>> And of course for experimentation purposes you could use a different
>> syntax for entity references.
[...]
>But this is all leading down the road towards little private languages
>with mini application-specific parsers layered over the XML parse.
Well, I did say "for experimentation". It could then be standardised
using the usual &-syntax.
Even though I'm not convinced of the need for a new mechanism, I would
be happy to go along with one if it was clear that it was sufficient
to satisfy 90% of those who say they need one. My fear is that there
is no set of requirements that is coherent enough to be satisfied. A
proof-of-concept is essential for overcoming that, and the fact that
undefined entities are often a well-formedness error should not be
allowed to stand in the way of it.
-- Richard
