Archived

This topic is now archived and is closed to further replies.

the_grip

Coding Structure...

Recommended Posts

Question... i know the whole game loop thingie, so my structure question is not on that. i''m just curious how people typically code their games. Like classes, structures, etc... i''m thinking of overhauling a breakout game of mine for fun and for future use (making a reusable pseudo-template of code that can be reused). "You call him Dr. Grip, doll!"

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
fundamentally classes are types, just like int and double. Sure they have some differences but the main thing to understand is that they are types. Ever want to make a new type? Now you can. For example I have an angle type. It has only one variable: double value; I gave it all the operators but they make sense for angles. So no matter what you add to an angle you''ll end up with a value between 0 and 2pi. It works just like a real angle. For simplicity I have a constructor that takes a double and a conversion operator that converts to a double. So they can be used in any function that takes a double and they can accept a value from any function that returns one. Is it slower? I doubt it, later I''m going to inline everything and that should get rid of all the overhead. What I get is a nice new value type.

Later you''ll want to make more complex things, like monsters. Treat these guys a little differently, for one thing you''ll want to pass them by pointer. Start thinking in terms of behavior and/or data, depending on the purpose of the class.

Well I''m short on time but here is an important piece of advice: avoid inheritance. Inheritance is great when you need it, but most OO programmers overuse it greatly, because they were taught that way in school.

Oh and don''t focus on reuse. A much beter strategy is to just make the game and then extract the reusable parts from it. Thus reuse evolves naturally and eventually you will be doing a lot of reuse and it won''t have taken you much effort.

Share this post


Link to post
Share on other sites
Well that''s one school of thought, but it pisses me off to no end when I see broken code that I know an OOD would have prevented. For instance, if Quake(I/II/III) had used an interface to define it''s networking protocols, you could play on a 1.17 server with a 1.29 client, because it would be a *trivial* process to use the old networking code.

If you''re interested in designing reusable objects, check out "Design Patterns" by Erich Gamma & etc. al.

There is no object oriented design until there is a virtual function and inheritence, consequentially polymorphism. Many modern scene graphs are implemented with polymorhphic trees of various types of nodes.

You can check out these engines for examples of OO game engines:
www.auran.com look for Jet
www.RadonLabs.de look for The Nebula Device


Magmai Kai Holmlor
- Not For Rent

Share this post


Link to post
Share on other sites
"
There is no object oriented design until there is a virtual function and inheritence, consequentially polymorphism.
"

Not sure I would agree with that

I have designed several small OO applications that did not require the use of inheritance, yet they were still quite object-oriented. In some cases the scope of the project makes inheritance an unnecessary burden.

I think a better way of putting it is that there is no OOD without encapsulation

There are also cases when inheritance should be used exclusive of virtual functions. Sometimes you only want to inherit the data

But in the scope of games inheritance/polymorphism is generally a good idea I think.


Seeya
Krippy

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
I''d say that polymorphism is the key to object oriented programming. Inheritance is used to support it. The problem is that sometimes people use inheritance even when there is no compelling reason for it. Design Patterns agrees with me, somewhere in chapter 1 it says to cut back on inheritance and use variables instead.

Also you have to consider the person asking the question. Reuse is great but if you don''t have experience with OOP you don''t know how to go about it. I think the correct way to learn this stuff is the way I did it: trial and error, including starting over again and again. Then picking up Design Patterns and realizing that you learned half the patterns through trial and error, it will make it a lot easier to understand. Now I do reuse code but the best way to get started is to not focus on reuse. In the C++ programming language Stroustrup says that instead of setting out to make a template you should first make it for the case you are using it in and only after you get it working should you generalize it. Basically there is no point to making reusable code your first time through. As for Quake yeah they should be writing reusable code but newbies should start with the simple goal of getting something to work.

Share this post


Link to post
Share on other sites
Interesting question

My current game engine uses classes with COM-style interfacing... using function pointers (similar to V-tables I guess) I can switch between graphics and sound and network code very easily. This isn''t a huge improvement over conventional class-based code, because I''m still writing code in a module that is very encapsulated and controlled by external-ish class methods.

On another node, my game objects will have combined structure and string-variable data, that will allow in-game objects to dynamically add and remove properties... this is done mainly for a planned scripting system in the future.




MatrixCubed
http://MatrixCubed.org






Share this post


Link to post
Share on other sites
True, but be warned, good sir - these are all peripheralia. Sound/graphics/input has been worked over so extensively that simply using COM for them does not squeeze the full benefit out.

I would advise you to direct your sights more towards what you aim to do. Do you have designers working for you? Do you intend to reuse the code directly (with a clear focus)?

More importantly, what do you seek to achieve with the current project?

Share this post


Link to post
Share on other sites