Re^2: Parser Performance Question

[dave_the_m]

I agree with LanX: we need a short, self-contained example. For example I tried with this, but it doesn't reproduce the problem:

$n = 'x' x 50 . "\n";
$p = "=foo $n";
$np = ($n x 50) . $p;
$_= $np x 100_000;

1 while m/\G ( = [a-zA-Z] .* ) \n/xgc;

In fact for me, 5.20 is 3 times faster than 5.18 with that example. Since for 5.20.0 I heavily reworked the part of the regex engine which is giving those debugging messages you show, I'd be very interested to have access to real working examples of where my changes made things go slower rather than faster.

Dave.