Not being an SGML expert, I can't really comment on this one, though
I've heard just the opposite opinion from others who do know SGML.
However, I'm not convinced this should delay the RFC, for two reasons:
1. Although being able to make a richtext DTD would be nice and was a
minor goal, I don't regard it as critical, by any means. I would lose
no sleep over the loss of this capability.
2. If some small changes to the richtext definition would ease the
definition of a DTD, we could easily include such changes in the final
Internet Standard version -- it certainly isn't critical if a few
details of richtext change between proposed standard and Internet
Standard.