Yves: Cache is not only for clients, but to ... on the server side. It would be good to have exemption of the partitioning for sites that are explicitly asking for public cache. ... control public... do we want a fake delay retrieval in that case?

Sangwhan: For CDN?

Yves: Not just CDN... also for sites with a lot of traffic...

... Mostly non-CDN and non-big-company websites.

... Wondering why they are ...

Tess: Worth raising the CDN, etc. question. Tension between goal of partition, namely hiding from the sites that they are both embedded in the same resource, preventing data sharing.

... As far as eTLD+1 vs origin... this is me guessing at rationale... general desire to avoid building new platform features that depend on the public suffix list. Arguments either way. Not obvious what the better move is.

Yves: Need to be an alignment between ??? -

Hober: prefer them to match up...

Yves: I'll write an issue on the cache control public... server-side... that's all for me on that specification

Lukasz: (copied from slack)

Looks like a nice privacy-improving spec. Is there any way to standardise partitioning keying across implementations? Seems there is an Apple precedent and this of course opens a question: should this precedent be followed? Are there any drawbacks of having different implementations using different keying techniques?

... Would we want ... extension to this partition?

... People might be questioning what kind of privacy or security in case an optional artificial delay in delivery (retrieval) will be deployed so maybe not ask about this particular thing and see how it goes

Tess: we have a number of good comments here to make on the issue.

Peter: Folks want to post those comments on the issue?

Tess: sounds good to me.

Peter: so, who...

Tess: Yves is writing up the cache control one. Lukasz is adding his. I can try asking again about origin vs. registerable domain decision.

Peter: Ok, I'll mark this as pending feedback, and we can come back when we get feedback.