RE: how to prune subtrees in an XQuery result

I'm hoping to do some XSLT vs XQuery performance comparisons over a wider
range of queries over the next few weeks. I don't think you can draw too
many conclusions from one example. However, I think that queries of the kind
"copy the whole tree except for X" are more natural to express in XSLT than
in XQuery and it wouldn't be surprising if such queries also performed
better in XSLT. There are other queries (the more SQL-like queries) that are
more natural to express in XQuery.

Obviously you can't extrapolate any performance results from Saxon and
assume that they are characteristic of the two languages. That requires a
wider set of comparisons.

I suspect most people will choose between the two languages based on
perceived usability rather than performance. Perceived usability depends
largely (in my view) on what other languages you have used before.

Michael Kay 

> -----Original Message-----
> From: www-ql-request@w3.org [mailto:www-ql-request@w3.org] On 
> Behalf Of Mukul Gandhi
> Sent: 21 March 2005 08:06
> To: Michael Kay; www-ql@w3.org
> Subject: RE: how to prune subtrees in an XQuery result
> 
> 
> Hi Mike,
>   I did a performance comparison between XQuery
> solution and the XSLT (1.0) solution (using Saxon 8.3;
> with -t option).
> 
> The equivalent XSLT solution is(I guess it can be
> written more efficiently):
> 
> <?xml version="1.0"?> 
> <xsl:stylesheet
> xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
> version="1.0">
>  
>  <xsl:output method="xml" indent="yes" /> 
>  
>  <xsl:template match="/ASet">
>    <Results>
>      <xsl:for-each select="A">
>        <xsl:if test="B[(. = 'red') or (. = 'yellow')]
> or C[(. = 'chair') or (. = 'table')]">
>          <A>
>            <xsl:copy-of select="@*" />
>            <xsl:copy-of select="B[(. = 'red') or (. =
> 'yellow')] | C[(. = 'chair') or (. = 'table')]" />
>          </A>
>        </xsl:if>
>      </xsl:for-each>
>    </Results>
>  </xsl:template>
>  
> </xsl:stylesheet>
> 
> XQuery solution's (average) execution time is - 313
> milliseconds
> XSLT solution's (average) execution time is - 63
> milliseconds
> 
> XSLT seems quite faster than XQuery. Can we conclude
> that XSLT will always be faster than XQuery? 
> 
> I believe XQuery is geared more towards database
> rather than flat XML files.. Should we still prefer
> XQuery sometimes than XSLT (when doing a normal flat
> file XML transformation). 
> 
> Regards,
> Mukul
> 
> --- Michael Kay <mhk@mhk.me.uk> wrote:
> > Here's a solution that doesn't involve doing the
> > comparisons twice.
> > 
> > <Results> {
> >    for $A in doc("data.xml")/ASet/A
> >    let $matches := $A/(B[.=("red","yellow")]
> >                      |C[.=("chair","table")])
> >    where ($matches)
> >    return <A>{$matches}</A>
> > }</Results>
> > 
> > Michael Kay
> > http://www.saxonica.com/
> > 
> > > 
> > > Hi Sergio,
> > >   Probably this XQuery is required..
> > > 
> > > <Results>
> > > {
> > >       for $A in doc("data.xml")/ASet/A
> > >       where ($A/B = "red" or $A/B = "yellow")
> > >             and ($A/C = "chair" or $A/C = "table")
> >    
> > >  
> > >       return 
> > >       <A>        
> > >         {$A/@*}
> > >         {$A/B[(. = 'red') or (. = 'yellow')]}
> > >         {$A/C[(. = 'chair') or (. = 'table')]}
> > >       </A>
> > > }
> > > </Results>
> > > 
> > > Regards,
> > > Mukul
> > > 
> > > --- Sergio Andreozzi
> > <sergio.andreozzi@cnaf.infn.it>
> > > wrote:
> > > > 
> > > > Dear all,
> > > > 
> > > > I'm wondering what is the best way to prune
> > subtrees
> > > > that did not match 
> > > > the where clause of a FLOWER expression.  See
> > the
> > > > example below for 
> > > > clarification:
> > > > 
> > > > SAMPLE DATA
> > > > 
> > > > <ASet>
> > > > 	<A name="one">
> > > > 		<B>red</B>
> > > > 		<B>yellow</B>
> > > > 		<B>green</B>
> > > > 		<C>chair</C>
> > > > 		<C>table</C>
> > > > 		<C>sofa</C>
> > > > 	</A>
> > > > 	<A name="two">
> > > > 		<B>red</B>
> > > > 		<B>green</B>
> > > > 		<C>chair</C>
> > > > 		<C>sofa</C>
> > > > 	</A>
> > > > 	<A name="three">
> > > > 		<B>green</B>
> > > > 		<C>chair</C>
> > > > 		<C>table</C>
> > > > 		<C>sofa</C>
> > > > 	</A>
> > > > </ASet>
> > > > 
> > > > QUERY: for each A, list all the name of A, their
> > B
> > > > elements that are 
> > > > either "red" or "yellow" and their C elements
> > that
> > > > are either "chair" or 
> > > > "table"
> > > > 
> > > > EXPECTED RESULT:
> > > > 
> > > > <Results>
> > > > 	<A name="one">
> > > > 		<B>red</B>
> > > > 		<B>yellow</B>
> > > > 		<C>chair</C>
> > > > 		<C>table</C>
> > > > 	</A>
> > > > 	<A name="two">
> > > > 		<B>red</B>
> > > > 		<C>chair</C>
> > > > 	</A>
> > > > </Results>
> > > > 
> > > > 
> > > > in practice what I would like to understand is
> > how
> > > > to prune all the 
> > > > subtrees that do not contain any match with
> > > > components of the WHERE 
> > > > clause in the XQuery result.
> > > > 
> > > > The general pruning rule is "if an element does
> > not
> > > > participate in the 
> > > > satisfaction of the WHERE clause and all its
> > > > children elements (if any) 
> > > > don't participate as well, then prune it from
> > the
> > > > result".
> > > > 
> > > > A possible starting point is the following
> > query,
> > > > but the pruning action 
> > > > is missing.
> > > > 
> > > > <Results>
> > > > {
> > > >       for $A in doc("data.xml")/ASet/A
> > > >       where ($A/B = "red" or $A/B = "yellow")
> > > >             and ($A/C = "chair" or $A/C =
> > "table")
> > > >       return ...
> > > > }
> > > > </Results>
> > > > 
> > > > 
> > > > 
> > > > Thanks,
> > > > 
> > > > 	Sergio
> > > 
> > > 
> > > 
> > > 		
> > > __________________________________ 
> > > Do you Yahoo!? 
> > > Yahoo! Small Business - Try our new resources
> > site!
> > > http://smallbusiness.yahoo.com/resources/ 
> > > 
> > > 
> > 
> > 
> > 
> 
> 
> 		
> __________________________________ 
> Do you Yahoo!? 
> Make Yahoo! your home page 
> http://www.yahoo.com/r/hs
> 
>

Received on Monday, 21 March 2005 09:04:03 UTC