Re: How expensive is eval { }?

If you don't know, just try:

use strict;
use warnings;
use Benchmark qw(cmpthese);

no warnings 'void';

sub do_something {
    for (1..10){
        $_*$_;
    }
}

cmpthese -1, {
    eval    => sub { eval { for (1..10){$_*$_} } },
    sub     => sub { do_something() },

};
__END__
         Rate  sub eval
sub  446836/s   --  -0%
eval 446836/s   0%   --
[download]

This is not a very good benchmark, but if eval BLOCK was very slow you'd see it here.

Comment on Re: How expensive is eval { }? Download Code

Replies are listed 'Best First'.
Re^2: How expensive is eval { }? by mscharrer (Hermit) on Apr 17, 2008 at 09:18 UTC
Thanks, that's a good idea. I'm getting results from -3% to -6% in favor for sub on my machine. I might write a bigger benchmark and post the results here.	[reply]
Re^3: How expensive is eval { }? by moritz (Cardinal) on Apr 17, 2008 at 09:25 UTC
I rerun the test several times and with different perl version (5.8.8 and 5.10.0), and got results from 0% to -4%. You can see that our both results have higher noise than the difference, so I wouldn't base a design decision on it. If exceptions are cleaner from a programming POV I'd go with exceptions.	[reply]