Re-land http://trac.webkit.org/changeset/83007. Needed to lock v8 before using.
Created attachment 88726 [details] Patch
Adam, Darin: The only change from the previous patch you reviewed is the addition of v8::Locker in WebFrameTest.cpp > FrameForEnteredContext.
Oh, and the reason I didn't see this earlier is that the build bots randomize the order of the test runs and in order for this CHECK to be hit, v8 has to be used with a locker once before my code ran. There is one test that does that.
Comment on attachment 88726 [details] Patch Yep. Sorry I didn't catch the Locker issue. It's bene a while since I worked on code that enters V8 from an empty stack.
With bug 58110 resolved, the Lockers are no longer needed. I've verified Aaron's patch w/o lockers does not fail on: webkit_unit_tests --gtest_repeat=-1 --gtest_shuffle which was the case before. So the patch, sans the Locker line, can be landed now.
Created attachment 88883 [details] Patch New pach with removed references to v8.h and v8::Locker. This is the same patch that was previously reviewed and landed.
Attachment 88883 [details] did not build on chromium: Build output: http://queues.webkit.org/results/8377161
/Projects/CrMacEWS/Source/WebKit/chromium/tests/WebFrameTest.cpp:136: error: 'HandleScope' is not a member of 'v8' Looks like we still need v8.h.
Attachment 88883 [details] did not build on chromium: Build output: http://queues.webkit.org/results/8377172
Created attachment 88893 [details] Patch fffff...