[rules-users] Efficiency of rules with XPath...

[Schwenker, Stephen]

Hey all,
 
I'm trying to make my rule more efficient but I'm not sure how.  I've
designed my rules based around xml objects.  Basically, I have these
documents and XPath queries are executed against the Object to determine
if the rule should be executed.  The problems is, is that we have over
900 rules that need to be executed.  The performance isn't really that
bad but I would like to Make it even better.
 
To do the job,  I have written an XmlUtils class to execute the
expression.  The class first gets the XPathFactory.  Then compiles the
expression and then executes the expression.  This happens for every
expression, and will happen each time the same expression is executed.
I want to make the rules more efficient by getting the XPathFactory once
and compiling the expression once and then execute the expression n time
on the same compiled object.
 
I would like to know if there is a way I can have the rule object that
is generated by the package build do the compile of the XPath Expression
and then use it instead of compiling it every time.
 
Here is a sample of one of my rule criteria.
 
    asset: AssetDocument()
    eval( XmlUtils.evaluatePathBoolean( asset,
"boolean(//ass:asset/ass:properties/ass:property[@name =
'Desk']/ass:value/text() = 'column')") )
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20070417/8e2a6961/attachment.html