make 3.82 considered harmful

[Roland McGrath]

> Roland McGrath <roland@hack.frob.com> writes:
> 
> > How it happened to be happens to be wrong in a way that happens to coincide
> > roughly with the change to the behavior.
> 
> Is that a question?

No.

> > But the old behavior was easy to understand while the new behavior is
> > not.
> 
> The new rule is more intuitive since it doesn't depend on order.

My intuition is to expect that order is the only thing that matters.