[RFC] Tightening up the type system

[Richard Kenner]

    My question to the FE folks is: what are the right semantics for
    checking MODIFY_EXPR?  Is compatible_types_p too strict?  Should we
    have had a cast operation in the above assignment?

My view is the compatible_types_p is exactly the right condition.  We
certainly ought to be able to get there for scalars, though aggregate
types sometimes have some corner cases that might be tricky.

    In fact, I would like to get to the point where we can simply call
    'gcc_assert (lang_hook.compatible_types_p (dest, orig))' when doing
    propagation.  Ideally, all the necessary type conversions should be
    exposed in the IL.

I agree (strongly).  I've been pushing for this sort of consistency for
a long time.