commons-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dmitri Plotnikov" <dmi...@apache.org>
Subject Re: JXPath Performance
Date Fri, 06 Feb 2004 23:14:05 GMT
All this is absolutely correct.  I am using lower level APIs in my
integration.  However, that is work in progress and it is too early for me
to make any promises.

- Dmitri


----- Original Message ----- 
From: "Snehal Khanna" <snehal_haridas@yahoo.com>
To: "Jakarta Commons Users List" <commons-user@jakarta.apache.org>
Sent: Friday, February 06, 2004 4:54 PM
Subject: Re: JXPath Performance


> Thank you for this bit of input. My group had run
> tests using XPathAPI and the numbers were quite bad
> compared to JXPath. That is why I was curious to know
> what you would gain by integrating with Xalan. I had a
> hunch that CachedXPathAPI would probably what you had
> in mind.
>
> However it seems that CachedXPathAPI is good so long
> as the source document does not change. To compile
> XPath expressions to use over any document, one would
> need to use some other low level APIs.
>
> Regards
> Snehal
>
> --- Dmitri Plotnikov <dmitri@apache.org> wrote:
> > It is a little bit like comparing apples and
> > oranges:
> >
> > Xalan is an incredible XSLT processor. Its
> > implementation of XPath is
> > really, really fast (just don't use XPathAPI, use
> > CachedXPathAPI instead).
> >
> > JXPath is a highly flexible interpreter that applies
> > the XPath syntax to
> > mixed data models: it is not designed to be the
> > fastest on DOM structures.
> > Rather, it is optimized for general object graph
> > traversal.
> >
> > To summarize: if you need a high-performance XPath
> > processor for DOM, go
> > with Xalan.  If you need to traverse heterogeneous
> > object graphs, go with
> > JXPath. 'Simple as that.
> >
> > I hope this helps.
> >
> > - Dmitri
> >
> > ----- Original Message ----- 
> > From: "Snehal Khanna" <snehal_haridas@yahoo.com>
> > To: "Jakarta Commons Users List"
> > <commons-user@jakarta.apache.org>;
> > <dmitri@apache.org>
> > Sent: Friday, February 06, 2004 4:24 PM
> > Subject: Re: JXPath Performance
> >
> >
> > > Hi Dmitri,
> > >
> > > Have you done any tests comparing the performance
> > of
> > > JXPath to Xalan? It is imperative in my project to
> > use
> > > a really fast XPath engine so, it would be good to
> > > know how much gain in performance we are talking
> > about
> > > here - especially for the case I mentioned
> > earlier.
> > >
> > > Regards,
> > > ~Snehal
> > >
> > >
> > > --- Dmitri Plotnikov <dmitri@apache.org> wrote:
> > > > JXPath is designed to work with heterogeneous
> > > > models, including
> > > > JavaBeans, Maps, Collections etc, as well as
> > DOM,
> > > > JDOM, etc.  To
> > > > accomodate all this variety, JXPath inevitable
> > > > produces some overhead.
> > > > For example it creates a bunch of NodePointer
> > > > objects during the
> > > > evaluation of an XPath.  You wouldn't need such
> > > > objects if you only
> > > > worked with XML (DOM, JDOM).  Thus, JXPath is
> > slower
> > > > than for instance
> > > > Xalan, which manages to allocate almost zero
> > > > objects.
> > > >
> > > > I am currently working on some behind-the-scenes
> > > > integration between
> > > > JXPath and Xalan.  Once that's complete, JXPath
> > > > performance will be
> > > > similar to that of Xalan, as long as you are
> > working
> > > > exclusively with
> > > > XML.
> > > >
> > > > I hope this helps.
> > > >
> > > > - Dmitri
> > > >
> > > >
> > > > --- Snehal Khanna <snehal_haridas@yahoo.com>
> > wrote:
> > > > > We are in the process of evaluating two
> > > > > implementations of XPath namely JXPath and
> > > > jd.xpath
> > > > > from www.aztecrider.com
> > > > >
> > > > > In our evaluations we have noticed that for
> > XPath
> > > > > expressions resulting in a node set with 100
> > > > nodes,
> > > > > the performance of jd.xpath is superior to the
> > > > > performance of JXPath. It looks like JXpath
> > loses
> > > > out
> > > > > on performance while iterating through the
> > > > results.
> > > > >
> > > > > For XPath expressions resulting in a single
> > node
> > > > > output, the performance is comparable if we
> > use
> > > > > compiled expressions.
> > > > >
> > > > > Has anyone else performed a similar analysis?
> > Are
> > > > > there any known performance issues with JXPath
> > > > that
> > > > > are currently being addressed? In general what
> > is
> > > > the
> > > > > development plan for JXPath in terms of
> > > > performance
> > > > > improvements and enhancements?
> > > > >
> > > > > Regards,
> > > > > Snehal
> > > > >
> > > > >
> > > > > __________________________________
> > > > > Do you Yahoo!?
> > > > > Yahoo! SiteBuilder - Free web site building
> > tool.
> > > > Try it!
> > > > > http://webhosting.yahoo.com/ps/sb/
> > > > >
> > > > >
> > > >
> > >
> >
> ---------------------------------------------------------------------
> > > > > To unsubscribe, e-mail:
> > > > commons-user-unsubscribe@jakarta.apache.org
> > > > > For additional commands, e-mail:
> > > > commons-user-help@jakarta.apache.org
> > > > >
> > > >
> > > >
> > > >
> > >
> >
> ---------------------------------------------------------------------
> > > > To unsubscribe, e-mail:
> > > > commons-user-unsubscribe@jakarta.apache.org
> > > > For additional commands, e-mail:
> > > > commons-user-help@jakarta.apache.org
> > > >
> > >
> > >
> > > __________________________________
> > > Do you Yahoo!?
> > > Yahoo! Finance: Get your refund fast by filing
> > online.
> > > http://taxes.yahoo.com/filing.html
> > >
> > >
> > >
> >
> >
> >
> >
> ---------------------------------------------------------------------
> > To unsubscribe, e-mail:
> > commons-user-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail:
> > commons-user-help@jakarta.apache.org
> >
>
>
> __________________________________
> Do you Yahoo!?
> Yahoo! Finance: Get your refund fast by filing online.
> http://taxes.yahoo.com/filing.html
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: commons-user-help@jakarta.apache.org
>
>
>



---------------------------------------------------------------------
To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-user-help@jakarta.apache.org


Mime
View raw message