Russ Allbery <rra(_at_)Stanford(_dot_)EDU> writes:
But by the time the protocol extension reaches that point, it should
*already* be an I-D and the IETF consensus process can begin.
Exceptions of headers that reach widespread use without ever going
through IETF standardization can be dealt with separately. Encouraging
people to take that route rather than properly documenting their
extension in an IETF document strikes me as counter-productive.
I see no purpose served by the provisional registry.
That wasn't sufficiently clear, or rather the above is from a particular
perspective that's not clearly noted.
I see no purpose served by a provisional registry under the aegis of the
IETF. If you want a registry of every header that someone's claimed,
Dan's already running a perfectly usable one. If you have no intention of
ever writing an IETF standards document for your header, I don't see why
you'd use an IETF header registry either.
In other words, an IETF registry is going to serve a particular body of
users, and for that use the above-quoted arguments apply. If you don't
actually want to cooperate with the IETF, then the existence of an IETF
registry is going to be irrelevant to you, and you may as well just use
Dan's list.
It seems to me that one of the goals for a provisional header registry is
to actually provide some sort of "fast-track" process to register a header
with the IETF without actually going to the work of writing a formal
description of its semantics and subjecting it to group review. I can
appreciate that sentiment; however, I don't think that's what the IETF
does. If you want something quick and dirty, you can use the existing
lists of headers that don't guarantee anything beyond "someone's using
this." If you want the process of writing an IETF standards document to
be easier, well, adding an end-run around the process isn't solving that
problem in the most direct or straightforward manner.
--
Russ Allbery (rra(_at_)stanford(_dot_)edu)
<http://www.eyrie.org/~eagle/>