Backport fix for spurious anonymous ns warnings PR29365 to 4.2?

[Ian Lance Taylor]

"Aaron W. LaFramboise" <aaronavay62@aaronwl.com> writes:

> I discovered PR29365 today while testing 4.2.0 RC2 on a client's
> codebase.  This causes some of my code, based on the popular pimpl
> idiom, to generate warnings, even with no warning flags specified.  If
> there's a way to turn it off without patching the source, I can't find
> it.
> 
> Given the pervasiveness of the pimpl idiom in C++, and the increasing
> use of anonymous namespaces in modern code, it is possible that this
> will be a serious problem for some C++ developers, that may hinder the
> adoption of 4.2.  It's not hard to imagine that this could come up in
> a lot of other weird places that uses anonymous namespaces, too.
> 
> Does this qualify as a regression?  Does anyone else think this is a
> serious problem?

I agree that it would be appropriate to backport the patch to gcc 4.2.

Ian