1) Outlook working with Exchange Server 5.0 (which converts the MIME
type to proprietary gunk in the MTA), correctly displayed text/paragraph
as text/plain. Thus, Outlook's error with direct Internet access (for
which it decodes MIME itself) appears to be an inconsistency rather than
a direct design decision by Microsoft to go against the standard.
Interesting -- that's the exact opposite of what I would have predicted. Just
goes to show the essential foolishness of trying to predict the behavior of
other people's code...
2) Let's not lose sight of the natural split inherent in text/* vs.
application/*. The text/directory effort, for instance, was debating
last year the crux of the issue, which is whether uniterpreted
information should be displayed to the user or made available to save
(e.g., octet-stream). We can argue over any specific subtype, like
text/html, but the dichotomy clearly makes sense and is an integral
feature of MIME. Conformance to this dichotomy is essential to
continued growth of MIME , as shown in this situation.
Hear hear!
3) I would like to heartily second Paul's idea for a MIME conformance
program. I think the MIME conformance statement in RFC 2049 is an
important direction in Internet standards. What is now needed is for an
independent and trusted third-party to start conformance testing.
Complete agreement here as well.
P.S. I am probably missing someting obvious, but I don't see "interpret
unrecognized subtypes of text/ as text/plain" in the 10 conformance
requirements of RFC 2049. This would seem worthwhile to add as MIME
moves to Standard. When is that, BTW?
Well, for one thing, it's implicit in the first conformance requirement for
text when the charset is US-ASCII:
-- Recognize and display "text" mail with the charset "US-ASCII."
This doesn't say anything about the subtype being "plain".
But then there's an explicit conformance requirement covering this case:
-- For unrecognized subtypes in a known charset, show or offer to show
the user the "raw" version of the data after conversion of the content from
canonical form to local form.
Note that you are allowed to make display conditional on user acceptance.
Ned