I tried to optimize Ruby a bit more. The results vary quite a lot with
benchmarks,
but typical improvement is about 5% less time and 4% less memory.
More about it:
* http://t-a-w.blogspot.com/2007/02/ruby-and-judy.html
And a patch against Ruby 1.8.2-p12:
* http://zabor.org/taw/ruby-performance-patch-judy-2007-02-26.patch.gz
--
Tomasz Wegrzanowski [ http://t-a-w.blogspot.com/ ]
I wrote a naïve patch to use Judy for ruby's internal tables in 2002; unlike
yours, it replaced all st_tables IIRC, but some of the feedback from Judy's
author might still apply, see [48365].
matz took a look at Judy last year, see
http://www.rubyist.net/~matz/20060628.html
He seemed pretty excited about it, but the license (LGPL) would make
integration a bit cumbersome.
Normally, a patch like this should be tested in HEAD before it gets in the 1.8
branch, but it's knu's call, of course.
PS. please don't include all of configure when just configure.in will do 
On Tue, Feb 27, 2007 at 10:31:49AM +0900, Tomasz Wegrzanowski wrote:
I tried to optimize Ruby a bit more. The results vary quite a lot with
benchmarks, but typical improvement is about 5% less time and 4% less
memory.More about it:
* taw's blog: Ruby and Judy
And a patch against Ruby 1.8.2-p12:
* http://zabor.org/taw/ruby-performance-patch-judy-2007-02-26.patch.gz
--
Mauricio Fernandez - http://eigenclass.org - singular Ruby
** Latest postings **
More logical programming in Ruby: solving Einstein's riddle (Zebra puzzle)
http://eigenclass.org/hiki.rb?einstein-riddle-solved-in-ruby
Bounded space memoization, another new twist
http://eigenclass.org/hiki.rb?bounded-space-memoization