Re: [ddraw/tests 2/2] Mark some more tests are broken() on W2K

[Stefan Dösinger]

Am Mittwoch, 11. März 2009 12:07:27 schrieb Paul Vriens:
> - ok(hr == D3D_OK ||
> - broken(hr == D3DERR_SCENE_IN_SCENE), /* W2K */
> - "IDirect3DDevice7_BeginScene failed with %08x\n", hr);

I don't like that one. I think I wrote those tests on Win2K partially, so I'd 
blame it on Vmware. Its ok to skip checking the colors to reduce follow-up 
tests, but I dislike accepting random errors because Vmware returns them. If 
you find a real Windows hardware driver that returns this error its ok to 
accept D3DERR_SCENE_IN_SCENE as a return value

[ddraw/tests 2/2] Mark some more tests are broken() on W2K

[Paul Vriens]

Stefan Dösinger wrote:
> Am Mittwoch, 11. März 2009 12:07:27 schrieb Paul Vriens:
> > - ok(hr == D3D_OK ||
> > -
> >   broken(hr == D3DERR_SCENE_IN_SCENE), /* W2K */
> > -
> >   "IDirect3DDevice7_BeginScene failed with %08x\n", hr);
>
> I don't like that one. I think I wrote those tests on Win2K partially, so I'd 
> blame it on Vmware. Its ok to skip checking the colors to reduce follow-up 
> tests, but I dislike accepting random errors because Vmware returns them. If 
> you find a real Windows hardware driver that returns this error its ok to 
> accept D3DERR_SCENE_IN_SCENE as a return value

So what do you propose? Making it broken(FAILED(hr)) ?

[ddraw/tests 2/2] Mark some more tests are broken() on W2K

[Henri Verbeet]

2009/3/11 Stefan Dösinger :
> Am Mittwoch, 11. März 2009 12:07:27 schrieb Paul Vriens:
> > - ok(hr == D3D_OK ||
> > - broken(hr == D3DERR_SCENE_IN_SCENE), /* W2K */
> > - "IDirect3DDevice7_BeginScene failed with %08x\n", hr);
>
> I don't like that one. I think I wrote those tests on Win2K partially, so I'd
> blame it on Vmware. Its ok to skip checking the colors to reduce follow-up
> tests, but I dislike accepting random errors because Vmware returns them. If
> you find a real Windows hardware driver that returns this error its ok to
> accept D3DERR_SCENE_IN_SCENE as a return value

Isn't that the point of marking it broken() though?

I'm pretty sure the failures are specific to vmware rather than w2k.
D3DERR_SCENE_IN_SCENE suggests there was a previous BeginScene()
without a matching EndScene() though, so you'll want to make sure this
isn't caused by eg. a skip() in an earlier test.