What I''m doing for my game now is working on the editor/tool for the game; when the game is ready to release, I can simply remove a lot of the functionality, so that the program executes like a game rather than an editor.

Mind you, the game is different than yours (a side-scroller JumpMan Jr-style game), so maybe this option is not a good idea for your project.



MatrixCubed
http://MatrixCubed.cjb.net

My current engine is fully 3D, but object based more than level based (like The Sims for example, with lots of dynamic world objects that can be placed anywhere, but in a fully 3D world).

Since I can''t do much in the way of preprocessing world geometry in this case (everything is too dynamic), I just made the editor part of the engine and it can be enabled with a command line switch or key combination (for any game release I will likely disable this just to avoid having people be tempted to jump into the editor and cheat while in-game).

Since the game I''m writing on top of this engine is UI heavy anyway, I have a full UI layer that sits on top of a D3D display which is really useful for the editing functionality.

The point of telling you all this is just to confirm what others have said..it depends on what you are doing. It also depends upon your personality to some degree too. Some people are very visually oriented and are better off working on editor & engine simultanously so they can see in-game results as soon as possible. Other people can work more abstractly for a longer period of time without being bothered by it. There''s no one-size-fits-all right answer.

The first game I made was 6 years ago and that had no concept of game or engine or whatever. It was just a lot of code that made up the game. If you''ve never made a game before and already you''re trying to abstract it then you''re trying to run before you can walk I''m afraid.

But to answer your question in terms of a starting solution - I first made the character control code (just got a platform on screen with my character bouncing around it), then I made the level editor so he could jump around some stuff, then I finished the game and refined the collision detection.

It was at that stage I knew what was involved in making a game and could hence abstract in the right places, rather than what I *thought* were the right places.