Pete Resnick wrote:
On 1/15/04 at 7:09 PM +0100, Arnt Gulbrandsen wrote:
6. I don't like Bruce's changes to subject. They mix in RFC 1036
syntax, which IMO does not belong in 2822.
I agree. These should be removed.
IMO, Subject should be unstructured, period. Unfortunately, RFC 1036
introduced two
hacks:
1. "Re: " (which it got wrong!), which *requires* certain actions when
encountered, i.e. it is
effectively part of the syntax. See RFC 1036 sections 2.1.4, 2.2.5.
2. "cmsg", which *requires* certain actions when encountered. See RFC
1035 sections 2.2.6
and 3 (including subsections).
This is the same "Subject" field described in RFCs 822 and 2822.
Now, if an RFC 2822 successor were to repudiate those requirements
spelled out in RFC 1036,
that would be fine.
As things stand now, it is necessary to recognize the "re: " and "cmsg"
hacks in order to comply
with RFC 1036, and there is at least one draft nearing RFC status that
would add "Auto: " to the
list of Subject field hacks.
RFC 2822 currently has verbiage (section 3.6.5) which falls short of a
requirement, however it
does nothing to remove the RFC 1036 requirements. The verbiage also:
1. implies that presence of "Re: " indicates that a message is a reply,
which is incorrect (I daily
receive several spam messages having Subject fields beginning with
"Re: " which are not replies).
2. seems wholly unnecessary. It might as well state that when not used
in a reply, Subject MAY
begin with "Re: ", when used in a reply, Subject MAY begin with
"Qwerty%$@&^%#:_",
when not used in a reply, Subject MAY begin with "Qwerty%$@&^%#:_", etc.
If an RFC 2822 successor were to repudiate the RFC 1036 Subject hack
requirements, then
subject = "Subject:" unstructured CRLF
would be accurate (provided no other hacks are introduced in the interim).