Re: Recursion performance (fibonacci numbers)

On 1/17/06, andrew welch <andrew.j.welch@xxxxxxxxx> wrote:
> > Or if you want timing information:
> >
> > java net.sf.saxon.Transform -t fibonacci2.xsl fibonacci2.xsl n=30
> >
> > though you need to be aware that for anything less than a second or two,
the
> > timings will be meaningless as they are dominated by Java warm-up time.
>
> If you use easytransformer (a graphical front end for Saxon) then the
> timings below 2 seconds should be more accurate than from the command
> line as it doesn't include java start up or stylesheet parsing - it
> starts only when the transformation begins (when the XML is parsed).
>
> http://sourceforge.net/projects/easytransformer
>
> (easytransformer uses SaxonB 8.6.1, so an XSLT 1.0 stylesheet would be
> running with an 2.0 processor, if that's significant.  In this case
> you would ideally give the root matching template a name and then use
> the "standalone" tab, so you dont have to apply the stylesheet to
> itself)

I'm not sure why I thought everyone was using XSLT 1.0......

To use a stylesheet like this (one where no input XML is required) its
best to give the the root matching template a name, eg:

<xsl:template match="/" name="main">

then it can be run either in the traditional way by applying it to
itself (or any other XML doc) or by specifying the initial template
with the -it switch:

java net.sf.saxon.Transform -t -it main fibonacci2.xsl n=30

The "standalone" tab in easytransformer can auto-detect the available
named templates and parameters.