Schema grammar chaching for Xerces1

[Vl...@bmo.com]

I have a huge XSD schema file (around 500Kb) which I need to use to 
validate a stream of XML requests (in OFX format). It takes about a minute 
to finish validation of one request on my PC. I tried to do schema file 
cashing by implementing EntityResolver, but it didn't make much difference 
- parsing was still too long (about 40 sec).
I modified method resolveSchemaGrammar() of XMLValidator class to cache 
schemas as parsed XML documents and even that didn't help, because 
creating of TraverseSchema(...) is taking too long (~ 20 sec).

Has anyone tried to implement grammar caching for Xerces1?

Another question:
Is it possible to include Ken's RevalidatingDOMParser.java into the next 
release of Xerces1, so that people wouldn't have to add it to every new 
release of Xerces1?

I appreciate any comments.

Regards,
Vlad