Re: [xmlblaster] Which XPathFilter is used

[Marcel Ruff <mr at marcelruff.info>]

Stefan Lischke wrote:
> Hi there,
>
> I'm just wondering which XML Filter technique is used within xmlBlaster.
> You guys use FSM (finite state machines) like used in XFilter or YFilter 
> or you have your own way?

For the internal DOM we use com.fujitsu.xml.omquery.DomQueryMgr:
"XPath Version 0.92.2 from Takuki Kamiya, Fujitsu Limited, 2000-april
 XPath interface for XT provides for DOM query API facilities on top of XT.
 It is implemented in Java language and complies with W3C's XPath Proposed
 Recommendation as is currently implemented in XT."

For the additional filter plugins ('full text search') there is
one from Peter Antman:
 http://www.xmlblaster.org/xmlBlaster/doc/requirements/mime.plugin.access.xpath.html

You can easily implemented other plugins.

> Can I filter with the XPath 1.0 compliant expressions or is it reduced 
> to a subset of XPath (without some axes).
>
> You have any actual throughput measurements?
> Is your throughput depending on the number of Filters? I found an 
> independing approach which has over 1 Million Xpath filters and reached  
> a throughput of 5.4MB/s

Just write a little testcase for your problem domain.
If you find solutions to improve bottlenecks please
report them to us.
Your above mentioned filter approach you can probably
easy embed into an xmlBlaster mime plugin.

regards,

Marcel

> thanx a lot for further informations
>
> Stefan


--
http://www.xmlBlaster.org