RE: [xsl] Performance with Saxon and Kernow.

Subject: RE: [xsl] Performance with Saxon and Kernow.
From: "Fabien Tillier" <f.tillier@xxxxxxxx>
Date: Wed, 1 Sep 2010 08:58:48 +0200
Hi Michael.
Thanks for your input.
My (so far) only concern is that I am using DOM everywhere to store XML
Documents, and I don't know, for testing purpose, how to create the TinyTree
from a DOM document easily.
If there is no simple way, I may consider getting to tinytree directly (but
lots of code refactoring needed to get rid of DOM tree, thus I need to be sure
I will get better perfs).

Best regards,
Fabien


-----Message d'origine-----
De : Michael Kay [mailto:mike@xxxxxxxxxxxx]
Envoyi : mardi 31 ao{t 2010 17:54
@ : xsl-list@xxxxxxxxxxxxxxxxxxxxxx
Objet : Re: [xsl] Performance with Saxon and Kernow.

  I generally tell people that running Saxon using its native tree model
(TinyTree) is 4-10 times faster than running the same transformation on
a DOM, so if you are getting a ratio of 2, you are doing quite well.

If you want performance with Saxon, don't use a DOMSource.

Michael Kay
Saxonica

On 31/08/2010 4:28 PM, Fabien Tillier wrote:
> Hi List.
> I know some very knowledgeable people are members of this list, so I
> will ask a question that is not really directly an XSL one.
> I am using Kernow (1.6.1, Saxon 9.1.0.3 Basic) for testing my XSL
> Transforms. When using the same XML file and XSL template, the
> transformation is done in 31 seconds.
> When I use Saxon (9.2 HE) in my Java Servlet (within Tomcat), The same
> transformation is measured at 65 seconds (same computer, everything is
> local).
> The only difference is that the source is a file in the first case, and
> an XML Document in the second (so StreamSource (I assume in Kernow) in
> the first case and DOMSource in the second).
> I am using javax.xml.transform.Templates for caching XSL, but it doesn't
> improve things...
> Anything I should have a look at ?
>
> Thanks in advance,
> Best regards,
> Fabien

Current Thread