On 2014.04.04 at 17:48 +0100, Nathan Sidwell wrote:
> On 04/04/14 17:38, Markus Trippelsdorf wrote:
>
> > I'm not sure that this is a good idea. This changes the existing
> > behavior of -Wnon-virtual-dtor and causes hundreds of new warnings when
> > building LLVM. Wouldn't it make more sense to move the 3rd ed Weffc++
> > behavior to the -Weffc++ flag alone?
>
> IIUC you're using -Wnon-virtual-dtor on its own, is that right? What are the
> class shapes where you're seeing this behaviour?
>
> AFAICT the -Wnon-virtual-dtor warning was an attempt to separate out that
> particular -Weffc++ warning, but it was broken.
For example:
markus@x4 tmp % cat test.ii
class Option
{
public:
virtual ~Option ();
};
template <int> class opt_storage
{
};
template <int = 0> class A : Option, opt_storage<0>
{
};
template class A<>;
markus@x4 tmp % g++ -Wnon-virtual-dtor -std=c++11 -c test.ii
test.ii: In instantiation of ‘class A<>’:
test.ii:12:16: required from here
test.ii:9:26: warning: base class ‘class opt_storage<0>’ has accessible
non-virtual destructor [-Wnon-virtual-dtor]
template <int = 0> class A : Option, opt_storage<0>
^
markus@x4 tmp % clang++ -Wnon-virtual-dtor -std=c++11 -c test.ii
markus@x4 tmp %
(Before your commit)
markus@x4 tmp % g++ -Wnon-virtual-dtor -std=c++11 -c test.ii
markus@x4 tmp %
--
Markus
