Original Post
How easy is it to develop with existing components? I'm kind of in the process of exploring glueing together / desiging an RTS game, which will be used for scientific research. To keep costs low we selected a number of open source candidate components: OGRE (graphics), OpenAL (sound), torque networking library ( network), python (scripting), ODE (physics). Problem is only that we don't know how easily these components can be glued together and there is no way to find out before actually integrating it. As our requirements might evolve we might end up rewriting large parts and we would have been better of writing our own component anyway. Would it be better to buy componetns that offer more (e.g. integrated graphics/physics) so minimze the number of integrated components? and save implementation costs? are there some useful strategies that we can use? What kind of integration problems can we expect and can we come up with an architecture design that will minimize the coupling between componens giving us some more runtime flexibility?
