Thanks - I'm still missing a second generating agent, but now that you've
claimed generation for both PMDF command-line and the PMDF ALL-IN-1 gateway,
I believe we have one (from the same codebase).
The reason for the fuss is the change from 1602 to 1602bis, which we are
currently operating under:
I understand what text we're operating under. What I do not understand, do not
see, and do not agree with is how this text is being interpreted as being a
requirement that certain specific sorts of generation agents exist. As I said
previously, I do not believe that many if not most features of IETF
specifications can possibly meet such a "generating agent" requirement. And
even if this is a valid reading of the new text, I do not like this being
brought up at this point in the process for a document that is undergoing a
recycle. It should have been brought up quite a while back.
John gave some examples at the extremes of the spectrum for how
this can be handled. His examples are intended to be ridiculous extremes, but
they do illustrate the range of possibilities here.
What John did not mention, however, is that the MIME specification itself
provides detailed information on what MIME conformance and interoperability
really mean. In the case of multipart/alternative, for example, a display agent
is said to be MIME conformant if it is capable of displaying only the first
part. Agents are encouraged to do better, and many of them do in fact do
better, but this is not required for conformance. I have already noted that a
conformant MIME display agent doesn't have to do anything with parallel other
than treat it as mixed, and this is why we don't have a lot of people jumping
up to say that they "support" parallel. If we think we need to have more
support for parallel then we need to change the requirements, not remove the
concept from the specification.
The specification is totally silent on the subject of generation agents. This
is because the specification intentionally imposes no existance or completeness
requirements on such things. As such, I believe the very concept of requiring
such things is out of order given the content of the present documents.
I also don't think you quite realize what a can of worms you're opening here.
If the existance of generation agents is going to be a requirement for MIME
interoperability then this needs to be explicitly stated in the documents
themselves, along with what agents need to be able to generate in various
circumstances. I can even see some logic in this -- there was at least one
agent out there that could display MIME just fine but was incapable of
generating it, yet could still claim to be MIME compliant.
But if such a requirement is in fact imposed I cannot see how we can advance
the documents without describing it in detail in the section on conformance.
And since this is an entirely new concept for MIME, which in fact we explicitly
agreed to leave out of the original specification and instead put in a separate
informational document (RFC1344), it will absolutely require a reset to
proposed. And given that non-trivial requirements for MIME generation agents
are likely to be a very dicey topic indeed it is possible that this will
require the formation of a new working group to debate them.
I therefore suggest the following alternative set of steps for us to take:
(a) Recycle the documents as-is. All this does is fix the many things in the
MIME specification that urgently need to be fixed. It doesn't change the
current situation vis-a-vis generation agents or support for various
complex combinations and permutations of MIME objects.
(b) Write up a specification for what generation agents have to be able to do
in order to be MIME compliant. I will undertake to do this if need be --
I think I understand what can be required and what cannot be. I doubt that
you'll going to get much in the way of required support for constructing
complex MIME objects into the requirements (you definitely will not if I
have any say in the matter), but you will end up with a set of things you
can test at both ends for interoperability.
(c) See if (b) flies. If it does put it on the standards track, get it to
draft status and move it into the MIME specification at that time. (I
REALLY want to avoid a reset to proposed for the specification as a whole.)
If it doesn't we need to look at why it doesn't and see whether or not we
think generation agent requirements are really such a good idea.
Ned