Re: New XSL Optimization

Subject: Re: New XSL Optimization
From: Paul Prescod <paul@xxxxxxxxxxx>
Date: Fri, 25 Jun 1999 22:29:18 -0400
S Manimaran wrote:
> 
> 1. First I extracted out the match patterns to be matched from the XSL
> stylesheet. Then the match patterns are converted to a context-free
> grammar whose language is the matching trees in pre-order linearized form.


You can do much better if you do not use a context free grammar and
instead use a regular tree grammar. Context free grammars are way, way,
way, more general than anything you can find in an XML document. Therefore
your algorithm (Earley) is more general than needed.

See http://www.prescod.net/forests/shorttut

I wonder, though, how much time is spent in match selection. My DSSSL
experience tells suggests that it is the code IN the rules that takes the
most time. Still, every little bit helps.

-- 
 Paul Prescod  - ISOGEN Consulting Engineer speaking for only himself
 http://itrc.uwaterloo.ca/~papresco

"Scarcely a discussion goes on, between linguistic groups or regions or 
even within cities and towns, which is not a struggle between competing 
myths of victimization."  
		- John Ralston Saul, "Reflections on a Siamese Twin"


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list


Current Thread