Re: transfer-encodings on subtypes of "message"

> I would like to request that this substantive change be un-done.
I see a lot of validity on either side of this argument, so this is one issue
I'm going to stay out of. I'll implement whatever the group decides is
appropriate, of course, but this was clearly a change from the earlier
documents, so unless someone wants to argue the other side I'll have to
reinsert the restriction.

                                Ned