From: Daniel Carrera <dcarrera@...>
Date: 2003-02-20T15:13:16+09:00
Subject: Re: optimization question

On Thu, Feb 20, 2003 at 02:58:08PM +0900, Bulat Ziganshin wrote:
> Hello Travis,
> 
> Thursday, February 20, 2003, 1:53:14 AM, you wrote:
> 
> TW> # This is ran about a million times
> TW> h = Hash.new(0)
> TW> data.scan(iptok).each do |tok|
> TW>   h[tok] += 1
> TW> end
> 
> are you heard about "sort|uniq" idiom?

I think that sorting is about as expensive as hashing.
Here is a quick benchmark for 5,000 iterations:

$ time ruby original_code.rb 

real    0m3.888s
user    0m3.520s
sys     0m0.020s

$ time ruby your_code.rb 

real    0m3.453s
user    0m3.140s
sys     0m0.040s

I still think that using C is a good idea.

-- 
Daniel Carrera
Graduate Teaching Assistant.  Math Dept.
University of Maryland.  (301) 405-5137