g++ and aliasing bools

[Richard Kenner]

    I would really like to see g++ doing better in C++-aliasing sensitive
    codes (i.e., Haney Speed) and I believe that any effort in this area
    should be supported and encouraged as much as possible.

I don't think anybody disagrees, but that doesn't mean that a proof of
correctness isn't needed.

As Mark points out, there are only a relatively small number of things in
GCC where we can prove things.  We ought to take advantage of those.

I am very much against the idea of defining a change to be "correct" if
it doesn't cause any regression test failures.  You have to be able to make
an argument that a change is correct independently and the regression tests
serve as a debugger of (among other things) that proof.