Russ went through the Chairs slides.
In particular, Russ reported that:
No discusion.
Speaker: Chongfeng Xie
Luigi: An implementation report and a seperate draft for the
RPKI-to-Router (RTR) Protocol Data Unit (PDU) are required before moving
to WGLC.
Di Ma: Should consider incremental deployment.
Job: Manifest is a complex feature of the RPKI toolset. I recommend
against changing its structure.
Rob Austein (chat): I tend to agree with Job's comment about leaving
manifests alone. Skipping payload objects one doesn't care about is not
a big deal for a validator; changing the core validation logic that has
to walk the CA certs and CRLs and manifests is a much bigger deal, and I
suspect it's not worth it unless and until we hit major scaling limits.
Tim Bruijnzeels (chat): I agree that we should not change the manifest
format, at least not lightly. On the other hand I do agree that with
additional object types we need to think about the growth of rpki data -
especially if those new types would be short lived (lots of churn). I
don't have an answer yet - but I think the Erik idea will help at least
somewhat with scaling and deduplicating.
Speaker: Job Snijders
No disussion.
Speaker: Jishnu Roy
Job: CCR is stable and waiting for implem reports.
Luigi: Noted the yangdoctor review, but an implem report is needed to
start the WGLC.
Speaker: Libin Liu
Job: concerened about complexity vs cost.
Speaker: Jeffrey Haas & Bob Beck
Nan Geng: what is the link with the work about BGP over QUIC?
Jeff: Will take on that and similar work.
Job: Target creating a tiny WG for this.
Jeff: This makes sense.
Speaker: Jia Zhang
No discussion.
Speaker: Jia Zhang / Chongfeng Xie
Rob Austein (Chat): I am extremely skeptical of cost/benefit.
Speaker: Minglin Jia
Nan Geng: this is not a traditional SAV as there might be multiple hops
from subscribers. Maybe consider destiantion dimension.
Minglin: Maybe complicated
Lancheng Qin: Issue with the name and the solution does not leverage
RPKI.
Speaker: Sitong Ling
Job: Concerns with self-asserted info.
Warren Kumari (chat): This looks much more reasonable than I had
thought... Yes, I have to trust the assertion, but if X says it does
ROV, and Y doesn't, well, I might want to use X over Y.
Rob Austein (chat): I probably missed something (it's late) but this
smells a bit like the DNSSEC AD bit, which is useful in certain limited
contexts but only in those contexts.