in reply to regex for negating a sequence

might be faster to just loop (5 times) through the next two characters in the string and test for equality. lookahead/lookbehind assertions are rather expensive if the string being matched is long 
$matched = undef;
for($index=0,$index<11;$index+=2) {
  if(substr($test_string;$index,2) eq "ab") {
    $matched = 1;
    last;
  }
}

[download]
untested code
the hardest line to type correctly is: stty erase ^H

Replies are listed 'Best First'.
Re^2: regex for negating a sequence (lookaheads and lookbehinds are now fast)
by grinder (Bishop) on Oct 04, 2006 at 14:53 UTC
    lookahead/lookbehind assertions are rather expensive if the string being matched is long

    demerphq might be too shy to toot his own horn, but a couple of weeks he landed an awesome patch to the development version of Perl that teaches the optimiser to look inside those assertions and use the information to improve the matching strategy used. Which equates to some massive speedups in some cases, perhaps the cases to which you were alluding.

    Let's just hope that Perl 5.10 sees the light of day soon. There's lots of goodness in there waiting to be unleashed.

    • another intruder with the mooring in the heart of the Perl

[reply]

In Section Seekers of Perl Wisdom