So, this is my very first post on GameDev. I've already tried using the search engine, and even that I'm sure that this question was probably answered many times, couldn't get rid of my own specific doubt.

I've been into programming for quite some years, and recently I decide to jump into game development. I'm doing great so far, and have an almost made (simple, but working) game with C++ using SDL. Although everything is working just fine, I'm pretty sure some things can be improved.

Okay, so far I've been working with a main loop structure like that (I'll post just pseudo-code to make it easier):

Game::Execute()
{
    if(!InitializeEverything())
        return -1;

    while(Running)
    {
        Events();  //catch and process any inputs by the player
        Update();  //update any data related to the game and from the inputs
        Render();  //render all stuff based on what happened in the previous functions
    }

    CleanEverything();
}

According to what why have studied so far, this is a pretty common and efficient way to have your game working.

So, my problems started when things got too big. Let's say my game has three states: the start menu (with the options where the player can choose what to do next), the main game screen (where the game actually happens) and a game over screen (with a high score board or whatever).

The first option I thought (and the only one, so far) was creating a variable on the main game class to set what the current state of the game, so the event catcher, the updater and render stuff could act properly according to it. So I have:

int GameState; // holds the current game state

enum{
    GAMESTATE_START_MENU = 0,
    GAMESTATE_MAIN_GAME,
    GAMESTATE_GAME_OVER
};

Therefore, I could choose how to behave according to that state. So, my Render function, for example, would be like that:

Game::Render(){

    switch(GameState){

        case GAMESTATE_START_MENU:
            //Render stuff related to the main menu screen
            break;

        case GAMESTATE_MAIN_GAME:
            //Render stuff related to the game screen
            break;

        case GAMESTATE_GAME_OVER:
            //Render stuff related to the game over screen
            break;
            
    }

}

The same goes for the other functions. On Event catcher, I would check the game state too see if the clicking or button pressing by the user was related to an option chosen on the start menu or a an action on the game, etc.

This seems pretty good with this small game, but I wonder how it would work, for example, with a game with 10 states or more? I'm sure there should be a better way to do that, specially considering I have OOP by my side. I hope I'm not missing anything or making a totally dumb question. How would be a better way do handle that?

Ooh, this is the second time I get to link to this article today :)

State Machines by Lazy Foo You should find it interesting.

I am using for every game state a class, so you have seperated files, and only a switch when changing game state.

greetings

The way I always did it was using the StateChangeEvent only to change the state of the individual modules.
The components of the engine modules were reset and reinitialized for the new state.

Different listeners were registered, the entity manager was reset / new entities were registered and new resources loaded ... and so on.
Then the rendering logic, the entity manager and resource management etc. didn't care what the current gamestate was any more.
The loop just called the same functionality as in the previous gamestate.

That might be too big a step, though. Simply moving the game loop functionality to gamestates might be a better incremental step.

Ooh, this is the second time I get to link to this article today

State Machines by Lazy Foo You should find it interesting.

That's exactly what I was looking for!

The way I always did it was using the StateChangeEvent only to change the state of the individual modules.
The components of the engine modules were reset and reinitialized for the new state.

Different listeners were registered, the entity manager was reset / new entities were registered and new resources loaded ... and so on.
Then the rendering logic, the entity manager and resource management etc. didn't care what the current gamestate was any more.
The loop just called the same functionality as in the previous gamestate.

That might be too big a step, though. Simply moving the game loop functionality to gamestates might be a better incremental step.

It's an interesting idea. But indeed it's too much for now, specially considering the simplicity of the current games I'm making.

But I'll take a further look at that when things start to get big.

Thanks for your help guys!

Alternative reading: General Game/Engine Structure
Passing Data Between Game States


Additionally, inputs are not events. Don’t handle them as such. Your game loop is nothing but Update() and Draw() (with Update() actually performing an update 0 or more times each loop based off how much time has passed).

L. Spiro

Alternative reading: General Game/Engine Structure
Passing Data Between Game States


Additionally, inputs are not events. Don’t handle them as such. Your game loop is nothing but Update() and Draw() (with Update() actually performing an update 0 or more times each loop based off how much time has passed).

L. Spiro

Oh, I didn't know about that. Looks like, since I really come from event-based systems, I was mistaking the concepts.

Thanks for the advice and articles! They are my next reading now!

Generally speaking, every time you have a messy switch (or if/else if/else if...) you can refactor in some way using classes.

The way I always did it was using the StateChangeEvent only to change the state of the individual modules.
The components of the engine modules were reset and reinitialized for the new state.

Different listeners were registered, the entity manager was reset / new entities were registered and new resources loaded ... and so on.
Then the rendering logic, the entity manager and resource management etc. didn't care what the current gamestate was any more.
The loop just called the same functionality as in the previous gamestate.

That might be too big a step, though. Simply moving the game loop functionality to gamestates might be a better incremental step.

So you grew out of game states ?, interesting, got me confused.