RE: [GD-General] Architecture Design

[Chris C. <cca...@io...>]

> It'll be interesting to see where this debate goes when the Americans get on
> board;)

:)

For the last project I worked on, we were pretty hardcore -- UML, CRC cards,
and we were trying to incorporate Refactoring and some of the XP philosophies
into our workflow.  Unfortunately, the company went under before we could 
really see if it paid off.  Take that for what you will. :)

Personally, I'm more of a fan of (1) doing some rough UML-esque diagramming
to see how internal components fit together and how external clients will 
access the system, (2) getting the relevant programmers together to hash 
out the general design and make sure everyone's on the same page, and (3)
get going writing clean code that we can adapt as we better understand the
requirements (or as the requirements change).

More than a couple days of design work on any one "section" of code is
overkill, IMO.  Though large systems can have hierarchies of sections,
so that the physics system as a whole can be a section, the collision
representation can be a section, player control can be a section, etc.

One thing I did want to point out from earlier in the thread is that I 
think that *game* design documents (as opposed to *code* design docs)
are very useful to have fleshed out early in the project.  I agree that
huge sections will be axed by the "reality police" (read: programmers),
but they can be invaluable for doing requirements analysis of game
systems.

-Chris