Re: [PATCH 0/1] MR4017: d3d10/effect: Add support for shift instructions.

[Matteo Bruni (＠Mystral)]

Matteo Bruni (@Mystral) commented about dlls/d3d10/effect.c:
> - {
> -
>    unsigned int v = arg1[instr->scalar ? 0 : i] >> (arg2[i] % 32);
> -
>    retval[i] = *(float *)&v;
> - }
>
> +}
> -
>
> +static void pres_ushr(float **args, unsigned int n, const struct preshader_instr *instr)
> +{
> - unsigned int *arg1 = (unsigned int *)args[0];
> - unsigned int *arg2 = (unsigned int *)args[1];
> - float *retval = args[2];
> - unsigned int i;
> -
> - for (i = 0; i < instr->comp_count; ++i)
> - {
> -
>    unsigned int v = arg1[instr->scalar ? 0 : i] >> (arg2[i] % 32);

I guess this always does the right thing but wouldn't it be better to use a plain (signed) int here for 'v'?