> With a d-o-e solution I don't have to care, character maps may have be 
> adapted if the content changes.

You only have to worry about < and & (as they are the only characters
that need escaping) so you only have to find two unused characters out
of the 1114111 available unicode characters. Unicode has whole blocks of
characters that are explictly and permanently undefined for use
privately within applications such as this. the ones I suggested in that
mail are the first three of those so called private use characters.

There is the faint possibility that the content you are transforming has
it's own "private" use but you could check for this with contains() or
just use some slightly more obscure pair of PUA characters than the
first two (there are  thousands of the things to pick from)

With a d-o-e solution you always have to worry that your application
might be moved to a system that doesn't support it as even in xslt1 it
was always optional and not always supported.

David

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________