RE: transform optimization for a schema-constrained do

I became curious about this because for a each type of output I produce, I 
have a single stylesheet that transforms several types of documents.  
In most cases, the generic stylesheet is very readable and maintainable but
very inefficient.  I could optimize performance by creating a different 
stylesheet for each type of input document, but that would multiply the 
creation and maintenance overhead on the stylesheets.  Hence my desire to 
automatically optimize a transform for a given schema.  Are there any other 
tricks that help in this situation?

David

-----Original Message-----
From: Michael Kay [mailto:mhkay@xxxxxxxxxxxx]
Sent: Wednesday, July 25, 2001 12:13 PM
To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
Subject: RE:  transform optimization for a schema-constrained
domain


The Tamino XML database (from Software AG) uses schema knowledge to optimize
queries against a database (the queries are currently written in a language
that's somewhere between XPath and XQuery). But this is a rather different
scenario, because the database can do a lot of work at document loading time
to set up metadata that's useful at query time. An XSLT processor generally
has only raw XML as its input, so there's a real risk that optimisation time
will exceed transformation time - especially if it means reading and
analysing a schema.

Another problem is that you can't bind a stylesheet to a particular schema
at the time it is compiled - though I guess one could get around that.

In short, I don't think anyone is doing this in an XSLT processor today. I
might be wrong, of course.

Mike Kay

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