On Wed Apr 5 18:01:51 2023 +0000, Rémi Bernon wrote:
This could perhaps test whether `QueryInterface` is really needed in `ProcessMessage`. Otherwise I'd say if it's not tested then it's probably not needed on the implementation side. Also, any reason not to add the tests upfront? I think it'd make the changes more obvious if the tests were added upfront or in a lockstep, and the implementation incrementally fixing the todo_wine.
I am sorry, I am not entirely sure what do you mean here: how can we in principle avoid QueryInterface on the implementation side, once we receive an IUnknown but for now only going to support IDXGIDeviceManager (that probably can be also d3d9 manager in theory but I indeed didn't test it). The only reason I added this skip here as that `MFT_MESSAGE_SET_D3D_MANAGER`fails on Testbot machines with these errors while that doesn't happen here on real hardware (and that's what games apparently do).