Splitting this out from my work in bug 960671 for use in bug bug 1066361.

jryans, this is the same actor patch(es) you reviewed in bug 960671, but I just want to rubber stamp it again since it is in a new bug now. I split the toggling of recording into a start and stop method, as you asked. This is a combined patch of the allocations changes, the force GC/CC changes, and a tiny method exposing Debugger.Memory.prototype.takeCensus over the RDP. It was getting too hard to keep everything separate, sorry. https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=cba18023568c

Comment on attachment 8489689 [details] [diff] [review] memory-actors.patch Review of attachment 8489689 [details] [diff] [review]: ----------------------------------------------------------------- Still seems good to me!

Unless you're checking this in yourself, I think you should add a commit message to the patch.

remote: https://hg.mozilla.org/integration/fx-team/rev/cc361e77343d

sorry had to back this out for test failures like https://tbpl.mozilla.org/php/getParsedLog.php?id=48249937&tree=Fx-Team

Hrm, I can't repro locally. Digging into the test logs...

Ok just adding a check that the frame does exist in test_memory_allocations_01.html. Let's see how this turns out: https://tbpl.mozilla.org/?tree=Try&rev=d00453950900

And I just realized I didn't include mochitest-chrome in that try push. Sigh...

Looks like adding that null check fixed it!

landed as https://hg.mozilla.org/integration/fx-team/rev/f6c42abb5457