[Top] [All Lists]

Re: Restrictions on the Content-Type "message"

1999-01-04 11:49:32
The fact is that a news message is not an RFC 822 message.  It is
very similar to one, but it does not conform to RFC 822, and labelling
it in such a way that implies that it does is going to cause problems.

There is a need for a Content-Type that distinguishes RFC 822 mail
messages from RFC 1036 news messages.

(Yes, I'm talking about interpretation of the "Newsgroups" header,
among other things.)

Then I would suggest adding a paremeter to message/rfc822. Again, the cost of
adding new composite types is very high, and has to be taken into account
when making these choices.

Making the Content-Type that describes a news message not be a subtype
of "message/" is just unspeakably lame.

I agree.

And moreover, there is no reason for doing this. Look at it this way: The
reason for wanting to put it under application is to avoid the rule about
encodings that applies to message subtypes. The reason USEFOR wants to avoid
this is so that they can pass messages to restricted transports without
breaking their "cannot change the inner CTEs" rule. In other words, they want
nested encodings.

Now, we had a strong consensus in the MIME WG that such things were not to be
allowed, at all, ever, in any context. And if that consensus still exists (and
I suspect it  does) then this won't fly no matter what top-level type you put
it under. You cannot get there from here. So there's no advantage to not using
a subtype of message.

But suppose the consensus doesn't exist any longer, and moreoever we're will to
say "screw the installed base" that implemented to this consensus in good
faith. (I don't think this going to happen, but let's suppose.) In such a case
the right thing to do would be to lift the restrictions on message/rfc822. Or
failing that, on the message type. So again there's no advantage to not using a
subtype of message.

Bottom line: Regardless of what restrictions currently exist on message, the
right path  forward is for news messages to be a subtype of message. You solve
nothing by using some other branch of the type tree.

BTW, one argument I suspect to hear is that this is news, not mail, and what
applies to news doesn't apply to mail. But all you have to do is look at the
number of news to email gateways to see how this doesn't wash.