysostomos

• Re: [perl #104504] Variable "$bar" is not available by Dmitry Karasik
• [perl #104504] Variable "$bar" is not available by dk @ tetsuo . karasik . eu . org
• [perl #104504] Variable "$bar" is not available by Nicholas Clark via RT
• [perl #104504] Variable "$bar" is not available by Father Chrysostomos via RT
• [perl #104504] Variable "$bar" is not available by James E Keenan via RT
• [perl #104504] Variable "$bar" is not available by Eric Brine via RT