Re: [w3c/ServiceWorker] API surface for #920 (50c3cdf)

That's basically to avoid "readers–writers problem" on the registration map. Although we hugely serialized the register/update/unregister jobs using a job queue, some of the algorithms running out of the jobs (mostly still off the main thread's method steps) can race with those jobs. "Atomically" isn't entirely avoiding this problem, but I don't think we have any ultimate way to do this better. I just looked up some text in HTML, and it even removed the storage mutex I suppose? (Maybe because it wasn't what UAs implemented I presume.) We definitely need a better way to clarify this behavior platform-wise.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/w3c/ServiceWorker/commit/50c3cdf306996d38eec91bb14f599db3b3f033c8#commitcomment-19300511