Hello,

I am trying to design (and implement) simple software rendering engine, but I want to do it "by the book" - using some methodology (unified process or waterfall for example - this is one-man small project so I think waterfall is ok too). My primary goal is to complete project using chosen methodology.

Engine will have scene-node manager that will serve as interface to 3d application, so that renderer is decoupled from 3d application. So it's some kind of two-tier architecture:

application - scene manager - renderer

I don't want application to know about renderer.

That said, I have questions about requirements phase, that is similar in both aforementioned methodologies:

1) When I write use-case stories who should be an actor? I guess it is application, because it has behavior and it communicates with scene manager (that is system)?

2) If I choose application as actor, and scene manager as system, I will end up with use-case stories such as "Application calls scene manager to set up camera on some position". That way I will have good understanding of scene manager requirements, so I will be able to design it. But I am wondering about renderer itself. These stories will not give me enough information to design a renderer. If I want to get more information about renderer, I would need to chose scene manager as "actor" and renderer as "system".
In other words, I am wondering if it is ok to separate requirements phase in this way? First you chose one actor on higher level, and another actor as you go deeper into the system?


Thanks in advance!

p.s. sorry for my bad English.