I've posted the fourth beta of XOM, my absolutely correct, open source, dual tree-streaming API for procesing XML with Java.
This release is still backwards compatible with 1.0d25. However it does
deprecate four XSLTransform constructors that
will be removed in the next release. The upside is that XSLT
transformation should now be much faster and mmore
memory efficient since XOM
is no longer writing documents out into strings and passing those to the
transform. Instead it is streaming SAX events directly into the
transformer. Furthermore, implementing this exposed several bugs in the
SAXConverter (and indirectly one in the DOMConverter) that have now been
fixed. I've also fixed some minor cosmetic bugs innvolving serialization
of internal DTD subsets.
I've also completed the first draft of the XOM tutorial (XHTML compliant browser required). At this point I'm open for any comments anyone has, ranging from spelling errors, grammatical mistakes and code bugs to missing topics and poorly explained subjects.
For what it's worth, I've also updated the future directions document with some more thoughts about what may or may not appear in XOM post-1.0. Finally, I've posted some instructions for building XOM that should assist anyone who wants to build this from source.