(In reply to Tim Nelson from comment #12) > Oh, one more comment - is there anything else I can do to help track down > this bug further? Ideally you could use git to bisect between 3.6 and 3.7 to see what change caused the regression.