Thought about this while on the plane. I need to read the code again to prove to myself that this is indeed the case. The race I'm worried about is a failed compilation causing us to create a new CodeBlock before we iterate all the callbacks.
Created attachment 307119 [details] patch
Comment on attachment 307119 [details] patch r=me. Maybe we should also have the plan null out the tasks after it runs them.
Created attachment 307120 [details] patch for landing
Comment on attachment 307120 [details] patch for landing adding the clear lgtm.
Comment on attachment 307120 [details] patch for landing Clearing flags on attachment: 307120 Committed r215368: <http://trac.webkit.org/changeset/215368>
All reviewed patches have been landed. Closing bug.
http://trac.webkit.org/r215373