Anthony Jones wrote:


> Have you ever tested the depth to which this sort recursion can be taken, is
> there a limit?

No, that will certainly depend on the implementation, in particular if 
it has optimations to avoid the recursion on the implementation level.

> Isn't the need to write such a monster for what elsewhere is a fairly basic
> (even intrinsic) function indicative the perhaps XSL isn't a good choice
> here?  XSL may form part of a useful solution but this sort of work should
> be done somewhere else surely?

I admit that using named template and XSLT to write recursive templates 
is clumsy and the code looks bloated compared to other solutions but you 
have to judge the advantages (pure XSLT solution) against the 
disadvantages of possible other solutions (e.g. processor dependent 
script extension).
So it depends on the requirements, if you know you are using a certain 
XSLT processor having a minimum extension function or access to script 
with such a function then you can certainly go for that solution. But I 
would not rule out some XSLT code only because some xsl:with-param or 
xsl:choose/xsl:when is a bit bloated.


-- 

	Martin Honnen --- MVP XML
	http://JavaScript.FAQTs.com/