> One way I can imagine an efficient way of using XSLT to
> select and transform a few nodes from a gigabyte XML
> document is indexing it - a database thing.

This is a very cool idea, and something I've been doing a variant of for a
while in SQL Server. Basically, I read my documents (caveat: no mixed
content, no ANY content, but elements and attributes are OK) and store the
information in my database in a branch-leaf metaphor, with the leaves
containing the two or three bits of information I'm interested in searching
on. Then, I can query my DB with something like "give me any leaf of type
'model' that has style='sedan' and whose parent has maker='Chrysler'." The
query pops out XPath expressions which I can then use from a
dynamically-generated XSLT stylesheet (built using the DOM) to grab just the
bits of information I want from the doc. In addition, if I'm only interested
in summary information - like just how many Chrysler sedans there are - I
don't have to go to the document at all anymore.

It remains to be seen whether Oracle and Microsoft (the two big RDBMS
players in my space) build stuff like this in natively - but it sure would
be nice if they did.

- Kevin

Kevin Williams
XML Architect
Ultraprise Corporation


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