Re: d3d9/tests: add check for D3DFMT_V8U8 support to avoid possible test crash.

[Stefan Dösinger]

On which GPU does that occur? Pixel Shader 1.x mandates V8U8 support afaik, so I am surprised this happens. I’m not opposed to this change conceptually though.
> - IDirect3D9_Release(d3d9);
> - if(FAILED(hr)) {
> -
>    skip("D3DFMT_V8U8 not supported\n");
> -
>    return;
> - }

Please watch the formatting. Add a space between if and (, and put the { on the next line. Yeah, I know, the existing code isn’t quite consistent :-( .

Am 02.11.2013 um 07:23 schrieb Alexander Dorofeyev :
> dlls/d3d9/tests/visual.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> <0002-d3d9-tests-add-check-for-D3DFMT_V8U8-support-to-avoid-.txt>

d3d9/tests: add check for D3DFMT_V8U8 support to avoid possible test crash.

[Alexander Dorofeyev]

On 11/02/2013 11:36 PM, Stefan Dösinger wrote:
> On which GPU does that occur? Pixel Shader 1.x mandates V8U8 support afaik, so I am surprised this happens. I’m not opposed to this change conceptually though.
>
> Am 02.11.2013 um 07:23 schrieb Alexander Dorofeyev :
> > dlls/d3d9/tests/visual.c | 10 ++++++++++
> > 1 file changed, 10 insertions(+)
> >
> > <0002-d3d9-tests-add-check-for-D3DFMT_V8U8-support-to-avoid-.txt>

The conditions under which the crash happened in generate_bumpmap_textures() were pretty artificial - a hack in wined3d/directx.c (select_fragment_implementation) to select the nvts/nvrc codepath and thus test/debug it, on Geforce8800. The only meaningful mention of D3DFMT_V8U8 I was able to find in DX8-9 sdk docs was in context of legacy/FF bump mapping - where they suggest checking formats before use. But if PS implies V8U8, then the check is unnecessary. A look at the online d3d9 caps database suggests de-facto PS and V8U8 are indeed linked.