Re: gtkmm2.4 and MSVC (again)

["Timothy M. Shead" <tshead k-3d com>]

Murray Cumming wrote:
> This is strangely similar to a dynamic_cast<> problem that existed in g++
> 2.96:
> http://www.gtkmm.org/docs/gtkmm-2.4/docs/FAQ/html/index.html#id2543380
>
> So it might be a compiler bug with use of dynamic_cast<> in the
> constructor of with classes that use multiple inheritance. But to reach
> that conclusion we would need a non-gtkmm test case.

This begins to look like the problem - enclosed is a simplification of 
Matt's test case that should be clearer to look at, I am currently 
fiddling-around to try and create a C++-only test case.

> In the meantime, please make sure that any extra project file changes are
> in our CVS, so that we are starting at the same point. I have access to
> MSVC++ .Net 2003, so I might take a look if your own efforts don't reach a
> conclusion.

Unless I've got the wrong branch, the MSVC files aren't in gtkmm CVS 
yet, are they?

Cheers,
Tim

#include <gtkmm/main.h>
#include <gtkmm/window.h>
#include <iostream>

// As written, this program will generate a "no RTTI data!" at runtime when
// compiled with Visual Studio .NET 2003
//
// There are two ways to eliminate the exception, either:
//
// * Comment-out the member variable derived::x, or
// * Comment-out the call to show() in base::base(), and un-comment the call to show in derived::derived()

class base :
	public Gtk::Window
{
public:
	base() :
		Gtk::Window(Gtk::WINDOW_TOPLEVEL)
	{
		try
			{
				show();
			}
		catch(bad_typeid x)
			{
				std::cout << "Exception " << x.what() << std::endl;
			}
	}
};

class derived :
	public base
{
public:
	derived() :
		base()
	{
		// show();
	}

protected:
	int x;
};

int main(int argc, char **argv)
{
	Gtk::Main m(&argc, &argv);

	derived w;
	m.run();

	return 0;
}