Cross-Origin Isolation in S&P section

[yoavweiss]

We got feedback from @camillelamy that the spec should mention the granularity of timers in CrossOrigin Isolated contexts. We should add wording to that effect, as well as point to HR-time-3 rather than 2 (as it already includes COI in its processing model)

[npm1]

Why should the spec mention the granularity of timers? This should be under HR-Time, and we don't really mention timer granularity at all in this spec. DOMHighResTimeStamp already links to HR-Time, where it should state granularity of timers.

I do agree with changing the links to HR-3.

[yoavweiss]

@camillelamy - Would the above be sufficient from your perspective?

[camillelamy]

Sorry for the late reply, just back from being OOO. I think my main issue is that HR-3 does not mention that the threat model in crossOriginIsolated contexts vs regular contexts allows for different timer resolution. Now wether this information should be in this spec or we should consider adding it to HR-3, I don't know.

[npm1]

Ok, I think it makes more sense in HR-Time. Performance-Timeline is not the only spec that uses DOMHighResTimeStamp, and all users of that would find that information relevant. @yoavweiss is it possible to move this issue to hr-time?

[yoavweiss]

We can add something to https://www.w3.org/TR/hr-time-3/#clock-resolution that outlines that timer resolution should vary based on COI status (although this is already described in the normative processing model)