Re: Xalan performance

Subject: Re: Xalan performance
From: "Blond Jean-Baptiste" <pblond@xxxxxxxxxxxxxxxx>
Date: Tue, 14 Nov 2000 16:59:53 +0100

Maybe you have to work on your stylesheet, too... I got one I did optimize a
few times. The last version is almost 70 times faster than the first... with
almost the same functionnality (from 17-18 s to 260 ms)!

----- Original Message -----
From: Peter McEvoy <peter.mcevoy@xxxxxxxx>
To: Harbaugh, Alan (FUSA) <AlanHarbaugh@xxxxxxxxxxxx>
Cc: <XSL-List@xxxxxxxxxxxxxxxx>
Sent: Tuesday, November 14, 2000 4:05 PM
Subject: RE: Xalan performance

> > Peter,
> >   We have a very similar situation.  We are about to
> > investigate using XT.  Our preliminiary evaluation is that it
> > is MUCH faster.  We need to look into whether it has an
> > API and how well it conforms to the standard.
> Alan,
> Thanks for your feedback - it's kinda reassuring that others are seeing
> this as an issue. Today I've actually managed to do quite a bit of
> and have investigated already:
> XT - certainly an alternative, but they freely admit that it was not built
> with performance in mind.  They have a Java API and it is possible to
> a servlet to use it.
> Saxon - I am currently looking at this - it really does look like it is
> about 400% faster than Xalan.  It has a full java API, and they even have
> sample servlet which does transforms (although the servlet needs a bit of
> work to be as fully featured as DefaultApplyXSL).  From initial
> I think Saxon will take about 3+ seconds to do the transform that I have.
> Sablotron - this is a C++ library which SCREAMS along.  I will investigate
> next how I can make JNDI calls to the librabry file and transform my
> FYI, my 14second Xalan transform takes 0.5 seconds in this (yup, I said
> half of a second...!!!).  However, they don't fully implement the XSL
> standard yet.
> Translets - From Sun - will look at this.  You can compile your XSL sheets
> into classes and just run XML through it.  They say it is 3-10 times
> than XT.  See
> There were also a few commercial ones as well, I'd like to exhaust my
> options on the open source side and then start calling companies.  These
> are:
> DataPower - some XSL acceleration technology.  They have some interesting
> other stuff, though - worth a nosy.
> Non-runners:
> Cocoon - this just uses the Xalan jar anyway.
> I tried finding stuff on, and it seems that they too
> their XSL engine on Xalan.
> Thanks to everone who mailed me with tips - this seems to be a hot
> Peter McEvoy
> Senior Technical Analyst
> IONA Technologies PLC
>  XSL-List info and archive:

 XSL-List info and archive:

Current Thread