At 02:44 PM 2/3/2002 -0500, Keith Moore wrote:
A registry introduces a third possibility - documenting practice that
isn't either existing or desirable.
if someone wants to document it and share it, then it is desireable to
somebody.
the fact that somebody desires that a protocol extension be documented and
shared does not imply that it benefits the internet community for IETF
to help them.
it helps to stay with specific, stable criteria.
you previously said "desireable".
now you want to move to using the same criteria that are used for IETF
standards.
No, I've said on multiple occasions that it's reasonable and useful to
register extensions that have consensus but don't meet the 2026 requirements
for standards track. I believe I've also said that it's generally useful
to document existing practice. It's the "register any old random
extension" idea that I have a problem with.
it seems clear that you do not want to discuss a registry. rather, you
want to impose a standards process on all header name publications.
It seems clear that you want to persuade others by misrepresenting my
views, and resorting to ridicule, rather than by explaining why your
approach will improve reliability of Internet applications that use
header fields. Frankly, I expect better from you.
should IETF help "document and share" Fleming's IPv16 proposal?
yes.
Perhaps we simply disagree. I don't think that documenting stupid
ideas that have no community support is a useful exercise of either
IETF's or the RFC Editor's time, energy, or network bandwidth.
Doing so just serves to degrade the value of the RFC document series
and confuses the marketplace.
you seem to be forgetting the difference between the IETF openness to
sharing ideas with the IETF approval process for a subset of those ideas.
you also seem to be forgetting that the model in the msg header registry
proposal is already in use by IANA, so we are not inventing some high-risk
thing here.
You seem to forget that different protocols have different extension
mechanisms, and different registries have different criteria - for
good reasons. I'm arguing about the criteria for *this* registry.
Keith