Keith Moore wrote:
>
>> IMO, Subject should be unstructured, period. Unfortunately, RFC 1036
>> introduced two
>> hacks:
>
>
> Usenet is not email, and email should not be expected to inherit every
> feature (or mistake) of Usenet.
Transfer details are irrelevant to the message format.
But protocols and applications are not.
The format is essentially the same,
Similarity does not make them equivalent. Nor does it mean, as Keith
points out, that one must cover all aspects of the other.
and
in many contexts (e.g. via IMAP) a message is just a message -- there is
no way to differentiate
"Usenet" and "email".
Which at most makes it convenient for the grammers to be equivalent. It does
not rise to the level of requiring it.
"Re: " is a mistake of Usenet that RFC 2822 has
picked up and run with.
Only to the extent of describing a prose convention. It does not appear in the
ABNF, nor should it IMO.
("cmsg" can for the moment be ignored here, since it is only significant
for Usenet transport
software) From the point of view of library code handling the message
format, the fact that
1036 requires "Re: " (and cmsg) to be recognized effectively makes that
part of the syntax;
library code has no way of knowing whether the calling application is
Usenet-specific software,
an MTA, an email-only MUA, a combined email/news UA, or an IMAP client
(for which there
is no Usenet/email distinction).
You're making all sorts of assumptions here that just do not hold water: That
it is necessary to handle "Re:" at the same syntactic level, that a library
even needs to concern itself with this detail, that all implementations must
allow for the conflation of netnews material and email material.
The bottom line as far as I'm concerned is that I see no requirement
that "Re:" be handled in the 2822 ABNF. And since it is not a requirement,
it becomes a matter of costs versus benefits. And from what I've seen
so far the costs far exceed the benefits.
Ned