Hello,
For what it's worth, just before the Xerces-J 2.0.0 release I ran some tests that showed a significant difference between a parse with the schema validator in the pipeline and one without it in the pipeline, even when there was no schema grammar to validate.
The results I saw were roughly 25-50% worse with the idle schema validator in the pipeline than with it removed from the pipeline.
Thanks,
Henry
------------------------------------------------------------------
Henry Zongaro XML Parsers development
IBM SWS Toronto Lab Tie Line 969-6044; Phone (905) 413-6044
mailto:[EMAIL PROTECTED]
| Andy Clark <[EMAIL PROTECTED]>
02/04/01 12:31 PM
|
To: [EMAIL PROTECTED] cc: Subject: Re: XNI optimized parser config? |
Dennis Sosnoski wrote:
> Does anyone have a sample of how to configure an optimized version of
> the Xerces2 parser? I'd like to try it out with validation disabled to
> see how much difference this makes to performance.
There's an XNI sample that builds a parser configuration
without any validators in the pipeline. So you could use this
to see if there's a noticable performance difference. Check
in the samples/xni/parser directory.
However, more can be done with a little programming. For
example, if the documents being parsed are generated by
machine and are guaranteed to be well-formed, then a much
smaller scanner could be written that doesn't have to check
all of the WF constraints in the XML Specification. But the
code to do this hasn't been written.
--
Andy Clark * [EMAIL PROTECTED]
