Re: [5/5] WineD3D: Add sincos support to arb shaders

[Ivan Gyurdiev]

Aren't most of these 2.0 and 3.0 instructions ?
What's the goal of adding them to ARB - you won't be able to implement 
full 2.0/3.0 support in ARB.

I think most of these were left unimplemented on purpose.

Ivan

[5/5] WineD3D: Add sincos support to arb shaders

[Stefan Dösinger]

Am Donnerstag, 27. September 2007 02:57:01 schrieb Ivan Gyurdiev:
> Aren't most of these 2.0 and 3.0 instructions ?
> What's the goal of adding them to ARB - you won't be able to implement
> full 2.0/3.0 support in ARB.
> We can implement 2.0(but not 2.a/b). Whenever it's easy to do, I'm adding 
> 3.0-Specific behavior too. The reason is that GLSL is a bloody pain on MacOS, 
> with ARB I can bypass many abstraction layers that have a huge number of 
> software fallback pitfalls.

[5/5] WineD3D: Add sincos support to arb shaders

[Ivan Gyurdiev]

Stefan Dösinger wrote:
> Am Donnerstag, 27. September 2007 02:57:01 schrieb Ivan Gyurdiev:
> > Aren't most of these 2.0 and 3.0 instructions ?
> > What's the goal of adding them to ARB - you won't be able to implement
> > full 2.0/3.0 support in ARB.
>
> We can implement 2.0(but not 2.a/b). Whenever it's easy to do, I'm adding 
> 3.0-Specific behavior too. The reason is that GLSL is a bloody pain on MacOS, 
> with ARB I can bypass many abstraction layers that have a huge number of 
> software fallback pitfalls.

Are you saying you intend to advertise 3.0 support on ARB only?
If not, then why add any 3.0 features at all ?

Ivan

[5/5] WineD3D: Add sincos support to arb shaders

[Stefan Dösinger]

Am Donnerstag, 27. September 2007 13:02:31 schrieb Ivan Gyurdiev:
> Are you saying you intend to advertise 3.0 support on ARB only?
> If not, then why add any 3.0 features at all ?
> No, I certainly do not intend to advertise 3.0 because we cannot implement 
> things like branching, dsx/dsy or vertex textures, but I'll propably 
> advertise 2.0 some day, if I feel like causing a lot of extensions(Or I think 
> that our shader tests are good enough to safely do this).

The ARB shader backend tries to eat everything it is fed with though, even 3.0 
shaders. Some games require a specific shader model, but use only a subset of 
it that is supportable. Having the code to support some 3.0 specials makes my 
life easier since I can just go and lie about the caps on macos for specific 
games(e.g. in crossover). I don't put special focus on SM 3.0 support, but if 
I come across something that is easy to add, I'm adding it. Also, some day we 
may implement the 3.0 features using the NV extensions.

Of course if it's preferred to keep the arb shader backend clean from version 
== 3.0 checks I can keep the things I have in my junk branch, as long as I 
don't have to explain why I'm not submitting $GAMEFIX to Wine. Pretending SM 
3.0 support is a hack, implementing the parts of 3.0 we can implement 
properly isn't IMHO.