Re: Re: Up for Critique

[biograd]

There were a couple reasons I went for the wholesome all-array approach...one of them is I have just never been as comfortable with hashes as I have been with arrays. This is no excuse, of course, and I've been meaning to use them more. I just always see the array algorithm first. (Can you tell I'm new???) Second, I'd read in Mastering Algorithms with Perl about the speed of arrays over hashes, so I felt justified. (Webadept noted this below too.)

However, I can see the utility of hashes from many of these examples, yours and others below, so I know I'll have to bite this bullet soon. Thanks for the example, tested or not. :)

-Jenn

[jeffa]

Speed vs. Maintainability/Scalability

If the difference in speed is small, i say drop the arrays and use hashes ... but i am lazy. I have this adversion to something called 'typing'. ;)

But, arrays are not always faster than hashes. It all depends upon context - if you have to scan the entire array to find something, a hash implementation will probably be faster. But if you have to iterate over all elements - then an array is probably the better choice. Consider the following Benchmark code:

use strict;
use Benchmark;

my %h;
my @a  = ('a'..'zzz');
@h{@a} = (1) x @a;

timethese('-10', {
   find_array    => sub { return (grep /^ccc$/, @a) ? 1 : 0 },
   find_hash     => sub { return $h{'ccc'} },
   iterate_array => sub { do {} for @a },
   iterate_hash  => sub { do {} for keys %h },
});


__END__
yields on my dual proc 400 linux box: (YMWV)
find_array:
10 wallclock secs (10.11usr + 0.04sys = 10.15 CPU) @ 45.62/s (n=463)

find_hash:
13 wallclock secs (11.41usr + 0.07sys = 11.48 CPU) @ 1183591.72/s (n=1
+3587633)

iterate_array:
11 wallclock secs (10.98usr + 0.09sys = 11.07 CPU) @ 70.28/s (n=778)

iterate_hash:
11 wallclock secs (10.78usr + 0.05sys = 10.83 CPU) @ 16.53/s (n=179)

jeffa

L-LL-L--L-LL-L--L-LL-L--
-R--R-RR-R--R-RR-R--R-RR
B--B--B--B--B--B--B--B--
H---H---H---H---H---H---
(the triplet paradiddle with high-hat)