RE: problem using dyn:evaluate with MSXML

> However if I suggest that, Mike is bound to say that saxon 
> can spot if an
> RTF variable is always used as a string and so optimises away its
> internal node structure. so doing this explictly makes no difference.

I'm getting close to that, but not quite there yet. I do some special-casing
of constructs like 

<xsl:variable name="x">fred</xsl:variable>

or

<xsl:variable name="x"><xsl:value-of select="'fred'"/></xsl:variable>

but it's still a lot more expensive than

<xsl:variable name="x" select="'fred'"/>

Michael Kay
http://www.saxonica.com/