Thank you both for your answers. They are very helpful. My thinking of our implementation would be quite similar to what you describe, Santiago. We have a series of XSL templates which would be cached until a change is made. If, as you suggest, we can possibly gain roughly a 100% increase in performance, that would surely be worthwhile.
The sad part is that XSLTC does not support xalan:component and xalan:script or Extension element. Though we don't currently use either of these, I'd recently done some prototyping to provide several extension elements. It would have been nice to gain the benefit of XSLTC on top of that. Thanks again, Ray On Tue, 2007-09-11 at 10:08 -0400, Santiago Pericas-Geertsen wrote: > > > On Sep 11, 2007, at 8:59 AM, Erin Harris wrote: > > > > > > > Hi Raymond, > > > > 1) If you are looking for conformance test results then I can > > probably dig something up but I don't think there is anything for > > performance. > > > > > > Our internal testing shows that on average XSLTC is 2-2.5X faster > than Xalan over a set of about 40 stylesheets. The benchmark simulates > a server scenario where stylesheets are precompiled (using Templates) > and ran over a period of time in which a throughput measure (TPS - > transforms per second) is computed. In this type of scenario, I'd > expect XSLTC to easily outperform Xalan. But as the saying goes: "your > mileage may vary". > > > -- Santiago > > > > > > 2) If you go to the following page it details the limitations of > > XSLTC extension support: > > http://xml.apache.org/xalan-j/extensions_xsltc.html > > > > Thanks. > > > > Erin Harris > > > > > > > > > > Raymond <[EMAIL PROTECTED]> > > > > 07/09/2007 07:19 PM > > > > > > To > > xalan-j-users > > <xalan-j-users@xml.apache.org> > > cc > > > > Subject > > translets > > > > > > > > > > > > > > > > > > Hey All, > > > > 1) Are there any statistical analysis between interpretive and > > "translet" driven XSL processing? > > > > Even a comparison: > > > > Xalan XSL Interpreted VS. Xalan XSLTC Translet > > > > would be great. > > > > 2) Are there limitations in the support of "exslt" when XSLTC > > Translets? > > > > Thanks, > > > > Raymond Auge <[EMAIL PROTECTED]> > > Software Engineer > > Liferay, Inc. > > Enterprise. Open Source. For Life. > > > > > > > > Raymond Auge <[EMAIL PROTECTED]> Software Engineer Liferay, Inc. Enterprise. Open Source. For Life.
