[RFC] Tightening up the type system

[Richard Kenner]

    It's an interesting view.  I'm pretty sure that we violate this is the
    Java FE in a few places, but perhaps we shouldn't.  The trouble is
    that the GENERIC type system has never been so well-defined.

That's true, however, I've always treated GENERIC's type system as
being very strict: the types must *exactly* agree.  Certainly the Ada
front end goes to a lot of trouble to make that happen in the trees it
generates.  I think there has been some code that assumes
type-correctness, but not much.

In GIMPLE, we relax this to allow any types that are type_compatible_p.

    I'm sure that enforcing this would break things and it would take some
    time to find and fix them all. 

On the other hand, *not* enforcing this looks like it's causing problems
for the optimizers.  Those problems are going to be harder to find and
fix than typing violations in the front end.