From: E S <eero.saynatkari@...>
Date: 2004-12-31T10:35:27+09:00
Subject: Re: seeking hints for fast io

> L�hett�j�: jm <jeffm@ghostgun.com>
> P�iv�ys: 2004/12/29 ke PM 05:04:06 GMT+02:00
> Vastaanottaja: ruby-talk@ruby-lang.org (ruby-talk ML)
> Aihe: seeking hints for fast io
> 
> I have a class which holds data while it's being manipulated. The 
> trouble is that I call to_s() quite a few times printing the resulting 
> data to a file. This increases the run time of the script a surprising 
> amount raising the runtime from about 6 minutes to about 8 and a half. 
> The code looks roughly likely the following,
> 
> class Example
> .
> .
>   def to_s
>     sout << "a=#{@a}"
>     sout << "b=#{@b}"
>     sout.join("\n")
>   end
> end
> 
> # in main
> File.open(outfile,File::CREAT|File::RDWR|File::APPEND) do |fp|
>   the_data.each do |d|
>     fp.puts d
>     fp.puts   # add blank line
>   end
> end
> 
> Anyone care to suggest a way to speed this up? Is it faster to
> 
>   sout << "a=" + @a
> 
> instead of
> 
>    sout << "a=#{@a}"

So are you doing something like this now?

s << "a=" << @a << "\n" << "b=" << @b

You could try to just interpolate, which may or may not be faster.
There's no reason to have an explicit String variable, though, if
you're not storing it:

def to_s
 "a=#{@a}\nb=#{@b}\n" # <- the 'empty line'
end

That's probably as fast as you're going to get. You say the data
is being manipulated so it's probably not possible for you to
prewrite/cache whatever @a and @b may be? The best thing you can
do is not write the data object by object like you're doing now,
but not knowing what you're *actually* doing we can't really offer
any concrete optimization advice.

> Jeff.

E