Re^2: Optimize debugging subs

This benchmark indicates that you are wrong.

sub ENABLE_DEBUG () { 0 }
sub debug_inelegant { print $_[0]; }
sub debug_elegant { print $_[0] if ENABLE_DEBUG; }

use Benchmark qw(:all) ;

cmpthese( 30000000, 
    { 
        'inelegant' => "debug_inelegant 'test' if ENABLE_DEBUG;",
        'elegant'   => "debug_elegant 'test';"
    },
);
[download]

Output:

                Rate   elegant inelegant
elegant    3337041/s        --      -95%
inelegant 62500000/s     1773%        --
[download]

--Dan

Comment on Re^2: Optimize debugging subs Select or Download Code

Replies are listed 'Best First'.
Re^3: Optimize debugging subs by ikegami (Patriarch) on Apr 05, 2006 at 05:07 UTC
According to your benchmark, `elegant` takes 0.0000003 seconds (0.3 microseconds) to execute. What's the problem?	[reply] [d/l]
Re^4: Optimize debugging subs by danb (Friar) on Apr 05, 2006 at 06:10 UTC
Good point; what about this? Lots debug() calls inside a long-running loop, inside many very frequently called subs, running in a virtual server program (with access to only 10% of the CPU), running on a commodity server (Pentium-III 1.3 Ghz), being accessed by many simultaneous clients. That's a lot different than the Benchmark I just ran on a Athlon64 X2 2.4ghz. Besides, what about the enjoyment of knowing that there is no debugging code in your program? --Dan	[reply]
Re^5: Optimize debugging subs by ikegami (Patriarch) on Apr 05, 2006 at 14:22 UTC
Besides, what about the enjoyment of knowing that there is no* debugging code in your program?* Enjoyment? In practice, I've found debug vs production optimizations a source of headaches. Also, I've found the production version is not sufficiently tested. People assume the production version will work because the debug version does.	[reply]