Hi everybody, while I think I've already found a working solution for the problem, I still don't really understand why it actually does work. I was hoping somebody could clarify the following for me. I was trying this on an ipod touch (OpenGL ES 1.1). simplified code:

void Foo::drawSome()
{
	GLfloat *vertices = new GLfloat[strLength*12];
	GLfloat	*coordinates = new GLfloat[strLength*18];

	// fill with correct verts and coords...

	// draw stuff
	glPushMatrix();
	
	glTranslatef(x, y, 0.0f);
	
	glBindTexture(GL_TEXTURE_2D, m_fontTex.name);
	glVertexPointer(3, GL_FLOAT, 0, vertices);
	glTexCoordPointer(2, GL_FLOAT, 0, coordinates);
	glDrawArrays(GL_TRIANGLES, 0, strLength*6);
	
	glPopMatrix();

	// here it crashes, basically complaining that I'm freeing memory still in use	
	delete [] vertices;
	delete [] coordinates;
}

The app crashes as soon as I reach a certain amount of verts. Obviously the array are still in use when I try to free the memory. The solution is to use static arrays like this:

void Foo::drawSome()
{
	GLfloat vertices[256*18];
	GLfloat coordinates[256*12];

	// fill with correct verts and coords...

	// draw stuff
	glPushMatrix();
	
	glTranslatef(x, y, 0.0f);
	
	glBindTexture(GL_TEXTURE_2D, m_fooTex.name);
	glVertexPointer(3, GL_FLOAT, 0, vertices);
	glTexCoordPointer(2, GL_FLOAT, 0, coordinates);
	glDrawArrays(GL_TRIANGLES, 0, strLength*6);
	
	glPopMatrix();	
}

I don't understand why this works and dynamically allocated arrays don't. Shouldn't the static arrays be gone as soon as the function returns? So what's the real difference? Does the data on the stack live longer? Long enough for it to work? Please explain...

Quote:Original post by V-man
Sounds like a bug but I don't know the details of the GL ES spec. Maybe you should ask on the GL ES forums
http://www.khronos.org/message_boards/viewforum.php?f=11

good advice, thx. if anybody's interested, here's the followup:
http://www.khronos.org/message_boards/viewtopic.php?f=11&t=1312

When you call gl[Vertex|TexCoord|BlahBlah]Pointer, you're telling the GL that your data is located at that address. OpenGL assumes that the data stays valid indefinitely. In this case, this is not true. The second example (local variables) works through luck - the memory for those variables is located on the stack. The stack remains part of your process's memory space for its whole lifetime, and therefore your program won't crash. However, there's no guarantee that you won't overwrite that data in the next function call.
There are several solutions that I can think of:
1) (Easiest) Call glFinish() before freeing the data. This will ensure that the GL finishes using your data before you free it. This will adversely affect the performance of your application.
2) (Medium) Make your buffers members of your Foo class, allocate them and use them, but don't free them until you're done rendering. You still need to make sure that the GL is finished before you overwrite the data, though.
3) (Harder, but worthwhile). Use VBOs. You can call glBufferData to copy data into the object, or glMapBuffer(GL_ARRAY_BUFFER, GL_WRITE_ONLY) to write directly into it. Then, you render from the VBO, not client memory. Assuming your data doesn't change too often, this will provide the best performance.

If your application crashes after implementing one of these solutions, then either you have a problem elsewhere, or there is a bug in the OpenGL-ES implementation.

Thanks Graz. So I was indeed just lucky that it works with static arrays. VBOs sound like the right solution for what I need, I'll look into that.

Quote:Original post by Graz
When you call gl[Vertex|TexCoord|BlahBlah]Pointer, you're telling the GL that your data is located at that address. OpenGL assumes that the data stays valid indefinitely. In this case, this is not true. The second example (local variables) works through luck - the memory for those variables is located on the stack. The stack remains part of your process's memory space for its whole lifetime, and therefore your program won't crash. However, there's no guarantee that you won't overwrite that data in the next function call.
There are several solutions that I can think of:
1) (Easiest) Call glFinish() before freeing the data. This will ensure that the GL finishes using your data before you free it. This will adversely affect the performance of your application.
2) (Medium) Make your buffers members of your Foo class, allocate them and use them, but don't free them until you're done rendering. You still need to make sure that the GL is finished before you overwrite the data, though.
3) (Harder, but worthwhile). Use VBOs. You can call glBufferData to copy data into the object, or glMapBuffer(GL_ARRAY_BUFFER, GL_WRITE_ONLY) to write directly into it. Then, you render from the VBO, not client memory. Assuming your data doesn't change too often, this will provide the best performance.

If your application crashes after implementing one of these solutions, then either you have a problem elsewhere, or there is a bug in the OpenGL-ES implementation.

This is not how OpenGL treats the client side buffer data, and I cannot find anything in the OpenGL ES specification that suggests it works otherwise (although I only checkd quickly, I may very well be wrong). On a call to glDrawArryas, the data is fully processed and the client is free to do whatever with he want the buffer afterwards whenever he chose to as soon as the function returns. If OpenGL does not draw it immediately, it is OpenGL's responsibility to copy the content to internal buffers. The user is not responsible for synchronization. So it is perfectly valid, as far as OpenGL concerns (driver bugs is another issue), to delete the buffers immediately when glDrawArrays returns.

Quote:Original post by Brother Bob
This is not how OpenGL treats the client side buffer data, and I cannot find anything in the OpenGL ES specification that suggests it works otherwise (although I only checkd quickly, I may very well be wrong). On a call to glDrawArryas, the data is fully processed and the client is free to do whatever with he want the buffer afterwards whenever he chose to as soon as the function returns. If OpenGL does not draw it immediately, it is OpenGL's responsibility to copy the content to internal buffers. The user is not responsible for synchronization. So it is perfectly valid, as far as OpenGL concerns (driver bugs is another issue), to delete the buffers immediately when glDrawArrays returns.

Pretty interesting. I just can say that Apple includes some samples with the SDK where they're doing exactly the same thing with static arrays. I know that doesn't necessarily have to mean much, but it looks like it was intended to work that way.

Quote:Original post by Brother Bob
On a call to glDrawArryas, the data is fully processed and the client is free to do whatever with he want the buffer afterwards whenever he chose to as soon as the function returns.

Yep, on closer inspection, you're right. This is a driver bug. My suggestions will suffice as workarounds, but the original code as posted is correct. I've just seen a lot of bugs caused by freeing data that's still considered valid by GL.