Re: [xsl] How smart are the XSLT processors? Are there any XSLT processors that convert tree-recursive functions into efficient iterative procedures?

Subject: Re: [xsl] How smart are the XSLT processors? Are there any XSLT processors that convert tree-recursive functions into efficient iterative procedures?
From: Markus Flatscher <markus.flatscher@xxxxxxxxxxxx>
Date: Wed, 28 Apr 2010 22:17:12 -0400
Dear David,

Can you quickly point me to somewhere (just a list thread or something) where your pet peeve about using xsl:variable with content is explained? At my current learning stage, that would be very helpful to me, especially since I do what I'm not supposed to do a lot, and I'd like to learn how to do it better. Also, other than reading (and understanding) the specs, is there any recommended reading you have for intermediate users of XPath/XQuery/XSLT2? Or should we just keep re-reading Mr. Michael Kay?

Thanks!

Markus


David Carlisle wrote:
On 28/04/2010 23:20, Florent Georges wrote:
David Carlisle wrote:

Hi,

You are making it pretty inefficient in any case by using
value-of (or relying on your xslt engine to optimise this away)
by using value-of you are asking for a document node with a
text node child

I didn't follow the whole discussion in details, so I probably missed something, but why do you say "by using value-of you are asking for a document node"? I thought it just created one text node.

Regards,



er because it was the middle of the night and I'm so used to moaning at people for using xsl:variable with content rather than a select attribute that "document node with text node child" just slipped off my fingers, sorry. (If you look at the saxon list you'll see that I hadn't woken up enough this morning to notice my error:-)

but the main point is that it's really too hard to expect automatic conversions from O(n^2) algorithms to O(log n). My own attempt was the following which I'd hoped would beat dimitre's for speed, but it doesn't consistently, and his has the advantage of using exact integer arithmetic (I didn't do the analysis to see at what the point the following will succumb to rounding error)

<xsl:function name="ex:pow" as="xs:decimal">
<xsl:param name="a" as="xs:decimal"/>
<xsl:param name="n" as="xs:integer"/>
<xsl:choose>
<xsl:when test="$n=0">
<xsl:sequence select="1"/>
</xsl:when>
<xsl:when test="$n=1">
<xsl:sequence select="$a"/>
</xsl:when>
<xsl:when test="$n mod 2 = 0">
<xsl:sequence select="for $z in ex:pow($a,$n idiv 2) return $z*$z"/>
</xsl:when>
<xsl:otherwise>
<xsl:sequence select="for $z in ex:pow($a,$n idiv 2) return $z*$z*$a"/>
</xsl:otherwise>
</xsl:choose>
</xsl:function>


<xsl:variable name="phi" select="1.618033988749894848204586834365638117720"/>
<xsl:variable name="r5" select="2.2360679774997896964091736687312762354406"/>


<xsl:function name="ex:fibclosed" as="xs:integer">
<xsl:param name="n" as="xs:integer"/>
<xsl:sequence select="xs:integer(floor(ex:pow($phi,$n) div $r5 + 0.5))"/>
</xsl:function>



-- Markus Flatscher, Project Editor ROTUNDA, The University of Virginia Press PO Box 801079, Charlottesville, VA 22904-4318 USA Courier: 310 Old Ivy Way, Suite 302, Charlottesville VA 22903 Email: markus.flatscher@xxxxxxxxxxxx Web: http://rotunda.upress.virginia.edu/

Current Thread