Nathaniel:
I have to agree with this. One of the simplifications I would like to
see in the next version of richtext is the elmination of all character
set switching inside richtext. [...]
Terry:
This doesn't achieve the same effect, since it would enforce character
set switching at the object level rather than on the character level. In
general MIME has this problem [...]
I think if one is transmitting documents of this order of complexity,
one should be using SGML or some other structured text mechanism suited
for the purpose.
Fron my point of view I think I'd like to see MIME have an explicit type for
SGML, although the way the RFC is written implies to me that this was
discussed and rejected. The mechanism would presumably use the multi-part
capability to transmit DTD, SGML declaration and dcument instance separately,
with the instance first, and possibly a `flat ASCII' representation before
that.
It'd be nice to see one or more public DTDs designed for such use, so they
wouldn't need to be sent; I'd be interested in working on this if it helps.
Of course, this has veered a little from wide characters. The prevalant
practice in the SGML world is to use `entities' like &a-acute; for
transmission, but that doesn't help the implementation, and there is no
standard way of specifying the semantics of entities defined in a given
document instance outside the public entity sets provided by the ISO.
Sorry, that's more SGML than ought to creep in here, I suppose.
My point is just that one shouldn't need to require that a basic message
transmission mechanism cope with services priovided by a more complicated
transmission system, where both are available.
Lee
--
lee(_at_)sq(_dot_)com (Liam Quin) the barefoot programmer; SoftQuad Inc +1 416
239 4801
OPEN LOOK UI FAQ; Metafont list; HexSweeper NeWS game; lq-text text retrieval