Out of interest with the experimental ?>, I did a benchmark with the following little test:

use Benchmark;
$str1 = "_test (folloed by 1 or more spaces)";
$str2 = "_test < xxx >";

timethese ( 1000000,
            { 'p1' => '&p1;',
              'p2' => '&p2;',
              'p3' => '&p3;',
              'p4' => '&p4;',
             } );

sub p1 ()
{
    $str1 =~ /_test(?>\s+)(?!<)/;
}

sub p2 ()
{
    $str1 =~ /_test(?:\s+)(?!<)/;
}

sub p3 ()
{
    $str2 =~ /_test(?>\s+)(?!<)/;
}

sub p4 ()
{
    $str2 =~ /_test(?:\s+)(?!<)/;
}
[download]

I got the following results:

Benchmark: timing 1000000 iterations of p1, p2, p3, p4...
        p1:  3 wallclock secs ( 3.00 usr +  0.00 sys =  3.00 CPU)
             @ 333333.33/s (n=1000000)
        p2:  3 wallclock secs ( 2.79 usr +  0.00 sys =  2.79 CPU)
             @ 358422.94/s (n=1000000)
        p3:  3 wallclock secs ( 3.09 usr +  0.00 sys =  3.09 CPU)
             @ 323624.60/s (n=1000000)
        p4:  3 wallclock secs ( 2.82 usr +  0.00 sys =  2.82 CPU)
             @ 354609.93/s (n=1000000)
[download]

It seems that the ?> runs slower than ?: matching by as much as 10 percent. So am I correct to say that optimization wise, the ?> might not be the first choice?