Normally, Xalan will read the full document into storage (into a DTM rather than a DOM, if we're operating from a SAX stream) before processing it.
If you set the incremental flag (see http://xml.apache.org/xalan-j/dtm.html#incremental), we will "hold off" the incoming SAX stream as much as possible, building the model only as needed and only as far as is needed. This may reduce latence (start generating output sooner) and memory use (_if_ your stylesheet doesn't reference the whole document), but it will probably also reduce throughput (take longer to finish producing all the output) slightly, ______________________________________ Joe Kesselman / IBM Research
