First, a minor stylistic issue; the line
($limit < $default) ? $time_to_die = $limit : $time_to_die = $default;
Is more often written
$time_to_die = ($limit < $default) ? $limit : $default;
But back to the matter at hand. In general, any value that
has to be looked up in a hash or through a function is going
to be slower than just grabbing a scalar. A quick benchmark
I wrote:
package Apache::Test;
use strict;
use Apache::Constants qw(:common);
use Benchmark;
use vars qw($r); # Not a good idea in general, but I couldn't
# figure out how to pass parameters in a
# benchmark. Works fine for benchmarking purposes.
sub handler {
$r = shift;
$r->send_http_header;
$r->print("by_dirconfig:",timestr(timeit(250000,\&by_dirconfig
+)),"\n");
$r->print("by_var:", timestr(timeit(250000, \&by_var)), "\n");
return OK;
}
sub by_dirconfig {
my $time_to_die;
if ($r->dir_config('TimeLimit')) {
if ($r->dir_config('TimeLimit') < $r->dir_config('Defa
+ultLimit')) {
$time_to_die = $r->dir_config('TimeLimit');
} else {
$time_to_die = $r->dir_config('DefaultLimit');
}
} else {
$time_to_die = $r->dir_config('DefaultLimit');
}
}
sub by_var {
my ($default, $time_to_die, $limit);
$default = $r->dir_config('DefaultLimit');
$limit = $r->dir_config('TimeLimit') if ($r->dir_config('TimeL
+imit'));
$time_to_die = ($limit < $default) ? $limit : $default;
}
1;
Which leaves in my browser window:
by_dirconfig:14 wallclock secs (15.28 usr + 0.03 sys = 15.31 CPU)
by_var:11 wallclock secs (12.07 usr + 0.02 sys = 12.09 CPU)
Conclusions are about the same as yours, just thought I'd
add my results.