Adopting the calls for getProcessMemoryInfo on Linux was easy enough. Something like:

open(PID, "/proc/$$/statm");
return (split(' ', <PID>))[5] * 4;
[download]

but I found that the overhead of handling every statement with DB::DB was far too high. I tried to adjust the script to use DB::sub and just record subroutine calls.

I need to include something in my DB::sub function that handles the recursion caused by the calls to $Q->pending etc. but so far I have failed to come up with a mechanism that both breaks the recursion and actually keeps on executing. Any suggestions would be appreciated.

my $handling = 0;
sub DB::sub {
    return &$DB::sub if $handling;
    $handling = 1;
    scalar $Q->dequeue if $Q->pending > $OPTS{ LINES };
    $Q->enqueue( $DB::sub );
    $handling = 0;
    return &$DB::sub;
}
[download]