IETF-55

Message header field registry proposal

This is my personal proposal.  It has got through last-call fairly unscathed.  Mainly, we need to clarify the relationship with MIME, which is a special case.

Discussed with Ned Freed.  Agreed outline of how to handle MIME header fields, taking account of Jeff Mogul's comments about HTTP not really being a MIME-carrying protocol.  The plan is to treat MIME as a separate protocol, which email protocols like SMTP/RFC2822 are specified to use.  HTTP does not use MIME, just something that looks like MIME, so the headers HTTP uses will be registered separately.  At last, it will be clear that Content-length is not a MIME header!