Ah, that's very interesting. I think I prefer that method if for no other reason than the code makes more sense that way.
I'm not sure I really understood the whole mechanism before, and assuming I get it now...
What your saying is you create your surface, "attach" a rasterizer to it, and your good to go, you can draw on it just like you do when you draw to the DisplayDevice* returned by GetDisplay(). The only difference is the syntax you use (dot notation instead of ->)
Furthermore, and this is the part that's just dawning on me, what your really getting when you do DisplayDevice* display = GetDisplay(); is essentially a rasterizer which has automatically been attached to the screen display (I presume the back buffer surface) by PF itself.
Now this thread is getting kind of theoretical, but it's helped me to understand PF a lot better now, assuming I got the above two paragraphs right.
Thanks a million Thierry, this has been very valuable!