some sort of progress reports: continue ARB meeting notes

Hi,
i think we need some sort of pogress reports where OpenGL is going and what the plans are for the future. I understand that there were legal issues with the meeting notes, but there has to be a way to let the comunity know about this stuff. Maybe cut the critical stuff down but give feedback.

I feel that this is an important issue with a ‘rival’ API like D3D10. Many developers may switch because they think OpenGL may not deliver new features at the pace Microsoft can.

Some points may be:

  • virtual memory model (superbuffers continued?)
  • uniform shader adaptation (glsl 1.2, 2.0? or ARB_xxx_program continued?)
  • OpenGL3.0 (get rid of legacy stuff like pure OpenGL2.0 promised?)
  • could go on like this

For me a clean cut to a than again small und clean API (gl3) would be the biggest wish.

An active discussion is appreciated :wink:

-chris

There is not much to discuss as you are obviously right :slight_smile:

Hovewer, I doubt ARB will have the courage about designing a small and clear API, it would be too much stress and risk for them. The original 3dlab sproposal was already very good and could be implemented. Well, that didn’t happen.

Personally, I think they could rely more on the community in this questions. ARB is just too slow…

Many developers may switch because they think OpenGL may not deliver new features at the pace Microsoft can.
This isn’t a recent issue; it’s always been that way. The people who need new features faster have already switched to D3D.

The original 3dlab sproposal was already very good and could be implemented.
As an API, it was perhaps nice. As an actual specification from which implementations would have to be written? Not so much.

There’s more to designing a spec than making a decent API. You have to take into account the vagaries of different hardware, and 3DLabs doesn’t know enough about each other’s hardware to be able to design a good specification for all hardware.

Originally posted by Korval:
Many developers may switch because t
There’s more to designing a spec than making a decent API. You have to take into account the vagaries of different hardware, and 3DLabs doesn’t know enough about each other’s hardware to be able to design a good specification for all hardware.
I totally agree. But this API design could be (or have been) the starting point for the redesign.

I found a very good summary, why a redesign is needed and how this can make the API slim again, in the meeting notes of march 2004 (2 years ago!). There was this presentation. Look at the pages starting with page 12. Very good points in my opinion.

I still hope to get some feedback from the ARB.

Do you know where I can get the 3dlabs whitepapers on OpenGL 2.0? I would like to read them once again, do some experiments.

And by the way, I recently run into
this paper It’s about a functional SL. It’s genious! I wish I had learned the functional languages earlier. They are the best candidate for shading language, why wasn’t that adopted?

There was this presentation.
It says “why are texture attributes defined by the storage?”

width height border format are defined when you upload them, that’s why.
The format you get in the shader is defined in the spec and the spec offers a single function for each texture target (texture1D, texture2D, etc), . Of course, these return a vec4. If someone decides to invent a format that needs a vec5 or more, we will need a new function. :slight_smile:


Do you know where I can get the 3dlabs whitepapers on OpenGL 2.0?

If you can’t find them, I can email to you.

Originally posted by V-man:
If you can’t find them, I can email to you.
I also lost them around my HD… wouldn’t mind sending also to me? (Check personal forum message)

You’re meaning that papers (was 4 years ago?) about primitive programs, render task, timers, unified memory objects and such?

I’ve send you my email address too, thanks!

Originally posted by Korval:
[quote]Many developers may switch because they think OpenGL may not deliver new features at the pace Microsoft can.
This isn’t a recent issue; it’s always been that way. The people who need new features faster have already switched to D3D.
[/QUOTE]I never understood this argument. OpenGL has extensions that IHVs use to implement cutting edge features. RenderToTexture aside, I don’t remember a hardware-feature available in mainstream hardware that OpenGL didn’t support, that Direct3D did. I am aware though, of several features currently available in OpenGL, not available in D3D.

I sent you guys the 7 papers but I’m sure there were others. I remember some papers that had graphics to illustrate soem of the ideas.

I hope the ARB posts something soon, even if it is just to say that they are still there and supporting OGL. This is kind of important since a lot of changes have happened in the market, SGI’s close to going bankrupt, 3d labs biting the dust … so just wanna feel that OpenGL will not have the same fate as 3dfx.
:slight_smile: .

Originally posted by selwakad:
I hope the ARB posts something soon, even if it is just to say that they are still there and supporting OGL. This is kind of important since a lot of changes have happened in the market, SGI’s close to going bankrupt, 3d labs biting the dust … so just wanna feel that OpenGL will not have the same fate as 3dfx.
:slight_smile: .

with all that i hope nvidia takes on the job. just look at the people working for them. i think they could bring gl to a new level.

yes, NVidia can bring OGL to a new level, but as i recall there was a marketing WG in the ARB.
but so far, and maybe i am wrong, I haven’t seen anything that helps in marketing OGL.
So i wonder if they are (the marketing WG) still operational or not.
One thing that they can do is make a Contest where contenders would present demos written in OGL on any platform. and the prize can be something like a new NVidia card … just a thought :slight_smile:

Originally posted by selwakad:
I hope the ARB posts something soon, even if it is just to say that they are still there and supporting OGL.
The ARB is still here and supporting OpenGL. At the meeting last week one topic of discussion was the need to fill the information void which resulted when the publishing of meeting notes was discontinued. Stay tuned.