>Is it possible to come up with a version of XSLTC
>that generates translets that can process SAX events
>directly, rather than first buidling a DOM tree

XSLT requires random access to all parts of a document (consider XPaths such as //foo/ancestor::bar/baz, or the classic use case of styling an index, the body of the document, and then a table of contents). This generally means it requires a stored document model.

Some limited use cases can operate off a reduced data model, and there is ongoing interest in finding ways to recognize and optimize for those cases. Search archives of this discussion for the keywords "streaming", "pruning", and "filtering" for some discussions of those opportunities and the challenges of implementing them.

______________________________________
Joe Kesselman, IBM Next-Generation Web Technologies: XML, XSL and more.  
"may'ron DaroQbe'chugh vaj bIrIQbej"  ("Put down the squeezebox and nobody gets hurt.")

Reply via email to