Re: Do a named regex group in 5.8?

[Anonymous Monk]

The (?{}) and (??{}) constructs are available for use in perl5.8, but you'd have to ask yourself if such ugliness is really worth it:

$ perl -MData::Dumper -e 'q(abc de f abcdef) =~ m/(?<x>\w{3}).*\k<x>/ 
+and print Dumper {%+};'
$ perl5.8.8 -MData::Dumper -e 'q(abc de f abcdef) =~ m/(?{$+{x} = pos}
+)(\w{3})(?{$+{x} = substr($_, $+{x}, -$+{x}+pos)}).*(??{$+{x}})/ and 
+print Dumper {%+};'

[AnomalousMonk]

This can be made slightly less ugly by

c:\@Work\Perl>perl -wMstrict -le
"use Data::Dumper qw(Dumper);
 ;;
 print qq{perl version $]};
 ;;
 q(abc de f abcdef) =~ m/
   ( (\w{3}) (?{ $+{x} = $^N; }) .* (??{ $+{x}; }) )
   /x and print qq{\$1 '$1' \n}, Dumper \%+;
"
perl version 5.008009
$1 'abc de f abc'
$VAR1 = {
          'x' => 'abc'
        };

if it's really necessary to have named backreferences in the regex via a hash.

Otherwise, I'm not sure I see its advantage over something like

c:\@Work\Perl>perl -wMstrict -le
"print qq{perl version $]};
 ;;
 q(abc de f abcdef) =~ m/
   ( (\w{3}) .* \2 )
   /x and print qq{\$1 '$1'};
"
perl version 5.008009
$1 'abc de f abc'

or perhaps like

c:\@Work\Perl>perl -wMstrict -le
"print qq{perl version $]};
 ;;
 my $match = my ($whole, $first_part) = q(abc de f abcdef) =~ m/
   ( (\w{3}) .* \2 )
   /x
 ;;
 print qq{first part '$first_part'  whole '$whole'} if $match;
"
perl version 5.008009
first part 'abc'  whole 'abc de f abc'