Re: [w3ctag/design-reviews] First-Party Sets (#342)

@hadleybeeman Thank you for checking!

We announced a pause in our rollout of the feature in Chrome because we discovered a regression in some of our metrics. Our analysis suggests that the regression may be due to the fact that our implementation currently blocks all network requests on the completion of the set change handling algorithm [described here](https://wicg.github.io/first-party-sets/#handling-changes). The algorithm clears out all site data for any site that is leaving an existing set, so we had previously chosen the more conservative route of blocking all network requests until this operation was complete, in order to avoid being in a state where a site that is about to have its data cleared receives requests before the clearing is complete. We are currently waiting for our tentative fix (optimizing this implementation) to propagate, and hope to validate it in the next couple of weeks.

With regards to how that impacts this review; your question made us realize that we had previously neglected to specify this implementation detail in the algorithm, so we've now written up a PR at WICG/first-party-sets/pull/169 that describes this behavior, along with a note recommending that user agents optimize for request latency by being smarter about what requests/fetches to block.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/w3ctag/design-reviews/issues/342#issuecomment-1663097709
You are receiving this because you are subscribed to this thread.

Message ID: <w3ctag/design-reviews/issues/342/1663097709@github.com>