> It actually needs a tiny patch on windows (nonvirtual_fun is mangled as > static function). Unfortunately I did not find a way to sensibly declare > member "a" in D. There is another bug that a derived class is using a > wrong virtual function table offset, but that should be fixable.
It's very important that dmd will be able to call C++'s non virtual functions and it's ctors/dtors. But there must be some significant reasons why Walter didn't accept the patch.
