Anonymous types and [basic.link] for nonlocal variables

[Joe Buck]

I wrote:

> In my view, anything that breaks QT is a regression, no matter how
> many standards documents we can point to as justification.  Certainly
> -pedantic should object to such usage, and perhaps future QT versions
> should avoid such usage if possible, but we need to build QT.

On second thought, this is too strong.  There are certainly cases where
tightening up on C++ parsing might break a popular free software package;
I don't think such cases should be considered bugs.  But in this case,
we are essentially removing an extension, not correcting an accidental
mistake.