...
[01:51:07 CDT(-0500)] <Bosmon> Unknown.... we would have to "fiddle" that too : P Although testing performance on these micro-operations is always extremely fraught
[01:51:26 CDT(-0500)] <Bosmon> We really need a performance benchmark which assesses the performance of the framework as a whole on a sizeable task
[01:51:40 CDT(-0500)] <Bosmon> But that would reveal a number of other more serious problems
[01:51:54 CDT(-0500)] <Bosmon> I'm not sure that actually a "constructor" or "prototype"-based detection can be performant either
[01:52:09 CDT(-0500)] <Bosmon> Since if we require cross-frame resistance, it means calling "toString" on them
[01:52:14 CDT(-0500)] <Bosmon> Which can result in truly huge representations