<rdar://problem/18846532>
Created attachment 241829 [details] Also acquire the activity token when we have active network requests on progress.
Comment on attachment 241829 [details] Also acquire the activity token when we have active network requests on progress. It seems very wrong to keep the Web process from going to sleep for very long XHR/etc., seems like this will just result in a lot of very awake processes consuming all of your power, no?
(In reply to comment #2) > Comment on attachment 241829 [details] > Also acquire the activity token when we have active network requests on > progress. > > It seems very wrong to keep the Web process from going to sleep for very > long XHR/etc., seems like this will just result in a lot of very awake > processes consuming all of your power, no? With the changes in https://bugs.webkit.org/show_bug.cgi?id=138582, we won't track long XHR or sub-resources after the main frame has loaded, and these sub-resources won't keep the process awake.
It may be worth re-titling the bug for accuracy, "active network request" may not be the best description for what progress tracker does.
This patch has the risk of holding web process too long if there is a long lasting XHR that has started before main frame has loaded. I will look at a different approach.