From: Michael Neumann <neumann@...>
Date: 2001-10-30T05:30:06+09:00
Subject: [ruby-talk:23749] Re: Some inspirations from REBOL

Ryan Leavengood wrote:
> > Maybe great, but very slow. Why not make it a standard C-extension and
> > distribute it with Ruby? It would run on any platform.
> 
> Yes that is certainly a possibility.  But there is no guarantee it will
> run on any platform.  But I suppose if Ruby could be ported to that
> platform then that library could.  But pure Ruby is always more portable
> than C.
> 
> > I've implemented LZ77, LZ78, LZW, RLE and BWT (bzip) in pure Ruby and
> > they are really slow.
> 
> How slow is really slow?  Does it take seconds to process small data or
> is it just a lot slower than C?  Because sure, if you compare it to
> highly optimized compiled code it will be a lot slower, especially if
> you write it inefficiently in Ruby (for example using += instead of <<
> with Strings.)  But if it can process a meg a second or so then I think
> that would be sufficient for most people.

Well, I did not spent much time with optimizing it. But compressing a
5.3 MB large file with a blocksize of 4000 bytes using BWT (without any
other compression algorithm like RLE) took on my PC (Athlon 1000) 9 min.

bzip2 uses a block size between 100,000 and 900,000 bytes and compresses
the same file in only 5.7 (bs 100,000) or 7.5 (bs 900,000) seconds. 

Using a block-size of 500 bytes with the Ruby program speeds the whole
up by factor 2.5 (4 min), so the Ruby implementations largely depends on
the block-size.

But I must also note that the BWT compress (better: sort) routine is
only 9 lines long (decompress is 16).

Regards,

  Michael

-- 
Michael Neumann
merlin.zwo InfoDesign GmbH
http://www.merlin-zwo.de