Using extension points without registries, was: [Ietf-message-headers] Last Call Summary on draft-yevstifeyev-http-headers-not-recognized

[Julian Reschke <[e-mail]>]

On 09.01.2011 12:03, Ben Niven-Jenkins wrote:
> Agreed, we don't need to create a registry until we have something to put in it.
>
> If someone writes an I-D that creates additional Warning codes, that same I-D can:
> 1) Note that a registry needs to be created, or
> 2) Suggest creating the registry and contain the relevant IANA guidelines etc.
>
> In the case of (1) what I would expect is *if* the I-D progresses then at some point in the future before it is sent to IESG, either
> a) a separate I-D is produced&  progressed in parallel to create the registry and IANA guidelines
> b) the IANA guidelines etc are put in the original I-D (essentially the same as (2) above but the original I-D author does not expend time writing IANA guidelines before receiving feedback on their actual proposal).
>
> Ben

As a matter of fact, HTTPbis currently defines (at least) three new 
registries (method names, cache control codes, auth schemes). The 
absence of these registries hasn't caused these extension codes to be 
used before.

Summarizing: I'm reluctant to add new registries unless we *know* we 
need them. In the past 11 years, apparently nobody has asked for a new 
Warning code before. That might tell us something.

Best regards, Julian