Thread: Lay me a foundation

I always get so far with my game project then hit a design snag of some sort and it all falls apart. There are too many to list so I won't, I'll just say I've had > 10 false starts. I really suck at designing, which doesn't put me in a very good position if I'm working on my first game.

So I was wondering if anyone here could explain how they've laid out their games. BobMcGee123, I'm particularly interested in you since you mentioned you don't use any OOP: how have you structured everything? How have you managed to get data around and... things. I mean I start working on mine then realise I need such and such data from such and such class then start static'ing things, half of my objects end up being singletons and it quickly becomes too "spilly" to continue working.

The same things seem to happen every time no matter how I try and approach it.

I start pretty much the same way each time though. I write up a quick logging class (a singleton), well now, when I say write up I mean copy and paste from one of the last gazillion false starts. Then I move onto getting a frame. I create skeletons of classes I know I'll need, Player, Kernel, MessageReceiver, MapTile, MapManager, I think that's it. Then I fill them out with "what they need" as I realise what they need.

I don't do much on paper - I can sit there for an age with pen and paper and they won't touch. My mind just goes blank as if I've suddenly unlearned everything I knew (or thought I knew ) about OO design. I (used to) prefer manic hacking over design as everybody does when they start programming (I guess).

It's a bad habit which I'm finding frustratingly difficult to move away from, so I need inspiration. I know it's a little much to ask but even a few links to some articles on design (please nothing from gamedev, I've milked that for all it's worth) would be great.

Is there any shame in starting by using someone else's design as a base? The Enginuity articles over at gamedev are pretty good but I don't like using other peoples ideas like that. Even now as I use his Singleton base class I feel constantly... dirty.

That sounds like good advice. I just can't seem to get my ideas out unless I need them on the spot, while I'm implementing another. I'll try getting into the habit of jotting down ideas as they come to me while I work.

> I really suck at designing
It's an aquired skill - practice small and work your way up from there.

Figure out what sort of program you could write "straight from your head" without doing any kind of up-front design.

Pick a project which you estimate would be about 5 to 10 times bigger than that and start making some design notes. For this sort of scale, you should be able to figure out in a few hours pretty much everything you need to do. Focus on things like significant algorithms, data structures, class relationships, internal interfaces. Think about how you would test each component.
Keep refining the details until you have a list of say 10 to 20 "easy" things to do, and have a go at implementing it.
Depending on how well that goes, use that to calibrate whether you need more or less design for next time around.

Thanks Salem. Is this a method of design you use(d) in your work / hobbies (you work, right?) ?

I should be back in a few hours

Pretty much, except my designs take weeks or months to complete.
Projects with 100K+ lines don't just come out of your head in one sitting at the keyboard.

I'm unsure how to design the collision detection. If I have the Player class, I want it to check whether it's colliding with anything each frame, right? How do I do this without knowing what it'll be colliding with?

I was thinking of having a base class, SolidObject, which other objects can derive from. Something I've never been sure on: If an object inherits from one class, can I use the base class to check for a generic situation, if you see what I mean.

Something like:

Code:

class Player {
public:
    // ...
    bool CheckCollision(SolidObject& with);
    // ...
};

Assuming a class, lets call it Wall, has derived from SolidObject, would CheckCollision still realise the Player is colliding with a Wall by checking on a SolidObject?

Ok while typing this I've just realised, this is all well and good, but what the hell am I supposed to pass as an argument to CheckCollision each frame? I can't check for a collision with everything in the scene each frame... could I? Maybe by keeping a list of all SolidObject's in the scene. Meh, that just seems stupid to me.

How do frustrum's work in 2D? Or 3D for that matter - I haven't got a clue.
Guess I'll wiki it later too.

To ahluka: I totally understand I have done the same thing to man. As far as having the base class thing I basically generalize everything to a sprite then derieve everything on it considering everything you see on the screen is techniquely a sprite. It looks like this

Code:

#ifndef sprite_h
#define sprite_h
#include <vector>
#include "bitmap.h"

//A struct to hold x, y pos for screen location
struct coord
{
    short int x;
    short int y;
};

//A struct for bounding box collision
struct boundingBox
{
    short int top;
    short int bottom;
    short int left;
    short int right;
};

class Sprite
{
    public:
        Sprite();
        Sprite(const short int ani);
        Sprite(const Sprite &copy);
        Sprite &operator=(const Sprite &assign);
        //Return non-modifiable goods :)
        inline const boundingBox &Get_Collision() const;
        inline const coord &Get_Pos() const;
        //Set the X or Y position of the Sprite
        inline void Set_X(const short int x);
        inline void Set_Y(const short int y);
        //Return the current animation of the sprite
        inline const short int Get_CurAni() const;
        //Set the current animation
        inline void Cur_Ani(const short int ani);
        //Is the sprite currently offscreen
        bool isOffscreen();
        //Collision Detection
        bool operator==(Sprite &collide);
        //Draw to specified bitmap
        void operator<<(Bitmap &dest);
        void Draw(Bitmap &dest);
    private:
        //Update the bounding box collision detection
        bool Update_Collision();
        //All the Bitmaps for animations
        std::vector<Bitmap> animations;
        //Maximum animations for the Sprite
        short int max_ani;
        //The current animation of the Sprite
        short int cur_ani;
        //The bounding box collision of the Sprite
        boundingBox sBox;
        //The X, Y pos of the Sprite
        coord pos;
        //Is this an active bitmap
        bool active;

};

//Inline because they are just one liners
inline const boundingBox &Sprite::Get_Collision() const
{
    return sBox;
}
inline const coord &Sprite::Get_Pos() const
{
    return pos;
}
inline const short int Sprite::Get_CurAni() const
{
    return cur_ani;
}
inline void Sprite::Set_X(const short int x)
{
    pos.x = x;
}
inline void Sprite::Set_Y(const short int y)
{
    pos.y = y;
}
inline void Sprite::Cur_Ani(const short int ani)
{
    cur_ani = ani;
}

#endif//sprite_h