>If I have a SAX stream & I feed it to XALAN & want
>a SAX stream back internally does XALAN convert my
>input document into a DOM tree
Actually, in that case we use our own data models, which are
lighter-weight than most DOM implementations.. but as you noted, we do
need an in-memory model because stylesheets may traverse the document in
any direction at any time.
Searching the archives of this list for "pruning" and "filtering" will
turn up comments about ongoing efforts to find ways to reduce the amount
of that model which is built and/or retained. There are some opportunities
for clever coding here, but achieving them is emphatically nontrivial and
the optimizations may work only for some limited subset of stylesheets.
For example, I believe our SQL Extension does _not_ maintain a full model
-- but for that reason, only some sequences of processing can be
successfully applied against it. The "holy grail" here is an optimizer
which analyses the stylesheet and finds opportunities to reduce the
working set, and a model (or models plural) which can respond to that
information.
Meanwhile, turning on incremental mode (see the "features" page of the
documentation) may help in some cases, if your stylesheet doesn't have to
examine the whole document to produce its results.
______________________________________
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.")
