glBufferData does not delete existing data store - opengl

Question

The documentation states
glBufferData creates a new data store for the buffer object currently
bound to target. Any pre-existing data store is deleted. The new data
store is created with the specified size in bytes and usage.
So I code
// allocate storage in GPU and copy data
glBufferData(
GL_ARRAY_BUFFER,
myData.VertexCount()*sizeof(GL_FLOAT),
myData.Vertex(),
GL_STREAM_DRAW);
...
glDrawArrays(
GL_TRIANGLES,
0,
myData.VertexCount() );
which is executed every time I need to refresh the scene.
From time to time the number of vertices changes greatly. When the number of vertices drops, the old data appears to be still present and a garbled rendering of the old data appears.
I can workaround this problem by creating the buffer index at the start of each refresh, then deleting it afterwards
// construct buffer index
GLuint vertexbuffer;
glGenBuffers(1, &vertexbuffer);
...
glDeleteBuffers(1,&vertexbuffer);
This is fine for the moment, but I would like to optmize things later, including reusing some of the vertices in the buffer without needing to make a fresh copy on each refresh.
The problem does not show up on all my machines. It looks like the machines with better graphics cards do not have the problem
For reference, here is the entire render code:
void Render()
{
// Background color
glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
glEnable(GL_DEPTH_TEST);
glDepthFunc(GL_LESS);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT );
glUseProgram( myShaderID );
// Send our transformation to the currently bound shader,
// in the "MVP" uniform
glUniformMatrix4fv(myMatrixID, 1, GL_FALSE, &MVP[0][0]);
// enable vertices attribute buffer
glEnableVertexAttribArray(0);
// make current
glBindBuffer(GL_ARRAY_BUFFER, vertexbuffer);
// allocate storage in GPU and copy data
glBufferData(GL_ARRAY_BUFFER, myData.VertexCount()*sizeof(GL_FLOAT),
myData.Vertex(), GL_STREAM_DRAW);
// let shaders access buffer
glVertexAttribPointer(
0, // attribute. No particular reason for 0, but must match the layout in the shader.
3, // size
GL_FLOAT, // type
GL_FALSE, // normalized?
0, // stride
(void*)0 // array buffer offset
);
// 2nd attribute buffer : colors
glEnableVertexAttribArray(1);
glBindBuffer(GL_ARRAY_BUFFER, colorbuffer);
glBufferData(GL_ARRAY_BUFFER, myData.VertexCount()*sizeof(GL_FLOAT),
myData.Color(), GL_STREAM_DRAW);
glVertexAttribPointer(
1, // attribute. No particular reason for 1, but must match the layout in the shader.
3, // size
GL_FLOAT, // type
GL_FALSE, // normalized?
0, // stride
(void*)0 // array buffer offset
);
glDrawArrays(GL_TRIANGLES, 0, myData.VertexCount() );
glDisableVertexAttribArray(0);
glDisableVertexAttribArray(1);
glFlush();
myCanvas->SwapBuffers();
}
And here is the final, fixed production code. VertexCount() now returns actual count of vertices, not just the number of floats in the vertex vector.
/** Update GL display */
void Render()
{
// Background color
glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
glEnable(GL_DEPTH_TEST);
glDepthFunc(GL_LESS);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT );
glUseProgram( myShaderID );
// Send our transformation to the currently bound shader,
// in the "MVP" uniform
glUniformMatrix4fv(
myMatrixID,
1,
GL_FALSE,
myCamera.ModelViewProjection() );
// enable vertices attribute buffer
glEnableVertexAttribArray(0);
// make current
glBindBuffer(GL_ARRAY_BUFFER, myVertexBufferID);
// allocate storage in GPU and copy data
glBufferData(
GL_ARRAY_BUFFER,
3 * myData.VertexCount()*sizeof(GL_FLOAT),
myData.Vertex(),
GL_STREAM_DRAW);
// let shaders access buffer
glVertexAttribPointer(
0, // attribute. No particular reason for 0, but must match the layout in the shader.
3, // size
GL_FLOAT, // type
GL_FALSE, // normalized?
0, // stride
(void*)0 // array buffer offset
);
// 2nd attribute buffer : colors
glEnableVertexAttribArray(1);
glBindBuffer(GL_ARRAY_BUFFER, myColorBufferID);
glBufferData(
GL_ARRAY_BUFFER,
3 * myData.VertexCount()*sizeof(GL_FLOAT),
myData.Color(),
GL_STREAM_DRAW);
glVertexAttribPointer(
1, // attribute. No particular reason for 1, but must match the layout in the shader.
3, // size
GL_FLOAT, // type
GL_FALSE, // normalized?
0, // stride
(void*)0 // array buffer offset
);
glDrawArrays(GL_TRIANGLES, 0, myData.VertexCount() );
glDisableVertexAttribArray(0);
glDisableVertexAttribArray(1);
glFlush();
myCanvas->SwapBuffers();
}

Answer 1

Your code is just not correct. You set up the vertex attrib pointer to read 3 GLfloats per vertex, and use myData.VertexCount() as the number of vertices in the draw call, but you upload only myData.VertexCount()*sizeof(GL_FLOAT) bytes to the VBO. As a result, you will read way past the end of the buffer, and results are just undefined behavior.
.

Answer 2

What is happening is that when you call glBufferData the GL driver gives you a position in its memory which likely is the same as the previous position. You think you can update part of the whole thing with glBufferData, but that's not true. The rest may coincide with old data or may not.
I think you'd better use glBufferSubData instead of glBufferData to update new data (or part of it) without deleting and recreating a data store.
You need deleting and recreating when newSize > oldSize.