I'm wondering would it be considered acceptable:

The window class is responsible for drawing data, so it will have a method:

Window::draw(const Sprite&);
Window::draw(const Rect&);
Window::draw(const Triangle&);
Window::draw(const Circle&);

and all those primitives + sprites would be just public struct. For example Sprite:

struct Sprite {
    float x, y; // center
    float origin_x, origin_y;
    float width, height;
    float rotation;
    float scaling;
    GLuint texture;
    Sprite(float w, float h);
    Sprite(float w, float h, float a, float b);
    void useTexture(std::string file);
    void setOrigin(float a, float b);
    void move(float a, float b); // relative move
    void moveTo(float a, float b); // absolute move
    void rotate(float a); // relative rotation
    void rotateTo(float a); // absolute rotation
    void rotationReset();
    void scale(float a); // relative scaling
    void scaleTo(float a); // absolute scaling
    void scaleReset();
};

So instead of having each primitive to call their draw() function, which is a little bit off topic for their object, I let the Window class handle all the OpenGL stuff and manipulate them as simple objects that will be drawn later on.

Is this pattern used? Does it have any cons against it's primitives-draw-themself pattern? Are there any other related patterns?