Subject: Re: [xsl] measuring performance of XSLT stylesheet From: Florent Georges <lists@xxxxxxxxxxxx> Date: Sun, 22 Nov 2009 20:23:21 +0000 (GMT) |
a kusa wrote: > What little I know about using "//" is that the processor > checks a whole document when a "//element" is used. If we are > talking about a really big document (few mB) this would > definitely slow down the transformation processing time. > Right? No. There is nothing telling so in XDM, and when it comes to optimization, that does not mean anything to "theorize" about possible behavior. For an in-memory processor like Saxon, that reasoning is usually right, but I think Saxon has several optimizations for //, so maybe in some cases that could be the opposite. For an XML database in general that will be the opposite (for instance // is usually faster on eXist.) Ok, there are not yet an XSLT implementation supported natively by an XML database, but I predict that is going to change soon... What I want to say is just: when it comes to efficiency, there is no rule, just measurements. Regards, -- Florent Georges http://www.fgeorges.org/
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
Re: [xsl] measuring performance of , G. Ken Holman | Thread | RE: [xsl] measuring performance of , Michael Kay |
Re: [xsl] sort a node across a dyna, G. Ken Holman | Date | Re: [xsl] sort a node across a dyna, J. Argyl Plath |
Month |