In
<5(_dot_)1(_dot_)0(_dot_)14(_dot_)2(_dot_)20010919115724(_dot_)00a007d0(_at_)joy(_dot_)songbird(_dot_)com>
Graham Klyne <GK(_at_)ninebynine(_dot_)org> writes:
The usual IETF way to deal with these situations is to have an IANA
registry of values. But none exists for message headers (I can find no
reference to RFC822 or email messages in
http://www.iana.org/numbers.htm). The nearest to a registry we have seems
to be Jacob Palme's RFC 2076 and more recent Internet Draft updates
(currently draft-palme-mailext-headers-05.txt).
Yes, I agree that this is a task for IANA, but not tied irrevocably to RFC
2076, as others have said.
This issue has already been discussed on the USEFOR Working Group, which
could see much benefit in such a scheme. Essentially, there are three sets
of protocols which use headers in this style:
Mail (RFC 2822, plus many extensions for MIME and much else)
Netnews (RFC1036, or usefor in due course)
HTPP (RFC 2616)
There are many headers which are used in two, or even all three, of those
protocols with essentially the same meaning (or with small but carefully
documented differences).
Therefore, what we need is an IANA Registry for header names which would
record which protocol(s) used it, and the RFC(s) where its usage was
defined (that might be up to 3 RFCs if all of Mail, Netnews and HTTP were
involved, and even more if a header was defined in one RFC, but had some
extended usage added in another). Here I am speaking of both
standards-track and experimental protocol RFCs (since the prime objective
is to prevent two independent groups trying to define the same header).
But I would go further than that. I would also allow IANA to issue
provisional registrations for proposed new headers. There would have to be
stringent conditions for these, such as the existence of an internet draft
defining the proposed new header (thus provisional registrations would
lapse after 6 months in the same way as internet drafts). I understand
that there are already precedents for IANA to issue provisional
registrations (such as for Port Numbers, I believe).
The advantage of that scheme is that someone proposing a new header with
the intent that it would eventually see widespread use can get it
registered and immediately start using it on an experimental basis.
Currently, all he can do is to use an X-Header; then people start using
the X-Header and its name gets built into various pieces of software; then
someone decides to make it official in an RFC; and then they have to
change it to something different without an X-, and all the working
software has to change again :-( . We are currently going through that
process as regards the commands in the NNTP protocol. That is the
particular aspect of the matter which has been duscussed on USEFOR, and is
why I believe USEFOR would support such a move.
But of course, whoever proposes such a header has to show evidence of his
good intentions by at least producing an internet draft (which should be
enough to deter all but the most determined of the pranksters).
Please can someone enlighten us as to the mechanisms necessary to
establish a new IANA registry? Does one write an RFC
instructing/authorising them to do it, or is it done just by some edict of
IESG? If an RFC is needed, then I suspect this List is a good place to
prepare the draft, but of course we need to bring the News and HTTP people
on board (plus some input from IANA, of course).
--
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133 Web: http://www.cs.man.ac.uk/~chl
Email: chl(_at_)clw(_dot_)cs(_dot_)man(_dot_)ac(_dot_)uk Snail: 5
Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9 Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5