[Sidrops] Re: I-D Action: draft-ietf-sidrops-rpki-erik-protocol-06.txt

[Dirk Doesburg <[e-mail]>]

Hi all, nice to see so many responses on this draft!

I think it'd be better to not require clients to ignore trailing dot, if also requiring that they MUST not be there. Despite Postel’s law, *requiring* clients to accept non-compliant indexScope makes violations go unnoticed. Only if *one* side of this requirement is a MUST and the other is not, we can count on it actually being adhered to (I don’t mind which side is required versus which is suggested).

I now see that on the PR the client side already became SHOULD. Nice.

Related nit: the use of “.” at the end of the URI example lines in e.g. 5.2 and 6.3.1 may cause further confusion in txt/htmlized rendering of the draft, while being clear in the html representation. Perhaps just drop them?
Example from 5.2:

> As an example, the URI to fetch an ErikIndex for the rpki.ripe.net
> FQDN from a relay at relay.example.net would be:
> https://relay.example.net/.well-known/erik/index/rpki.ripe.net.


- Dirk


> On 14 Aug 2026, at 11:36, Loganaden Velvindron <[e-mail]> wrote:
> 
> 
> 
> On Fri, 14 Aug 2026, 11:58 Lancheng, <[e-mail]> wrote:
> > A trailing root label should be explicitly disallowed in indexScope
> > fields. (If we could start over with the RPKI it would've made sense
> > to mandate trailling dot always in the fields that contain FQDNs, but
> > that's not the universe we live in, so instead we yield to the deployed
> > reality.)
> 
> I agree. I suggest explicitly requiring that an indexScope (i.e., FQDN) MUST NOT contain a trailing root label. When fetching ErikIndex objects, a client MUST ignore a trailing root label as well. Perhaps this clarification could also be added to Section 5.2 and illustrated together with the existing example. The existing example in Section 5.2 is clear.
> 
> Thank you. Im updating the pr.
> 
> 
> 
> Best,
> Lancheng
> 
> 
> > -----Original Messages-----
> > From: "Job Snijders" <job=[e-mail]>
> > Send time:Friday, 14/08/2026 01:22:05
> > To: "Loganaden Velvindron" <[e-mail]>
> > Cc: "Braet, Kamiel" <[e-mail]>, "[e-mail]" <[e-mail]>
> > Subject: [Sidrops] Re: I-D Action: draft-ietf-sidrops-rpki-erik-protocol-06.txt
> > 
> > On Thu, Aug 13, 2026 at 09:12:46PM +0400, Loganaden Velvindron wrote:
> > > "4. Client-side Processing
> > > Clients start by acquiring and processing an ErikIndex, which
> > > represents the relay's current Merkle tree head for a given FQDN. A
> > > client MUST verify the requested FQDN exactly matches the indexScope
> > > value in the ErikIndex, and if not proceed to use a different relay.
> > > 
> > > Then, clients can decide whether or not to fetch ErikPartition objects
> > > listed on the ErikIndex, for instance, by checking whether the object
> > > associated with the hash was already fetched at some point in the
> > > client's past.
> > > 
> > > Before using a ErikPartition, the client MUST verify that all URIs in
> > > the accessLocations in the id-ad-signedObject accessMethod instances
> > > in the ErikPartition are encompassed in the requested indexScope. A
> > > client can then decide whether or not to fetch a given manifest
> > > object, by comparing the manifestNumber and thisUpdate with what's
> > > locally cached and what's offered by the remote relay."
> > > 
> > > Can we improve the section about FQDN and URI with something like this
> > > as a guide for implementers ?
> > > 
> > > A URI is encompassed by an `indexScope` if, after URI parsing and DNS name
> > > normalization, the URI host component is exactly equal to the
> > > `indexScope` value.
> > > Comparisons are case-insensitive. A trailing root label, if present, is ignored.
> > > Implementations MUST NOT use substring or suffix matching.
> > > 
> > > What do you think ?
> > 
> > A trailing root label should be explicitly disallowed in indexScope
> > fields. (If we could start over with the RPKI it would've made sense
> > to mandate trailling dot always in the fields that contain FQDNs, but
> > that's not the universe we live in, so instead we yield to the deployed
> > reality.)
> > 
> > Please note that Lancheng also made some suggestions in that same area
> > https://github.com/job/draft-rpki-erik-protocol/pull/7/changes
> > Feel free to send a pull request on top of the latest in the github repo.
> > 
> > Kind regards,
> > 
> > Job
> > 
> > _______________________________________________
> > Sidrops mailing list -- [e-mail]
> > To unsubscribe send an email to [e-mail]
> _______________________________________________
> Sidrops mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]