This is the mail archive of the xsl-list@mulberrytech.com mailing list .


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

RE: Measuring performance


> What is the ideal way to measure the performance of a specific bit
> of XSL code?  Are there any tools available for determining the
> areas of your XSL code that are causing the heaviest loads on the
> parser?

Saxon now has a hook, the TraceListener, that could be used to plug in
performance monitoring tools, but it would be rather laborious, and possibly
not than effective, since Java only seems to measure time in milliseconds (I
yearn for my old mainframe where I could get accurate counts of machine
instructions executed)

> Also, there are a few resources that indicate the efficiency
> of using certain commands over others (ie using count instead of
> xsl:number for certain applications).  Is there any place that
> provides a collection of these helpful suggestions?

Not that I'm aware of. Usually, you'll get the right answer by assuming that
the processor implements the algorithm given in the spec fairly literally,
without optimisation. Any optimisations are likely to be very processor (and
version) dependent - for example Saxon tries when it can to avoid n-squared
performance when using an <xsl:number level="single"> instruction to number
a sequence of nodes, but does it the hard way if there are any attributes
such as count and from.

Mike Kay


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]