[PATCH] Fix crash when TUI window creation fails

[Pedro Alves]

On 6/16/20 9:26 PM, Tom Tromey wrote:
>>>>>> "Pedro" == Pedro Alves <palves@redhat.com> writes:
> 
> Pedro> Is this "the Python function" referring to this PyObject_CallFunctionObjArgs
> Pedro> call (*)?
> 
> Yep.
> 
> Pedro>   std::unique_ptr<tui_py_window> window
> Pedro>     (new tui_py_window (win_name, wrapper));
> 
> Pedro>   gdbpy_ref<> user_window
> Pedro>     (PyObject_CallFunctionObjArgs (m_constr.get (),
> Pedro> 				   (PyObject *) wrapper.get (),
> Pedro> 				   nullptr));
> Pedro>   if (user_window == nullptr)
> Pedro>     {
> Pedro>       gdbpy_print_stack ();
> Pedro>       return nullptr;
> Pedro>     }
> 
> window-> set_user_window (std::move (user_window));
> 
> Pedro> Wouldn't it make sense to to move the tui_py_window
> Pedro> allocation until after the Python call instead?
> 
> No, because the TUI window object is passed to the user's constructor,
> and that constructor can write text there, etc.
> 
> Pedro> (*) - I think the comment would gain from mentioning
> Pedro> it more specifically, like 'the Python "contructor" function'
> Pedro> or something like that.
> 
> I'll update it.

Thanks for the clarification.  Feel free to merge with that
update then.

Thanks,
Pedro Alves