PDA

View Full Version : GLSlang approved by ARB



harsman
07-01-2003, 04:21 AM
3DLabs have released their GL2 SDK (http://www.3dlabs.com/support/developer/ogl2/index.htm) today and claim that the latest version of glslang was approved by the ARB at the June meeting. There's a revised version of the spec and some examples.

I got the simple shader example working on my radeon 9500 Pro by hacking around a bit. Basically I just took out some error checking (the ATI driver doesn't accept some of the newer enums like GL_OBJECT_LINK_STATUS_ARB) and changed the extension name suffixes to GL2. I also had to change the vertex_shader and fragment_shader tokens to 0x7010 and 0x7011 respectively. Only the wood and simple shader seems to work correctly, the other ones look strange.

I skimmed through the new spec but didn't notice anything new, probably just minor changes. Still neat to muck around with glslang! Hopefully the new extensions will come up at the registry soon and we'll be able to play with this in the next driver release...

[This message has been edited by harsman (edited 07-01-2003).]

V-man
07-01-2003, 06:54 AM
What did you do exactly?
Can you use the GL2 funtions in the drivers.
There was some thread about hidden GL2 extensions in the new catalyst drivers so I'm assuming they are usable.

stefan
07-01-2003, 07:24 AM
Isn't it a bad joke? The site says approved on 11th of june - and no one of us knew until 1st of july. Looks like perfect marketing and perfect communication!

Mazy
07-01-2003, 07:29 AM
I agree, how come that ARB needs one month to publish the meeting notes? its not that much text..

harsman
07-01-2003, 07:32 AM
The glslang stuff isn't exposed in the drivers, but you can retrieve the function pointers with wglGetProcAddress. There are minor differences between the ARB approved API used in the example from the 3DLabs sdk and the one exposed by ATI's drivers (like a GL2 versus ARB suffix for function names) and I just changed thos to get it to work.

Stefan, I'm not surprised it took this long for the info from the ARB meeting to get public. It usually takes about a month before the meeting notes are posted here at opengl.org for example.

stefan
07-01-2003, 07:40 AM
I can imagine that it takes some time to write down the minutes send it to all attendees wait until they mail back, make corrections etc.

It would have been enough to put a small (sticky) headline on the front page saying "GLSlang has been approved." and add that the minutes will be published in some weeks. Nothing more - just let those know who work with (and care about) OpenGL.

Lurking
07-01-2003, 09:46 PM
i guess the only question now is how fast ATI and Nvidia will write drivers that can handle this new language. I would hope really soon. I cant wait to try it out (though it will only work on FX's and ati's high boards)

- Lurking

NitroGL
07-01-2003, 10:56 PM
Originally posted by Lurking:
though it will only work on FX's and ati's high boards

I had no idea ATI's 9500 was "high-end" http://www.opengl.org/discussion_boards/ubb/eek.gif
http://www.opengl.org/discussion_boards/ubb/biggrin.gif

Mazy
07-02-2003, 05:28 AM
Have i missed something, or am i just lazy here, but where are the inverse and transpose of matrices? Do i have to do a inverse by myself now? i think that the language could have followed the arb_vertex and fragment program a bit more, especially in the in and out states and variables.

m2
07-05-2003, 03:22 AM
Originally posted by Mazy:
where are the inverse and transpose of matrices?

if you are looking for (MV)^t^-1, it's right there: gl_NormalMatrix.

Mazy
07-05-2003, 04:19 AM
but sometimes you really need the inverse, not only the inverse transpose ( and gl_NormalMatrix ins only the upper 3x3 vallues anyway, so no translation is possible)