Matteo Bruni (@Mystral) commented about dlls/d3dx9_36/tests/surface.c:
},
},
/*
* Same as decoding for BMP/DIB, if the alpha channel is set to a
* non-zero value on any pixel we get a different format.
*/
{ D3DFMT_A8R8G8B8, NULL,
{ { D3D_OK, D3DFMT_A8R8G8B8 },
{ D3D_OK, D3DFMT_X8R8G8B8 },
{ D3D_OK, D3DFMT_A8R8G8B8, .todo_hr = TRUE },
{ D3D_OK, D3DFMT_A8R8G8B8, },
{ D3D_OK, D3DFMT_X8R8G8B8, .todo_hr = TRUE },
{ D3D_OK, D3DFMT_A8R8G8B8 },
{ D3D_OK, D3DFMT_A32B32G32R32F, .todo_hr = TRUE },
{ D3D_OK, D3DFMT_A32B32G32R32F, .todo_hr = TRUE },
}, 0xff
Let's put the initial value before the expected results, the difference between this block and the previous one wasn't immediately noticeable to me (the comment helped sort it out quickly enough, but still I think it would be an improvement). I'd even explicitly write out the value `0` instead of e.g. using designated initializers for the `expected` part.