Re: [PATCH 0/4] MR7505: d3dx9: Partially implement ID3DXEffect::SetRawValue().

[Matteo Bruni (＠Mystral)]

Matteo Bruni (@Mystral) commented about dlls/d3dx9_36/tests/effect.c:
> -
>        float f[32];
> -
>        DWORD dword[32];
> -
>    } expected_value;
> -
>    BOOL todo_hr;
> - } raw_value_tests[] =
> - {
> -
>    { NULL, 0, 4,
> -
>      { .f = { 1.0f } },
> -
>      { .f = { 1.0f } },
> -
>      .todo_hr = TRUE
> -
>    },
> -
>    { "_2", 0, 8,
> -
>      { .f = { 1.0f, 2.0f } },
> -
>      { .f = { 1.0f, 0.0f } },
> -
>      .todo_hr = TRUE
> -
>    },

FWIW, I added one more test here before reading further (test 6 basically covers the same):
```suggestion:-4+0
        { "_2", 0, 8,
          { .f = { 1.0f, 2.0f } },
          { .f = { 1.0f, 0.0f } },
          .todo_hr = TRUE
        },
        { "_2", 0, 20,
          { .f = { 1.0f, 2.0f, 0.0f, 0.0f, 3.0f } },
          { .f = { 1.0f, 3.0f, 0.0f, 0.0f, 0.0f } },
          .todo_hr = TRUE
        },
```
I also changed the code below to have the `memcmp()` go for `param_desc.Rows * 4 * param_desc.Elements` DWORDs and cleared `tmp` before each iteration.

I don't think adding this test is particularly necessary, or even a good idea. Same WRT changing the inner `for` loop end condition. I'll come back to the other change below.