The better approach is to use John Cowan's DOMParser

 (http://www.ccil.org/~cowan/XML/) (* thanks for the link Michael)


  InputSource domSource = new org.ccil.cowan.sax.DOMSource(yourDocument);

  Parser inputParser = new org.ccil.cowan.sax.DOMParser();

  Parser sheetParser = "... your favorite XML parser ...";

  XSLProcessorImpl xt = new XSLProcessorImpl();

  xt.setParser(inputParser, sheetParser);

  xt.parse(domSource);


 This will avoid the unnecessary 'flattening' and parsing.



> Do you mean that for small document (less than 10Kb), it's faster to
> transform DOM into byte[] and then reparse it  ? Did I understand
> well?

  Flattening the DOM tree onto byte[]s and reparsing it will always
be costly, what Dan was suggesting was to 'pipe' :

 (1)  DOM -> DOMWriter -> InputStreamWriter -> PipedOutputStream ->
PipedInputStream -> SAXParser

instead of  :

 (2)

  flattening :

  DOM -> DOMWriter -> InputStreamWriter -> byte[]

  and then inflating :

  byte[] -> ByteArrayInputStream -> SAXParser

  in two separate steps.

 (1) has the advantage of being memory efficient (the memory usage is the
size
     of the PipedInputStream's buffer).

  The overhead of (1) is that there is a that the threads are 'waiting' and
'notifying'
  a lot, but for big enough documents (2) will bust your memory.


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