> I'm still having trouble wrapping my head around 78703-2.  It's just huge.
>
> So the most obvious thing I see to break out is the should_warn_p
> changes.  If maybe_warn could be a distinct patch too, that would help.
>  It may not seem like much, but it really does help.  There may be more
> that sticks out after that.

As I mentioned in private, I don't think I can break up the patch
before stage 3 closes.  I spent pretty much the whole weekend
integrating it with the changes made on trunk since I started
working on it.  It's a very time consuming process mainly due
to the large number of tests and the amount of detail in the
diagnostics they are written to verify.

In hopes of helping with the review I've annotated the second part
of the patch with detailed comments explaining each hunk (look for
@@ at the beginning of the line).  Since the patch causes many
conflicts with the top of trunk I attach the file itself.  Please
let me know if there's something else I can do to make this review
less of a burden.

Martin