ietf-822
[Top] [All Lists]

Re: Default Dispositions

1999-10-22 06:27:22
--On Friday, 22 October, 1999, 11:34 +0100 Paul Overell
<paulo(_at_)turnpike(_dot_)com> wrote:

Failing to display text/plain to the user just because it is
not the first part seems perverse and unnecessarily
restrictive.

It prevents having a series of "real attachments" interspersed
with explanatory text.

But, Paul, you can't tell a "real attachment" from some other
sort of body part in any fashion other than heuristically.
"Content-type: text/plain" is a plausible heuristic, but there
are many cases where that heuristic fails.  For example, if I
wanted to attach (note that word creeping in) all of RFC 2046 to
this message, I would hope that it would be treated as an
attachment and displayed on request at the recipient end, not
just folded into the text.

Also, if a message arrives as 
     text
     image
     image
     text
     text
it is at least as likely, a priori, that the sender expected it
to be displayed in exactly that order, or even with the images
next to the first text part, as that 
   text
   text
   text
  (attachments)
   image
   image
or even
   text text text
  (attachments)
   image
   image
was expected.  Sometimes you'll be right, sometimes wrong --
just can't tell.

So when displaying a part we respect the sender's wishes, if
they don't put in a final line break then we don't either.

Seems sensible.  How do you respect whatever wishes the sender
was trying to express when she sent
     text/plain
     text/plain
rather than a single body part.  If something was implied by
that distinction (see example about attaching a large RFC,
above), then it seems you are ignoring that "wish" while
respecting the one that might have just been an error in
communication between the sending user and the composing agent/
sending MUA.

I'm not really suggesting that what you are doing is wrong, only
that this problem isn't easy, that users who really care should
probably be using Content-Disposition (and sending MUAs that
support it in a careful and intelligent way on a per-message
basis-- a set that I think is nearly null), and that any
implication of "X got it right and everyone else is wrong"
(which I note you did not say)  is probably a little excessive
in this context.

regards,
    john


<Prev in Thread] Current Thread [Next in Thread>