>Xalan constructs an internal model of the whole document to
>perform its transformation on,

True, though we build only as much as is referenced and hope to eventually
prune away parts that are no longer needed.

>this takes up lots of memory (more than the input file).

Not necessarily more. We maintain only a single copy of each node name or
element name, for example, which can be a significant compression in large
documents.

Reply via email to