Pixel width

Hi! I’m editing a source written with OpenGL ES. I noticed that if I draw a square drawing every pixel with integer coordinates, I get some parts which are not drawn. I tried to draw on integer coordinates and the same coordinate plus 0.5 using GLfloat data type and it seems every part of the square is drawn this way. My question is: why? Maybe this is set when creating the context? Is it possible to disable this and having only to draw on integer coordinates to completely fill a square?
Thanks!