On 11/14/02 at 12:50 PM -0500, Keith Moore wrote:
uh, no. RFC 2822 is concerned with the user agent protocol, and it
does place a few requirements on user agents. RFC 2821 is
concerned with the mail transport protocol, so it places
requirements on MTAs.
Well, that's true, but the rest of Charles's message was exactly
right: RFC 2822 does not require (or even recommend) that user agents
unfold folded headers for any reason. It simply says that for
syntactic or semantic evaluation, they should be treated as if they
were unfolded (e.g., if you compare to Subjects for identity, compare
the unfolded form).
I don't think either 822 or 2822 is clear about the semantics of
folding of header fields.
If you've got problems with 2.2.3, I'm open to suggestions. But I
think 2822 is *very* clear on the semantics of folding. What it
doesn't say is what (if anything) a UA should do as far as unfolding
goes for presentation to the user.
but longstanding practice is that both UAs and MTAs can change the
folding from that supplied by the sender.
"Can change"? Of course. But that wasn't the question that was being
asked here. The question is "Should folding be changed?" It certainly
shouldn't by MTAs (the "don't touch the contents of the message"
maxim). It should be changed by UAs if the user has requested a very
long header field that needs to be folded. Whether it should be
changed to unfold a header field that is received is a trickier
business. My feeling is yes, but it's at least an interesting
question.
pr
--
Pete Resnick <mailto:presnick(_at_)qualcomm(_dot_)com>
QUALCOMM Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102