--On Tuesday, February 03, 2009 18:42 +0200 Jari Arkko
<jari.arkko@xxxxxxxxx> wrote:
> Brian,
>
>> Almost, because I'm not sure that the "MUST NOT publish"
>> should apply to Experimental. I think SHOULD NOT is strong
>> enough for Experimental; we already have a set of guidelines
>> for Experimental publication.
>>
>
> I think I agree with this.
Setting my other misgivings aside, +1.
However, I just tried to think about the cases in which we would
want to publish something as experimental that we had concluded
were OBE.
(1) In the late 80s and early 90s, a large fraction of the
community (probably a majority) had concluded that TCP/IP was
OBE. We had an entire IETF Area devoted to OSI Transition.
Fortunately, we did not shut down all of the WGs that depended
on TCP or IP as this document would suggest if it has been
applied at that time.
(2) Using the criteria in the document, we would have shut down
(or never started) any work on XMPP because it appeared at the
time that the marketplace had decided on a set of incompatible
and proprietary protocols.
(3) Given those two examples, and probably others, I think there
may be a place for publishing specifications that appear to be
dead ends as Experimental or even, with appropriate disclaimers,
as Proposed Standards. The conclusion that the technology is
OBE could be wrong and the community and IESG should be able to
make judgments about the risks of being wrong, the advantages of
having alternate technologies well-specified in case the
marketplace choice turns out to be unworkable in practice.
So, again, I'm not quite sure what problem is being solved here
and I think an argument should be made for "more good sense and
judgment and fewer rules that then turn out to over-constrain
us".
john
_______________________________________________
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf