Keith Moore wrote:
Actually, no. There was a difference in intent between the two. 1123 was
imposing new requirements on existing protocols, 1341 and 1342 were not
doing so. MIME implementation was not mandatory (as a matter of standards
compliance) for existing user agents.
Yes, in theory 1123 is mandatory and MIME compliance is not. In theory,
SMTP MTAs had to be compliant with RFC 821 as reaffirmed by 1123. In
practice most are not. The Internet SS isn't rounding up MTA authors
and throwing them in concentration camps. Certainly a UA can be
constructed which is not MIME-compliant (and indeed there are many of
those). However, while there is no standards requirement for MIME
support (and no enforcement even if there were a requirement), those
UAs which do provide MIME support are used in vastly greater numbers
than those that do not. Market forces are infinitely greater than
standards enforcement as far as the Internet is concerned.
The world changed a lot between the time that 1123 was written and the time
that MIME was written. At the time 1123 was written the Internet was small
enough, and there was still enough vestage of the ARPAnet mentality, that
imposing a few new requirements on existing protocols was conceivable. By the
time MIME was written it was obvious to us that trying to insist on
everybody's MUAs or MTAs being upgraded - no matter how trivially - as a
condition for making MIME work was simply untenable, and there was
considerable resistance from some parties to expecting English users to
upgrade their mail tools for the sake of the "few" (at that time) in the net
that didn't speak English. For that matter, most of the changes in 1123 were
not likely to cause the degree of pain that MIME did.
Anyway, I hope the difference in intent is clear by now.