Curious to who is sticking with OpenGL now...

[Brandon J. Van Every]

Personally, I want khronos to make a library so that these kinds of posts would stop.

I'd like that, but I've decided it's not my battle. Reading some other threads, the consequences of monolithic legacy ICDs sunk in. Retaining all that backwards compatibility is expensive. It means that development of newer stuff I want doesn't get done, like an actually standard Khronos blessed wrapper library on Windows. DX gets far more testing because it's got the critical mass now, not OpenGL. So the OpenGL drivers are buggy and give game developers fits. I don't need this. I will do my next project in DX11 and worry about porting when and if it makes money. By then, maybe I'll be more impressed with the OpenGL situation; I am not today.

Go to their website and see if there is a contact.
Or visit http://www.opengl.org/registry/
there are emails of driver developers in each extension. Those guys play a role in the ARB.

Bet they, and you guys, pay more attention if I become rich. As is, I'm just one voice among the many, that they've all heard before. If I become rich, will I care about this anymore? If I make enough money with DX11, will I need to bother with cross-platform anything? Maybe I'll prefer to work on game AIs or game designs or whatnot. Who knows.

Here's what I suggest you guys do instead, the ones that have a reason to care about OpenGL, that want to see it prosper. All you extension wrapper writing guys, sit down in one place, and SOLVE THE EFFING PROBLEM. Not to denigrate anyone's individual work, but you are past a point of software engineering where your divided efforts simply don't matter anymore. Most people can't / won't spend a week looking at the gory details of all your libraries to decide whether you did it right or not. Progress "in the large" requires people to combine resources and standardize. It should be a SDK, it should be unit tested, it should have automated nightly builds, it should be the only entry displayed on the opengl.org web menu. Whatever you do, make the whole thing MIT licensed. Don't GPL the generator code. Believe it or not, commercial entities want to modify the entire wrapper generation process. They don't want to be stuck with just your .h output.

If you guys can't work together or simply don't care to, then why would you expect much more out of Khronos? They don't have a lot of money to pay people to work together, so they have to contend with design-by-committee politics as well. You can spend time with an email campaign trying to convince them, or you can write it up yourselves.

Hmm, this sounds familiar. Did I write something like this around here a few years ago? Maybe it was for something else, with the same political dimensions. Anyways, good luck. I'll take a look at OpenGL again in a year or two.

[glfreak]

I will do my next project in DX11 and worry about porting when and if it makes money.

If I make enough money with DX11, will I need to bother with cross-platform anything?

They don't have a lot of money to pay people to work together, ...

Remember one powerful feature in OpenGL is the extension mechanism. They can come up with an extension that makes money: GL_EXT_make_money

But the question is any clue why on earth there has not been update to the DirectX SDK since the last one? Well I mean it's been unusually long...?

[mhagain]

Remember one powerful feature in OpenGL is the extension mechanism. They can come up with an extension that makes money: GL_EXT_make_money

Thing is though that the extension mechanism is a major part of OpenGL's current problem. It's gotten out of hand and there are far too many different ways of doing the same thing; a problem that just doesn't exist with D3D. You try to write a GL program that must run on any hardware within the last 5 or so years and you find yourself quickly being sucked into a world of multiple codepaths and buggy driver workarounds; that just doesn't happen with D3D.

On my current project (which is OpenGL) I needed to roll back to using GL_ARB_vertex_program and GL_ARB_fragment_program, for example, because I just could not guarantee consistent support and behaviour on all of the target hardware platforms using GLSL (owing to pre-2.0, 2.0 and post-2.0 compatibility issues and crappy drivers). That sucked and it shouldn't have happened.

The alternative was to provide multiple shaders, multiple code paths and a lot of time spent testing and bugfixing on different hardware platforms.

I'd love it if there was a single GLSL implementation that worked consistently everywhere, but there wasn't - not without bumping the hardware requirements beyond what was acceptable. I was stuck with testing for GL_ARB_do_it_this_way, GL_ARB_do_it_that_way, or GL_ARB_do_it_t'other_way and it was in danger of getting to the stage where I would have probably saved time (and gotten a better result) by doing a D3D renderer for Windows and a separate GL renderer for Linux... no way - my time is better spent doing stuff that's actually productive.

But the question is any clue why on earth there has not been update to the DirectX SDK since the last one? Well I mean it's been unusually long...?

Official word is that the DirectX SDK is being rolled into the Windows SDK and will no longer be updated separately. At least that resolves one massive headache from the too-frequent updates of the past (d3dx DLL hell).

[Foobarbazqux]

Bet they, and you guys, pay more attention if I become rich.

Probably not, rich people/companies (like iD and Blizzard) have been pushing to change OpenGL stuff for years and it has largely hasn't affected anything.

[Brandon J. Van Every]

Bet they, and you guys, pay more attention if I become rich.

Probably not, rich people/companies (like iD and Blizzard) have been pushing to change OpenGL stuff for years and it has largely hasn't affected anything.

Yes that is unfortunate. I might educate myself about what they did try to do in the past.

One thing a company with some cash can do, is bankroll the organizational resources for an extension wrapper SDK and clean Windows API, for instance. I'd like to think that the open source community could get that sort of thing done without the injection of cash, just on sheer idealistic discipline, but money has a way of focusing people and causing them to pay more attention to an effort. It's a way of catalyzing a de facto standard.

The thing is, any company who contemplates doing such a thing, asks how it benefits them. What's the business model? Is it just altruism? I think a company could actually benefit if they were shipping an open source 3d engine. The money would have to come from something in addition to that, but a proper wrapper would create notoriety for the company. I'm doubting that selling a 3d engine would be the right business model, as it seems there are far too many of those already.

I've heard one of the CMake developers say that they really don't make money on CMake. Not a profit center for them. But Kitware does have their Visualization Toolkit and though it's open source, I guess they do a lot of consulting for it. So CMake isn't exactly unjustified, as it made their cross-platform build jobs more sane than Autoconf. I don't think CMake is the best possible build system, but it is the most mature genuinely cross-platform alternative to Autoconf now.

I don't really mean to belabor my involvement here, as I thought I had unsubscribed, but I guess not. I just thought I'd point out that "when I'm rich" is not really meant as a tongue in cheek comment. When volunteers won't get things done, money is another option. The question becomes, how does the money make more money, enough to justify the effort. It's always weighed against all the other things a business could be doing, and survival pressure generally has to take precedence.

[V-man]

But the question is any clue why on earth there has not been update to the DirectX SDK since the last one? Well I mean it's been unusually long...?

You are taking the conversation into some weird directions.

The point he is trying to make is that why isn't there an official extension loader from khronos.
He wants to know why there are multiple libraries that seemingly do the same thing.
He says that there should be just 1 library and the developers should concentrate on it.

[thokra]

[..]why isn't there an official extension loader from khronos. He wants to know why there are multiple libraries that seemingly do the same thing. He says that there should be just 1 library and the developers should concentrate on it.

Why aren't there reliable, up-to-date, cross-platform profiling and debugging tools? Why isn't there an official, up-to-date common toolkit for various tasks like text and font rendering, common math operations, polygonal optimization tools? Why isn't there an official, cross-platform GLSL IDE?

One could go on. The point is that no company or other bigger entity seems to have an economic interest in providing something that is comparable to what is available for or as part of DX.

OpenGL developers have been reliant on community work for a long time and it is pointless to ask the Khronos members for a joint development effort to provide us with tools that have been maintained by the community for years. And if something that you specifically need for your project isn't available right now, write your own stuff, polish it and, if you want, publish it for the rest of us to use.

Edit: BTW, how is all this, including my own rant, even remotely tied to the forum category 'Suggestions for the next release of OpenGL'?

[Brandon J. Van Every]

OpenGL developers have been reliant on community work for a long time and it is pointless to ask the Khronos members for a joint development effort to provide us with tools that have been maintained by the community for years. And if something that you specifically need for your project isn't available right now, write your own stuff, polish it and, if you want, publish it for the rest of us to use.

It is not pointless to ask / organize / demand because the engineering caliber of OpenGL is clearly stagnant and many people observe that more needs to be done to make it robust.

Edit: BTW, how is all this, including my own rant, even remotely tied to the forum category 'Suggestions for the next release of OpenGL'?

I responded to the title of the thread. I've made concrete suggestions about OpenGL's future. If you can't see the relevance... well, this is why I'm not going to try to lead some kind of community project myself. Too many people in this community who don't understand what's wrong, who don't feel any particular need to address engineering "in the large."

[thokra]

It is not pointless to ask / organize / demand because the engineering caliber of OpenGL is clearly stagnant and many people observe that more needs to be done to make it robust.

'It'? The spec? The drivers? The dev tools? The apps using GL? And what exactly do you consider stagnating?

Too many people in this community who don't understand what's wrong, who don't feel any particular need to address engineering "in the large."

Yeah, that must be it. It's not like people got other stuff to do than conceive, implement, test and release a generally useful piece of software. I think to most half way committed developers it's painfully obvious what the missing parts are.

[Brandon J. Van Every]

Yeah, that must be it. It's not like people got other stuff to do than conceive, implement, test and release a generally useful piece of software. I think to most half way committed developers it's painfully obvious what the missing parts are.

We all have our business decisions. Mine is to ditch OpenGL in favor of more modernized DX11. An annoying decision as I'm an erstwhile OpenGL device driver engineer. I would have liked to see OpenGL rise to the challenge of the competition, but as you say, the community is too stressed to bother. I will now make a greater effort to unsubscribe as I've said all that needs to be said.