Re: [w3c/ServiceWorker] [question]: how do CacheStorage save encoded responses? (Issue #1641)

I should note that during discussions related to how to quota track Blobs/Files in multiple storage buckets and potentially applying limits to structured serialization payloads, we've been moving towards standardizing quota usage so that the accounting is deterministic and predictable from the data put in and decoupled from the on-disk representation, including any browser optimizations.

As an example, right now Firefox has a number of optimizations so that if you put() a Blob/File into a db multiple times, only a single copy of the blob is stored and a refcount tracks the number of uses of that blob.  The net effect is the blob's size is only charged once even if stored N times.  If we undertake a change like that, the blob would be charged all N times.  This would be beneficial in terms of avoiding potential side-channel information leaks and improving cross-browser web-compat.  Browsers could leverage implementation savings by being more generous with quota, ideally in a way that exposes limited side channel information.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/w3c/ServiceWorker/issues/1641#issuecomment-1096965286
You are receiving this because you are subscribed to this thread.

Message ID: <w3c/ServiceWorker/issues/1641/1096965286@github.com>