i agree that apps have to keep a moderated adoption rate at least until mesa do some catch up but khronos cant just slow down cuz that will efectively kill OpenGL.
understand that most of those 3D devs want an api able to squueze the last bits of feature that the highend cards can provide and for that they need an api that follow that rythm.
even if they would never use those new feature or even if is a software that not necessarily will run on that hardware in a near future.
everything here is as simple as DX(put your version here) have X,Y,Z function and hey OpenGL not = OpenGL is obsolete so it does not worth it or is not as future-proof enough as DX == OpenGL dead(well linux and mac will always keep it but you get the idea).
now i wanna clear something here as someone was mad cuz if something require X GL version cant run on X-y GL version, Dude OpenGL is not DirectX, in DirectX you have to recode every release(or at best partially) cuz DX is some sort of weird library paste and copy that get rewrited every now and then.
remember opengl is extension based aka GL dont have releases, it has revisions, aka the amount X of new extensions is packed as an incremental upgrade named for conveniece OpenGL X.y.
aka in most cases depending on the vendor libgl implementation i can make a software in GL2.1 and it will run perfectly fine on GL4.1 hardware and software or backwards just been careful of making those extension optional so i can adjust the visual quality for every generation (example errrmm unigine demos).
so yes tessalation in opengl is not a core function that need a massive rewrite, is only an optional extension that can be used if the hardware is present and you have the rigth revision that contains that specific function in you libgl library and that apply to almost anything in opengl with the exception of pbuffers wich should be avoided as possible cuz those are slooooow.
the only thing you may wanna chekout every "release" of opengl is the GsGL changes cuz some of those could make your life easier and your software more eficient but thats much it
so it matters if GL 5. 6. 7 get public next week? NO, should mesa bother in reach opengl X? NO for current generation of hardware, would be nice though for those who invested in the latest GPU cuz those new extension are thinked to optimize an specific workload in that new hardware, opengl 2.1 is enough for now? pretty much yes, i would love 3.4 cuz my hardware have those hardware bit present so i can get a bit more perfoamnce or visual quality but yes is enough for now. should i as dev think in use something more thatn GL 2.1? yes, but be careful when you use the extensions, with good dev skills and some care you app can scale from dx9/2.1 like hardware to dx11/4.1 hardware without any major issue(aka ati crappy gl lib)
openGL 1.x dont have shader cuz that hardware didint have shaders
openGL 2.0 have shader support but is very early, ok you can write a shader for 2.0 2.1 3.x 4.x so you can support dx8 class hardware too but not sure if that worth it these days.
theorically (especially since gallium) you can use gl 2.1 in dx8 or prior class hardware but shader have to fallback to a cpu code (gallium llvm) so dont expect those card to scream FPS but it can give a bit more life to that older hardware(already implemente in r300 gallium driver for laptop x1xxx cards that dont have shaders at all in hardware)
in fact is theorically possible to use shaders 4.1 in that hardware too (when gallium llvm get 4.1 support ofc), in fact you can do the same for opengl too (the driver just fall back to llvm in those extension not present in the hardware), so yes opengl is that flexible
and write an ray-tracing game engine based on OpenCL!
and use CPU/GPU for the exact same openCL-code!
thats kill this fucking DirectX/openGL shit ;-)
Blender Goes that way nice 64core ray-tracing engine ;-)
well nobody will ever make a game engine in opencl to begin with and opencl is not meant for that job either(for now, is too much work to be realistic), now opencl can be used for raytracing an many other things where is very efficient like you said.
so you cant get ride of opengl that easily, i wont bother is directx cease to exist though
as opengl is very efficient in certain jobs opencl is very strong in another area wich not necessarilly collide with opengl and backwards, so is not like opencl was meant as a replacement for opengl, it was meant more like a programatic language to take advantage of those massive fpu/parallel abilities of current gpu
It sucks that Mesa will be another version behind, but keep in mind that OpenGL 4.x will only run on DirectX 11 hardware. So it certainly won't be required anytime soon as even most Windows games still support DirectX 9.
Also, keep in mind that mesa and the open source drivers are open source so anyone that's interested in adding support for newer functionality is free to contribute. The the hardware and software specs are available, the only thing missing is more manpower.
ok, if i understand correctly current opengl4 cards (if any) are now obsolete? :>
Why? OpenGL4.1 will run on any card that 4.0 would. You guys are blowing this release out of proportion, it's not as big as a lot of the previous ones. It looks like the biggest change is just an API update to include some OpenGL ES functionality and compatibility. Plus better error handling when given bad data, being able to store shaders after they've been compiled, etc. Isn't that precisely the sort of thing point releases are supposed to be about?