explainers-by-googlers/chips-spec

Draft specification for Cookies Having Independent Partitioned State (CHIPS)

chips-spec's Issues

Computing cookie partition key in service workers

At the time of writing this, workers always use their own origin as their site for cookies, regardless of the context they are registered in. This will change when browsers ship storage partitioning, and service workers will derive their site for cookies from their storage key instead of the worker origin. This is a nice improvement to service workers since it prevents workers registered in 3P contexts from becoming SameSite leaks.

Given this effort, I think that it makes sense to specify somehow that partitioned cookies should use the service worker's storage key, not the worker origin, to compute the cookie partition key.

Should probably say "kibibytes" instead of "kilobytes"

See privacycg/CHIPS#59 (comment) and, if you're like me, https://www.logicmonitor.com/blog/what-the-hell-is-a-kibibyte

cc @annevk @DCtheTall

Adding Cross-Site Ancestor Chain bit

As discussed in the Privacycg, we are looking to add a cross-site ancestor chain bit to CHIPS to address the potential for click jacking attacks in an A->B->A context and align the keying of CHIPS with storage the storage partition key.