is this a reasonable driver?

[Niko Matsakis <ni...@datapower.com>]

Hello.  I'm attempting to develop a Xerces-C driver for some internal 
benchmarking of various parsers and (in particular) schema validators.
Let me first apologize for e-mailing this list even though I am not a
member, since I only have one question it didn't seem like subscribing was
necessary.

I've got a driver which seems to work for Xerces-C, but I'm not sure
whether the performance could be improved by tweaking the way I'm using
the API and I wanted to give you guys a chance to point at any mistakes
I might be making.

In particular, I have scanned the archives for speed related concerns and
noticed some people mentioned reusing the parsed DTD information.  I am not
validating against a DTD, but I thought that perhaps there might be a similar
flag to prevent unnecessary W3C Schema work when the same schema is being
used over and over, which is certainly the case here.  

The program takes a few commands as input on stdin which specify the schemas 
that are being validated against (and their namespaces) and the input files
to validate.  All of that parsing is taken care of by the base class.

Nothing is timed except for the xsdVerifyInput() function.  I seem to get
comparable performance for both DOM and SAX methods, which is somewhat 
surprising since the SAX handlers don't actually do anything.  I suppose that
this could indicate that schema validation is the bottleneck.

Anyway, I have attached the driver source as two files: one for the
driver itself, and the other is the base class that provides much of the core
driver functionality and which will be required to successfully build the 
driver.



thank you,
Niko Matsakis