Make HTTPCookieStorage operate in the UIProcess in absence of a WebProcessPool Currently it creates transient process pools, just to create networking processes, just for cookie operations. That's nuts - It should operate in the UIProcess and then just flush cookies if an associated WebProcessPool comes into existence.
<rdar://problem/31102330>
Created attachment 304819 [details] Patch
Comment on attachment 304819 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=304819&action=review > Source/WebCore/ChangeLog:8 > + No new tests (No behavior change). I'm not entirely convinced. > Source/WebKit2/UIProcess/API/APIHTTPCookieStore.cpp:63 > + callOnMainThread([completionHandler = WTFMove(completionHandler), allCookies = WTFMove(allCookies)]() { Cookies contain a lot of Strings. Should we do an isolated copy here?
(In reply to comment #3) > Comment on attachment 304819 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=304819&action=review > > > Source/WebCore/ChangeLog:8 > > + No new tests (No behavior change). > > I'm not entirely convinced. I assure you - "Setting cookies without a process pool" didn't work before, and still doesn't work after this patch. :) It just gets us closer. We can chat in person about what's left. > > > Source/WebKit2/UIProcess/API/APIHTTPCookieStore.cpp:63 > > + callOnMainThread([completionHandler = WTFMove(completionHandler), allCookies = WTFMove(allCookies)]() { > > Cookies contain a lot of Strings. Should we do an isolated copy here? I think that'd be more expensive, right? (I'm unsure why you're making the request)
Created attachment 304829 [details] Patch
Comment on attachment 304829 [details] Patch Clearing flags on attachment: 304829 Committed r214127: <http://trac.webkit.org/changeset/214127>
All reviewed patches have been landed. Closing bug.