Some tests fail because of missing reporting of frame loader callbacks. Some examples: fast/loader/stateobjects/document-destroyed-navigate-back-with-fragment-scroll.html fast/loader/stateobjects/document-destroyed-navigate-back.html http/tests/loading/state-object-security-exception.html
Created attachment 44337 [details] proposed fix
I would like to advocate this solution to the delegates reporting problem. It's also a very good way of experimenting with new API - we can get to know the signals we might need, and can promote them to proper API in the appropriate time. Xan? =)
Attachment 44337 [details] did not pass style-queue: Failed to run "WebKitTools/Scripts/check-webkit-style" exit_code: 1 WebKit/gtk/webkit/webkitwebview.cpp:2065: Use 0 instead of NULL. [readability/null] [5] WebKitTools/DumpRenderTree/gtk/DumpRenderTree.cpp:586: Declaration has space between type name and * in char *frameName [whitespace/declaration] [3] Total errors found: 2
I think you want this to be a WebKitLoadStatus instead of a new signal.
(In reply to comment #4) > I think you want this to be a WebKitLoadStatus instead of a new signal. As discussed in real life ;D, I prefer keeping it as a semi-private signal for now, because otherwise we will be adding stuff to a public enum, and there's no going back. We can probably add it to the load status when we decide it is a good API. Also, I think it is not a good idea to have 1.1.18 Since tag, so I'll remove it when committing, if I get an r+.
Comment on attachment 44337 [details] proposed fix I agree this is the best solution for the moment, since we'll need to add A TON of signals for all tests to pass and we shouldn't commit to their stability without a proper period of testing. So r=me
This first step just went in as r55316.