It's an argument, but I think the overall effect will be
negative and harmful to XSLT usage.
In particlular people (now, who could I be thinking of
here:-) will write and distribute stylesheets purporting to
be XSLT1 that work when they test them but that will fail to
compile and so fail to produce any result on any conformant
XSLT 1 processor. That way lies confusion....
I think the only confusion would come from seeing version="1.0" at the
top of the stylesheet and then assuming that it contained only 1.0 code.
As we know, a stylesheet can quite happily specify version="1.0" and
contain 2.0 code. So, by specifying version="1.0" I'm not in any way
guranteeing that my stylesheet will run on any conforming processor.
(This seems silly doesn't it?)
But then, as node-set() was an extension function, it would be rare that
a 1.0 stylesheet would work with *any* conforming processor, more like a
subset of conforming processors (exsl, xalan, msxml).
I think most of us have control of our target environments, so
portability comes last compared to functionality, performance and
maintainability. Having to convert RTF's to a node-set is a pain, the
difference between the two is only theoretical, and the sooner they're
forgotten about the better :)
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list