While the following code works, it offends me because the regex on the line marked with #!!!!! captures only a single line and I'd prefer to capture everything up to the next Smart Comment. Where's my "misteak?"

#!/usr/bin/perl

use Data::Dumper;
use strict;
use warnings;
use v5.10;

local $/;
my $code_s=<DATA>;

    while ($code_s !~ m{\G$}cg) {
        if ($code_s =~ m{\G^(\s*###+.*\n)}cgm) {
            # $1 is a Smart Comment upto and including the \n
            warn Data::Dumper->Dump([\$1],[qw(*1)]).' ';
            # Since the \n was captured we don't need to bump pos($cod
+e_s)
            #pos($code_s)++;
             }
        elsif ($code_s =~ m{\G((?:.*$)*)(?!\s*###+)}cgm) { #!!!!!
            # $1 only captures a single line 
            # why not multiple lines complete with their \n?
            warn Data::Dumper->Dump([\$1],[qw(*1)]).' ';
            pos($code_s)++;
             }
        else {
            my $pos=pos($code_s);
            my $residue=substr($code_s,$pos);
            die 'WTF:\n'.Data::Dumper->Dump([\$pos,\$residue],[qw(*pos
+ *residue)]);
             };
         };

exit;
__END__
no
 ### yes
 # ### no
no 
no 
 #### yes
[download]

yields

$1 = \'no';
  at Parse.pl line 20, <DATA> chunk 1.
$1 = \' ### yes
';
  at Parse.pl line 14, <DATA> chunk 1.
$1 = \' # ### no';
  at Parse.pl line 20, <DATA> chunk 1.
$1 = \'no ';
  at Parse.pl line 20, <DATA> chunk 1.
$1 = \'no ';
  at Parse.pl line 20, <DATA> chunk 1.
$1 = \' #### yes
';
  at Parse.pl line 14, <DATA> chunk 1.
[download]

Note I corrected \s###+ to \s*###+