Re: Fix disable-checking bootstrap problem on ppc-darwin

[Daniel Berlin <dberlin at dberlin dot org>]

> > If true, another solution is to ensure that for any commutative
> > operation that we're hashing, that we use a commutative hash.  So
> > instead of always
> >
> > 	for each operand (op) {
> > 	  hash <<= n;
> > 	  hash ^= munge(op);
> > 	}
> >
> > do
> >
> > 	if (commutative (operation)) {
> > 	  hash <<= n;
> > 	  hash ^= munge(operation);
> > 	  for each operand (op) {
> > 	    hash += munge(op);
> > 	  }
> > 	} else {
> > 	  ...
> > 	}
>
> This works for the _hash function, but the _eq function still needs to 
> somehow work out whether the two lists are really equivalent or just 
> happen to have the same hash value.

Right, and I believe the whole problem in your case was that you had 
two things that hashed the same, but did not actually have the same 
vuse lists, right?