I'm not ignoring it. I'm categorically stating that:
Indeed you are.
And other people are categorically stating the exact opposite :-( .
Well, I'm not accusing them of ignoring me :)
No, correct syntax alone isn't sufficient. The proposal must at least
appear to be sound and not harmful, and failure to address
security/privacy considerations is a frequent source of harm.
(cookies in HTTP, various frobs to request an automatic response in email)
Yes, but a registry is not the place to specify syntax or other technical
details. That is what the indicated specification or the ongoing
discussion is for.
It seems to me that we're discussing a mechanism for defining extensions
for certain protocols, for reserving the names used for those extensions,
and for making information about those names and extensions generally
available. But because extensions can be disruptive or otherwise
harmful, in order to minimize the potential for harm it needs to have
some degree of public review and oversight. And a certain degree of
specification is needed for such review to be effective.
The fact that we've been using the word "registry" to refer to this
extension mechanism doesn't change that; and the fact that we've been
using that word shouldn't compel us to choose an extension mechanism
that doesn't provide sufficient information with which to review
a proposal.
Agreed that, at the time of registration, some degree
of soundness and lack of harmfulness is needed. But that is just a matter
of insisting that the initial specification is clear enough to permit
experimental use in the wild, and that the forum proposed for its
discussion is the most appropriate one (e.g. that there is not already
another forum tackling the exact same problem).
I strongly disagree. A clear specification does not prevent harm, and
potentially having a separate forum for each new header field will
just diminish the effectiveness of community review. At most there
should be one forum for each protocol (e.g. mail, news, and http),
and the proposal should be reviewed by the forum corresponding to the
protocol for which the extension is proposed. But since header fields
tend to leak between protocols, I'd far prefer a single forum for
discussion of all new proposals.
Keith