GLSL programming...

drhalftone

I have a QGLWidget with a GLSL program that uses the RGBA values in a RGBA32F texture as a vertex buffer, reading the vertices using a texel fetch in the vertex shader. For the most part, everything works. The issue I am having is in the paintGL function. Here is what I think I should be using based on the Qt Cube example:
@
// BIND OUR VERTEX BUFFER SO THAT THE GRAPHICS CARD CALLS
// THE VERTEX SHADER FOR EACH AND EVERY POINT IN THE SCAN
glBindBuffer(GL_ARRAY_BUFFER, vboIds[0]);

// Tell OpenGL programmable pipeline how to locate vertex position data
int vertexLocation = program.attributeLocation("qt_Vertex");
program.enableAttributeArray(vertexLocation);
glVertexAttribPointer(vertexLocation, 2, GL_SHORT, GL_FALSE, 2 * sizeof(short), (const void *)(0));

// BIND THE INDEX BUFFER SO THAT THE DRAW ELEMENTS COMMAND KNOWS HOW TO CONNECT VERTICES
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, vboIds[1]);

// BIND THE SCAN'S XYZA COORDINATES AS TEXTURES
glActiveTexture(GL_TEXTURE0);
glBindTexture(GL_TEXTURE_2D, texture);

// PASS THE HANDLE TO OUR FRAME BUFFER OBJECT'S TEXTURE
program.setUniformValue("qt_texture", 0);

// DRAW THE ELEMENTS
glDrawElements(GL_TRIANGLES, numInds, GL_UNSIGNED_INT, 0);

@

However, this is not working. I have to replace the lines:
@
// Tell OpenGL programmable pipeline how to locate vertex position data
int vertexLocation = program.attributeLocation("qt_Vertex");
program.enableAttributeArray(vertexLocation);
glVertexAttribPointer(vertexLocation, 2, GL_SHORT, GL_FALSE, 2 * sizeof(short), (const void )(0));
@
with
@
glEnableClientState(GL_VERTEX_ARRAY);
glVertexPointer(2, GL_SHORT, 2sizeof(short), 0);
@
to get it to work. Does anyone know why this would be the case?

drhalftone

Okay, I found my mistake. In the vertex shader, I have the following:

@#extension GL_EXT_gpu_shader4 : require

uniform sampler2D qt_texture; // THIS TEXTURE HOLDS THE XYZ+TEXTURE COORDINATES
uniform mat4 qt_projection; // THIS MATRIX IS THE PROJECTION MATRIX
uniform int qt_rows; // THIS INTEGER HOLDS THE HEIGHT OF THE TEXTURES
uniform int qt_cols; // THIS INTEGER HOLDS THE WIDTH OF THE TEXTURES
uniform int qt_mode;

attribute vec2 qt_vertex; // THIS VECTOR HOLDS THE K=1 DFT COEFFICIENTS

varying float qt_magnitude; // THIS OUTPUT VALUE HOLDS THE TEXTURE COORDINATE
varying vec3 qt_normal;

void main(void)
{
// FIGURE OUT WHICH VERTEX THIS IS USING THE VERTEX ID
int row = gl_VertexID / qt_cols;
int col = gl_VertexID % qt_cols;
@

But I'm not actually doing anything with qt_vertex. I just declared it thinking it would receive the incoming vertices, but since I'm not using those directly, I just never use the qt_vertex in the code. So the compiler was pretty much deleting that variable since its not being used. So when I try to connect to it my paintGL function, its not making a connection since it deleted from the code. So I was getting a -1 returned for the qt_vertex handle.

My fix was to just include a useless reference to the qt_vertex variable in the main() function to read:
@void main(void)
{
// WE NEED TO ACTUALLY USE THE QT-VERTEX FOR IT TO CONNECT TO THE OPENGL CODE
qt_vertex;

// FIGURE OUT WHICH VERTEX THIS IS USING THE VERTEX ID
int row = gl_VertexID / qt_cols;
int col = gl_VertexID % qt_cols;

@

Now everything works fine.