Re: [5/5] WineD3D: Pixel shader 1.x constants are clamped to [-1;1]

[Alexandre Julliard]

Stefan Dösinger  writes:
> It is hard to judge if this patch will fix anything. A problem is that 
> clamping the constants on loading makes constants loading more expensive. 
> Luckilly though there aren't many ps 1.x constants(8 max). It is hard to say 
> if there are any games that use constants > 1.0, but incorrect constant 
> loading can have many bad effects.

This one fails make test for me:

../../../tools/runtest -q -P wine -M d3d9.dll -T ../../.. -p d3d9_test.exe.so visual.c && touch visual.ok
visual.c:3068: Test failed: quad 1 has color 00bfbf7f, expected 0x00bfbf80
visual.c:3070: Test failed: quad 2 has color 00bfbf7f, expected 0x00bfbf80
visual.c:3073: Test failed: quad 3 has color 00bfbf7f, expected 0x00bfbf80
visual.c:3077: Test failed: quad 4 has color 00bfbf7f, expected 0x00bfbf80
visual.c:3207: Test failed: quad 1 has color 007f7f00, expected 0x00808000
visual.c:3209: Test failed: quad 2 has color 007f7f00, expected 0x00808000
visual.c:3211: Test failed: quad 3 has color 007f7f00, expected 0x00808000
visual.c:3214: Test failed: quad 4 has color 00bfbf7f, expected 0x00bfbf80
make[2]: *** [visual.ok] Error 8

[5/5] WineD3D: Pixel shader 1.x constants are clamped to [-1;1]

[Stefan Dösinger]

Am Donnerstag, 13. September 2007 12:43:54 schrieb Alexandre Julliard:
> Stefan Dösinger  writes:
> This one fails make test for me:
> quad 4 has color 00bfbf7f, expected 0x00bfbf80 make[2]: *** [visual.ok]

This looks like a precision problem (7f vs 80). I will make the test less 
picky.