glVertexAttribP4uiv sets the constant current-vertex attribute. Since you passed "0" it should immediately dereference NULL and crash, not set all black.
glVertexAttribP4ui (no 'v') on the other...

You've specified a stride between vertices of 8*sizeof(float) = 32 bytes. But your vertex data is tightly packed and actually has a stride of 2*sizeof(float) = 8 bytes.

You've also drawn a QUAD...

bugreporter.apple.com

...where you can attach your system profile, so there is something more concrete to reproduce with than "ATI", which is any of a dozen+ different GPUs.

Upgrade to 10.6.8, and the GMA 950 will support GL 1.4.

This depends on your context version, profile, and possibly DEPTH_TEXTURE_MODE.

In Compatibility profile, or in an ancient GL2 context, DEPTH_TEXTURE_MODE defaults to LUMINANCE, so the expected...

GetTexLevelParameter(...TEXTURE_SAMPLES...), as per ARB_texture_multisample or GL3.2.

This works the same way as GetRenderbufferParameter(...RENDERBUFFER_SAMPLES...) from 2005.

Pedantically, yes you can, with EXT_gpu_shader4.

Time to report a bug to Nvidia, then.

Slap a #version 150 at the top of the vertex and fragment shader. "flat in" was not legal syntax in #version 110 ten years ago, so it fails to parse.

Also, at least in Apple's Core Profile, you...

It is more accurate to say that texture() implicitly calculates the LOD for you (based on ratio of texels to pixels.)
texture(...bias) adds a floating point bias to that implicit LOD....

That isn't how queries work.
You use GL_MAX_ enums to ask the driver what the limit is, like this:


GLint max_samples = 0;
glGetIntegerv(GL_MAX_SAMPLES, &max_samples);
NSLog(@"GL_MAX_SAMPLES:...

http://www.khronos.org/registry/gles/extensions/EXT/EXT_debug_label.txt

This is very easy. If the vertex shader reads an attribute that is not enabled (glDisableVertexAttribArray) then the shader reads the attribute from the "current vertex" instead of an array.

And...

The GL spec says:
"These error semantics apply only to GL errors, not to system errors such as memory access errors."

If you pass a pointer to the GL (glVertexAttribPointer) and then ask to...

Or call glWindowPos, to avoid exactly this problem.

See the documentation:

Working With Rendering Contexts
...

If you do not set GL_GEOMETRY_VERTICES_OUT, then the default is zero and you can't emit anything. Link should have failed, and the link log should tell you this.

You should also call...

Read the "Implicit Conversions" section of the GLSL spec and it should be clear why variant 2 and 3 don't work -- a uint will never be less than zero.

You've uploaded indices as uints, but are drawing them as ushorts. So this sequence:
0x00000001, 0x00000002, 0x00000003
is interpreted as
0x0000, 0x0001, 0x0000
which is a degenerate triangle,...

The spec is wrong. Get Khronos membership and go read Bugzilla #9517.

Depth is always clamped in core OpenGL (but not in Nvidia extensions.)

No, this is a spec bug. Despite removal of GLclamp, all depth values are always clamped to [0,1].

The restrictions on CompressedTexSubImage depend on the format.

The general rules are described in the ARB_texture_compression spec which disallows partial SubImage by default (see Issue 5).
This...

Search for "ARRAY_BUFFER_BINDING" in the specification and you'll see this explained in the "Vertex Arrays in Buffer Objects" section.

And in the "Vertex Array Object State" table, showing exactly...

The new 4.3 spec has an "Objects and the Object Model" chapter, but still describes the overall data flow through the pipeline, and leaves you to reference the State Tables appendix to figure out...

SAMPLE_BUFFERS and SAMPLES are framebuffer-dependent state, so can certainly be zero.
However, 3.0 and later require MAX_SAMPLES to be at least 4.

Any renderer claiming GL_VERSION = 3.3 needs to...