A glTexCoordPointer question

Since you can''t change a texutre inside a vertex array, you have to draw a quad, bind a new texture, and draw another quad, and bind another texture... and so on.

So wait a minute, I have to render lightmaps on all of the quads. Wouldn''t that completely defeat the purpose of using glDrawArrays?

Not if you pack the lightmap for several triangles into a single lightmap. Otherwise, yes, sort of.

So I''ve reorganized my render function to do this:
for( int a = 0; a < Map->GetNumVerts(); a += 4 )
glDrawArrays( GL_QUADS, a, 4 );

instead of this:
glDrawArrays( GL_QUADS, 0, Map->GetNumVerts() );

It works, and I see no immediate performance hit, so is this how most people do it when they''re going for best performance?

As a side note here is the whole render function so far:
Is there anything that you see horribly wrong with it?
bool CEngine::RenderScene()
{
TRACE_ENTER_FN( CEngine::Render )

OpenGL->BeginScene();
Camera->Render();


glEnableClientState( GL_VERTEX_ARRAY );
glEnableClientState( GL_COLOR_ARRAY );
glEnableClientState( GL_TEXTURE_COORD_ARRAY );

glVertexPointer( 3, GL_FLOAT, 0, Map->GetVerts() );
glColorPointer( 3, GL_FLOAT, 0, Map->GetColors() );
glTexCoordPointer( 2, GL_FLOAT, 0, Map->GetTexCoords() );

for( int a = 0; a < Map->GetNumVerts(); a += 4 )
glDrawArrays( GL_QUADS, a, 4 );

OpenGL->EndScene();

TRACE_LEAVE_FN()

return true;
}

That doesn''t make any sense. Calling glDrawArrays(GL_QUADS, a, 4); n times is exactly the same as calling glDrawArrays(GL_QUADS, a, 4*n); since you don''t perform any GL state changes between the calls to glDrawArrays.
Obviously, using a single call will be faster.

Apparently, you aren''t using different textures on the quads after all, you''re just using different parts of the same texture.

cu,
Prefect