From: Florian Frank <flori@...>
Date: 2005-06-22T08:23:52+09:00
Subject: Re: 1. Ruby result:  101 seconds , 2. Java result:9.8 seconds,          3. Perl result:62 seconds

Glenn Parker wrote:

> Well, I think it's valid to make the comparisons every once in a 
> while, since it highlights a weakness that Ruby implementations will 
> have a hard time overcoming on their own.  When you lag behind Perl by 
> a factor of two on *everything*, then you have a real problem.

Not in everything, only in the stupid "comparisons" and "benchmarks" the 
trolls keep posting here. Compare the Ruby version

(flori@lambda:~ 0)$ time ruby -e 'class A; def initialize(a,b) @a, @b = 
a, b end ; attr_accessor :a, :b; end ; 1_000_000.times { o = A.new(1, 
2); o.a; o.b; o.a = 2 ; o.b = 2 }'

real    0m3.838s
user    0m3.822s
sys     0m0.009s

to the wonderful Perl version here:

(flori@lambda:~ 0)$ time perl -e 'package A; sub new { my $p = $_[0]; my 
$c = ref $p || $p; bless { "a" => $_[1], "b" => $_[2] } => $c } sub a { 
@_ == 1 ? $_[0]->{a} : ($_[0]->{a} = $_[1]) } sub b { @_ == 1 ? 
$_[0]->{b} : ($_[0]->{b} = $_[1]) }; package main; for (1..1_000_000) { 
my $o = A->new(1, 2); $o->a ; $o->b; $o->a(2); $o->b(3) }'

real    0m9.420s
user    0m9.420s
sys     0m0.002s

And this doesn't include the qualitytime you can spent debugging Perl, 
while tracking down cyclic references in your object graph, or finding 
missing reference counter decreases in your extensions.

-- 
Florian Frank