Re: [saag] SSH & Ntruprime

[Eliot Lear <[e-mail]>]

Hi Rich,

On 25.03.2024 18:04, Salz, Rich wrote:
>
> A half-dozen emails came in while I was writing this and providing 
> examples for it, so some of this will be repetitive, sorry.
>
>   * Fortunately, we have a natural experiment here, because RFC 8446
>     explicitly allows the registration of TLS code points based on
>     I-Ds, so in five years I guess we can see how that worked.
>
> 8446 was published in August 2018, so we’re closing in on the first 
> five years of registry action.
>
> Allowing registry entries to point to a stable document source, rather 
> than requiring an RFC, was an unequivocably good thing. (I’m one of 
> the three TLS registry designated experts.) Look at the ALPN IDs[1] 
> and notice the several entries that are not RFCs. Assume that most of 
> those people would not go through the RFC process; would the Internet 
> be better off if those registries did not exist? Same question for the 
> Broadband form entry in exporter labels[2].  Same question for the ISO 
> identity-based encryption signature schemes. Our job is to make the 
> Internet work better, and those registry entries help that.
>
Fair enough.

> As for whether an Internet-Draft is acceptable, the IETF community 
> decided some time ago that drafts do not get deleted, so they meet the 
> requirements.
>
I don't think the one follows from the other for the reasons I already 
stated.


> This is also a good thing for several reasons. For example, several of 
> the examples I quoted above would not come to the IETF if they had to 
> write an RFC. Even if they do come, things change – SPDY becoming QUIC 
> for example – so should we not register SPDY?
>
SPDY is a weird case.  If someone wanted to publish it for historical 
purposes, with commentary about how things changed in the h2 process, I 
might at least entertain that as an independent submission for 
historical purposes.  Otherwise, a githash might be sufficient, or even 
a pointer to a paper in which it was presented at some academic 
conference.  These are all stable references.


> Also, not every individual or organization has the wherewithal to have 
> long-lasting links to documentation, and in many cases the IETF would 
> probably prefer those things be kept as an IETF resource.
>
Github is free.  Must be this tall to ride the Internet.

Eliot