From: "Simon Kröger" <SimonKroeger@...>
Date: 2006-05-23T05:57:41+09:00
Subject: Re: Zed and Luis drop the bomb on Ruby's poor performance

Sam Roberts wrote:
> Quoting SimonKroeger@gmx.de, on Tue, May 23, 2006 at 05:12:14AM +0900:
>> Hi Sam!
> 
> Hi Simon,
> 
>> Conclusion: method calls in general and object creation are the
>> bottlenecks in this (very, very) simple cases.
> 
> My conclusion as well.
> 
> My real app basically runs over all the lines in a set of files using
> IO#each, runs a regex, turns the matched strings into objects, so is
> pretty much dominated by these simple cases (I think, and profile
> suggests).
> 
> Your !-optimizations are interesting. I'll look back at my low-level
> code. Maybe by changing my "name = $2.upcase" statements to "name = $2;
> name.upcase!" I can get some good speedups. Depends on whether I need to
> keep both versions around, I'll have to construct some test cases and
> profile them.
> 
> Thanks,
> Sam

There were some performance tips posted here some time ago, keeping the
number of objects low is definitely a good start. I just don't like the
bang methods because you can't method-chain with them. (a lot of the
return nil if there is no change)

If the files aren't that big try to read them in one big read - should
be way faster.

Another one of the tips posted before was: use sqlite - and i can
wholeheartedly support that. If you have the chance to change your data
format to a sqlite database you can run a select on thousands of
lines in virtually no time. (and it's really, really painless to use)

cheers

Simon