I have tended to have a strong desire for greater modularity to Internet
specs, which includes breaking them out into separate docs, for
easier maintenance and cleaner citation. However, I take Nathaniel's
suggestion of the potential for RichText's good impact and believe that
he is correct that separating it out will make it easier for implementors
to treat it as lower priority.
Certainly it is a dangerous thing to make a change to a spec at this
late stage, but it should be noted that changes of this relatively minor
kind are, I believe, quite valid when going from Proposed to Draft, since
residence at Proposed IS SUPPOSED TO provide experience with implementation
and some testing/operation. The experience is hardly useful if the
spec isn't allowed to change, some.
The feedback from this list has been unusually strong and consistent. Even
those not wanting to disturb things, now, or claiming mild opinion, all
seem to support choosing Nathaniel's option 2. (Would that we had this
kind of consensus over ANY of MIME's other points!)
Hence, I think that it is quite reasonable to explore making the change
now, rather than later, as long as it does not affect the IESG/IAB
process. (Believe me, there is plenty of precedent for this level of
change, this late in the game. I can't guarantee that it's ok, but I
can tell you that I think it OUGHT to be...)
Dave