RFC 934 for multipart encoding

[Nathaniel Borenstein <nsb@xxxxxxxxxxxxxxxxxxxx>]

Well, my recreation for the weekend was to scrutinize RFC 934.  I've
concluded that it does 99% of what we need to extend RFC 1049
(Content-type) to include multiple types.  Here's what I'd propose to go
the rest of the way:

1.  We define & register a new content-type, "multipart", to distinguish
a multipart message where the parts are distinguished using RFC 934
encapsulation.

2.  Within such messages, we use 934-style encapsulation boundaries for
each separate content-type, e.g.

----Encapsulated Bodypart
Content-type: foo

or, more simply, just

----Content-type: foo

3.  We relax (remove) the restriction in RFC 934 that each encapsulated
piece must have From and Date headers, which are really silly and
redundant in this case.

4.  We define a standard quotation method for representing 8 bit data
within encapsulated parts.  I can think of many ways to do this but
don't want to get into an argument over which one is right at this stage.

5.  We define a way to distinguish "real" from "soft" line breaks.  This
is important for long binary data streams that don't really include line
breaks, but have to break lines to accomodate existing MTA's.  (The
Andrew solution, in this case,  was simply to say that the "real" number
of line breaks was one less than the apparent number of line breaks --
i.e. one or more real line breaks were quoted with an extra line break. 
A zillion other approaches are also feasable.)

All in all, it now seems to me that we could get everything we need from
multipart multimedia mail by using RFC 1049, RFC 934, and a few small
extensions like the one above, which should probably be written up as a
fresh RFC once we have something like a consensus.  Also, this scheme
should work perfectly with the "bottom-up" approach to multimedia mail
(metamail) that I've described in the past -- only minimal modifications
to my existing scheme will be required.)

I feel more strongly than ever that neither RFC 1154 nor new extensions
to RFC 821 are needed for multipart multimedia mail, and I thank Timo
Lehtinen for drawing our attention towards RFC 934.  -- Nathaniel