Re: Array containment

Wed, 11 Feb 2004 06:58:08 -0800

On Feb 11, 2004, at 12:14 AM, Jan Eden wrote:

James Edward Gray II wrote:

On Feb 10, 2004, at 2:39 PM, Jan Eden wrote: 

Rob, I read the perlfaq paragraph you mentioned and found that it
proposes a solution which does not make use of 'exists':


What's wrong with exists()?  I like exists() and you're going to hurt
it's feelings.  :)

It has six characters which can be left out to minimize typing.

All of the solutions posted added more than six characters, so you'll need multiple tests before you start saving keystrokes. Make sure you're lazy in the good way. ;)

You could also use map() I guess, if you don't consider that a loop:

@[EMAIL PROTECTED] = map { 1 } 0..$#blues;

For some irrational reason, I consider this more of a loop than your first/Rob's solution below.

map() is a loop, to be sure, but it comes from inside perl and thus should be faster much of the time than a loop you create yourself. (Compare 'foreach' and 'map_hash' in the results below.)

It's not for me to rewrite the docs (or perhaps it is?) but

for (@blues) { $is_blue{$_} = 1 }

does the same thing as

@[EMAIL PROTECTED] = (1) x @blues;

I finally came up with an idea of my own, just when I got up this morning:

@[EMAIL PROTECTED] = 1 .. @blues;

This way each of the hash keys gets a value different from zero, so

if ($is_blue{$_})

still works.

Clever.

Now it might be hard to determine which of these two is faster.

Why would that be hard? ;)

#!/usr/bin/perl

use strict;
use warnings;

use Benchmark;

my @blues = qw/azure cerulean teal turquoise lapis-lazuli/;
my %is_blue;

timethese( 3000000, {
        foreach         => '$is_blue{$_} = 1 foreach @blues',
        hash_slice      => '@[EMAIL PROTECTED]',
        all_ones        => '@[EMAIL PROTECTED] = (1) x @blues',
        map_ones        => '@[EMAIL PROTECTED] = map { 1 } 0..$#blues',
        map_hash        => '%is_bllue = map { $_ => 1 } @blues',
        range           => '@[EMAIL PROTECTED] = 1 .. @blues'
} );

__END__

And the results, from my G5:

Benchmark: timing 3000000 iterations of all_ones, foreach, hash_slice, map_hash, map_ones, range...
all_ones: 1 wallclock secs ( 1.76 usr + 0.00 sys = 1.76 CPU) @ 1704545.45/s (n=3000000)
foreach: 3 wallclock secs ( 2.39 usr + 0.00 sys = 2.39 CPU) @ 1255230.13/s (n=3000000)
hash_slice: 1 wallclock secs ( 0.94 usr + 0.00 sys = 0.94 CPU) @ 3191489.36/s (n=3000000)
map_hash: 2 wallclock secs ( 1.43 usr + 0.00 sys = 1.43 CPU) @ 2097902.10/s (n=3000000)
map_ones: 4 wallclock secs ( 3.91 usr + 0.00 sys = 3.91 CPU) @ 767263.43/s (n=3000000)
range: 1 wallclock secs ( 2.20 usr + 0.00 sys = 2.20 CPU) @ 1363636.36/s (n=3000000)

Looks like Rob is still the man to beat. :D

James


--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
<http://learn.perl.org/> <http://learn.perl.org/first-response>

Reply via email to