[1/5] D3D9: Deal with failing surface creation

[Stefan Dösinger]

Am Freitag, 3. August 2007 18:18 schrieb H. Verbeet:
> On 03/08/07, Stefan Dösinger <stefan at codeweavers.com> wrote:
> > +        ok(SUCCEEDED(hr) || hr == D3DERR_INVALIDCALL,
> > "IDirect3DDevice9_CreateTexture: %s\n", DXGetErrorString9(hr));
>
> Same comment as for yesterday's test (even though that one apparently
> made it in), I don't think that's a useful test. As the patch comment
> says, the point of the test isn't to test CreateTexture behaviour.
Yes, but on the other hand it does not cost much. I still prefer to have tests 
which are on the first sight useless. There may still be some windows driver 
which returns an unexpected value. Even if we don't do much about it, except 
changing the test to not bail out, it may be of use(e.g. a game doesn't work 
on that driver and the vendor says this happens because the card is not fully 
directx9c capable). The tests documents a tiny, neglicable bit of 
information, but it still documents something.

If the CreateTexture call was the core of that test I'd check the device caps, 
check if the pixel format is supported and check the return value based on 
the supported formats.