[PATCH] wined3d: Properly handle immediate NaN and Inf in GLSL backend.

[Guillaume Charifi]

Yes, of course. To better understand, here is a (more or less correct) 
equivalent between GLSL and C:

In GLSL:
uint a = 0xFFFFFFFFu;
float b = uintBitsToFloat(a);

In C:
unsigned int a = 0xFFFFFFFFu;
float b = *(float *)&a;

So you can see that the sign of the integer doesn't influence the float 
(and it is easier to use an uint than an int, because no extra - sign to 
deal with).


Le 14/06/2016 16:35, Cedric Blancher a écrit :
> Do you handle negative NaN and negative Infinity?
>
> Ced