Hello everyone,

Users have been reporting performance regressions in XalanJ 2.4.0 compared
to XalanJ 2.3.1 and other earlier releases.   Some of these have already
been fixed in the current CVS or have patches attached.  For example:

Bug #11661 [http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11661]:
xsl:key very slow!
Bug #13299 [http://nagoya.apache.org/bugzilla/show_bug.cgi?id=13299]
Transformation very slow compare to xalan 2.2D11
Bug #13501 [http://nagoya.apache.org/bugzilla/show_bug.cgi?id=13501]
Serious performance regression on select-child-last

We're also planning to review:

http://nagoya.apache.org/bugzilla/show_bug.cgi?id=12486: XPathContext.reset
() does unnecessary memory allocations
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=12481 Method
DTMManager.newInstance  causes unnecessary memory consumption
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=12687 Class
org.apache.xml.dtm.ref.ExpandedNameTable allocates String pools, but never
uses it

The XalanJ team is planning to have a XalanJ 2.4.1 release to bring the
performance back in line with (or better than!) earlier releases.  Since
several patches are already in, we are targetting this for end of next
week.

If anyone knows of other open bugs, related to performance regressions,
that should be included, let us know!

Oh, while we're at it, we'll  move up to  XercesJ 2.2.0.

Ilene.

Reply via email to