glEdgeFlagPointer vs GL_QUADS

You will probably have to post some of the code, its quite vague and the only thing anyone can suggest is just pure speculation. perhaps post the way you make the array, the way you use glEdgeFlagPointer, some of those things that are highly relevant to the problem.

Quote:

Original post by Xero-X2 You will probably have to post some of the code, its quite vague and the only thing anyone can suggest is just pure speculation. perhaps post the way you make the array, the way you use glEdgeFlagPointer, some of those things that are highly relevant to the problem.

And also what kind of error do you get? Is it an error obtained with glGetError() (if so which one), or a crash (if so did you try a debugger)?

Tom

So here are the critical lines of my code:

This first version works well when I use edgeflag pointer only for triangles.


When i modify the code to use edge flag for quads some problem occurs. glGetError returns with 0 but my cards crashes and switches off the monitor (only reset can help).



All arrays are stored in a Standard Template Library vector structure. I got the pointers by calling vector::begin() member function. I think it's correct because it works in the first version (and also everywhere else in my program). I've tried many versions leaving out different lines and I found out that it crashes when I use glEdgeFlagPointer and glDrawArrays with GL_QUADS together. Maybe my card's (Radeon 9250) GL implementation doesn't support it.

Thx

Quote:

Original post by Beco All arrays are stored in a Standard Template Library vector structure. I got the pointers by calling vector::begin() member function. I think it's correct because it works in the first version (and also everywhere else in my program).

I definately would not do that. I think that trick will only work for some compilers. If you're convinced that this is not the problem, I can't help much other than asking if you installed the latest drivers...

Tom

Quote:

Original post by dimebolt Quote: Original post by Beco All arrays are stored in a Standard Template Library vector structure. I got the pointers by calling vector::begin() member function. I think it's correct because it works in the first version (and also everywhere else in my program). I definately would not do that. I think that trick will only work for some compilers. If you're convinced that this is not the problem, I can't help much other than asking if you installed the latest drivers... Tom

No, don't use vector.begin(), since that will only work if your implementation uses raw pointers for its iterators. Instead use &vector[0], which will always work.

As to your problem, the only thing I can think to suggest is to double check that your quadEdge array is valid and large enough. I don't know much about edge flags, so I can't really help there.

By the way, why are you using multiple contiguous glDrawArray calls instead of just a pair of calls (one with GL_TRIANGLES and one with GL_QUADS) to draw all your objects?

Enigma

Quote:

Original post by Enigma No, don't use vector.begin(), since that will only work if your implementation uses raw pointers for its iterators. Instead use &vector[0], which will always work. As to your problem, the only thing I can think to suggest is to double check that your quadEdge array is valid and large enough. I don't know much about edge flags, so I can't really help there.

OK, I'll change the begin calls.

The size of the edge flag array is exactly the same as the vertex array. I use vertex, normal, color and texcoord arrays too with this method and all of them works except the edge flag with quads. The strangest thing that I do the same as with triangles and it's good but with quads it fails.

Quote:

By the way, why are you using multiple contiguous glDrawArray calls instead of just a pair of calls (one with GL_TRIANGLES and one with GL_QUADS) to draw all your objects? Enigma

The reason is that sometimes I don't want to draw all faces and they have flags signing if they are enabled. These flags are tested inside the "for" period. I posted only the main problematic lines.

Beco

Yesterday I found out the reason of the problem.

I searched for the Windows system error code that was generated after my card's crash. It sad that it is usually a hardver or driver error. The solution was that I pulled down hardver acceleration in display dialog from 100% and it works.

I spent more than a day debugging my code. I think it was a good lesson for me.