Re: [Sidrops] WGLC - draft-ietf-sidrops-rfc6482bis - ENDS: July 17, 2023 (07/17/2023)

[Job Snijders <[e-mail]>]

Dear Theo, WG,

On Wed, Jun 28, 2023 at 09:46:06AM +0200, Theo Buehler wrote:
> > > This should have had a bunch of chat/review/disagreement/agreement
> > > over the last ~6-9 months, so it should be relatively quick to
> > > read/respond about :) Please have replies finished prior to 17
> > > July 2023, so that if there are issues we can chat about them in
> > > SFO at the face-to-face-meeting.
> 
> This draft looks good to me.

Thank you for taking the time to review this internet-draft.

> I sent a bunch of small editorial suggestions in the form of a pull
> request:
> 
> https://github.com/job/draft-rfc6482bis/pull/2

Thank you for the contribution, I've accepted the changes, they'll be
included when I upload -04 of the draft.

> One thing I did not touch: it is a bit confusing that "Element" and
> "Field" are used interchangeably throughout the text. As an aside,
> I would probably prefer "Component" over "Element".

"Component" might relate to the ASN.1 syntax "WITH COMPONENTS { ...",
however, 'component' as a word in the natural language of RPKI-related
RFCs doesn't seem to be commonly used.

Being consistent (either Element or Field) might be a good thing.
Skimming through ASN.1 documents it seems 'element' is more commonly
used, I've adjusted accordingly:

https://github.com/job/draft-rfc6482bis/commit/83829dbc357ad918e9d39e76625011fc8621e709

The internet-draft now aligns better with (for example) RFC 3779 § 2.2.3.8.
Thank you for the suggestion.

> One regrettable point that probably can't easily be fixed (at least not
> in this draft) is that there is no order imposed on ROAIPAddress to
> match what is in RFC 3779 sections 2.2.3.3 and 2.2.3.6.
> 
> On the specification side this would be easy: use the same order as in
> these RFC 3779 sections, using the maxLenth as a tie breaker. The
> problem is that there are about 13k ROAs (11k by RIPE, 1.6k by ARIN and
> ~400 by LACNIC) that have more or less random orders in their
> ROAIPAddress lists - presumably the order in which the prefixes were
> added, though it seems that IPv4 always precedes IPv6.

Indeed regrettable, it seems that ship sailed 10+ years ago.

Kind regards,

Job