Comment 4 for bug 297563

After an IRC discussion with Mr. Collins, and an evaluation of all the decorators that are currently in use, I agree that forwarding attribute access is probably the right way to address this specific problem. The decorators in question don't really define any interesting attributes, and the mechanism is not in wide enough use to pose a serious problem with conflicting attributes; if that were to happen, making individual decorators be as non-intrusive as possible so they won't conflict with anything isn't a terribly difficult problem.

In any event, the main point is that coming up with a more general mechanism for test-run extensibility is out of scope for fixing this issue.