[Sidrops] Re: Mahesh Jethanandani's No Objection on draft-ietf-sidrops-rpki-crl-numbers-04: (with COMMENT)

[Job Snijders <[e-mail]>]

Hello,

On Wed, May 21, 2025 at 04:32:37PM +0530, Ketan Talaulikar wrote:
> I would recommend against going with option 1. As an implementer, I've
> always found it helpful when I know what to put into a field that is
> "deprecated" but cannot be "removed". Such guidance helps us get
> consistency across implementations and avoids people wondering what
> that number means, how to set it appropriately, and keep it backwards
> compatible, etc.
> 
> How about the following rephrase instead to remind the context (though
> I was perfectly happy with the text as is) ?
>
> SUGGEST:
> Per Section 5.2.3 of RFC5280, CAs issue new CRLs using a monotonically
> increasing sequence number in the "CRL Number" extension. However, as
> the ordering functionality provided by "CRL Number" is fully subsumed
> by the "manifestNumber", it is RECOMMENDED that the "CRL Number"
> matches the "manifestNumber" of the manifest that will include this
> CRL (see Section 4.2.1 RFC9286)

> The 2nd change in the same section already conveys that CLR Number is
> ignored, so that part is already well specified IMO.

But this suggestion does not resolve the question "what happens when the
'CRL Number' and the 'manifestNumber' are different values?" (as asked
by a handful of people). Your suggestion also introduces a nearly
verbatim repetition of what's already stated in the Introduction
section. :-/

I understand 'patch' RFCs are harder to read, but about 6 sentences
later there is an unambigious statement "RPs MUST ignore the CRL
Number".

I'd say leave text as-is or remove the sentence about the value
recommendation. 

Kind regards,

Job