recommended way to evaluate an xpath over 1000 docs?

[Christopher Giblin <CG...@zurich.ibm.com>]

hi

My program will have thousands of small XML docs ( ~1K) in memory.
I want to group documents according to whether they match a given xpath
expression.
The xpath expression will be applied to each, evaluating only for
match/no-match, that's it (can bail out once a hit is found).

Unfortunately I don't know much about optimizing Xalan for performance -
what would be the most efficient call sequence if I want to optimize on
memory? On speed? I may iterate through the 1000 docs many times.

While  I am aware of XPathAPI, its javadoc warns that it is intended for
convenience, suggesting it not being the best choice  for repetitive
evaluations. Any insights greatly appreciated!

thanks much, chris