Sign in to follow this  

Got the entity/component/system basics up, where do I go from here?

This topic is 1582 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I haven’t been into game programming for a long time, but I read up on entity/compent/system and thought it was an interesting idea. As an exercise, I’ve set up a basic framework to get this working:

void Game::load(ILoader& loader) {
	scene.spawn(
		PositionCp { {0, 0} },
		VelocityCp { {0, 0} },
		CollisionRectCp { {100, 100} },
		ArrowControlCp  { 150 },
		TextureCp { "sprites", {100, 100} },
		TextureSliceCp { {0, 0}, {100/256.0, 100/256.0} }
	);

	scene.spawn(
		PositionCp { {200, 300} },
		VelocityCp { {-120, -60} },
		TextureCp { "sprites", {100, 100} },
		TextureSliceCp {
			{100/256.0, 0},
			{200/256.0, 100/256.0}
		}
	);

	scene.run(LoadTextureSys, loader);
}

void Game::update(const UpdateContext& context) {
	scene.run(ArrowControlSys, context);
	scene.run(MovementSys, context);
}

void Game::render(IRenderer& renderer) {
	scene.run(RenderQuadSys, renderer);
}

At this point I’m a bit stuck. How do you go on from here to model game behaviour? The classic “If X happens, do Y” doesn’t really seem to map well to this design. I’d be super happy if someone could lay out some basic principles or points me to an article that does so.

 

One thing in particular that I’m wondering about is how specialised you should get with your components and systems. In this code I already have an “arrow control” component and system, but it seems to me like that wouldn’t scale. I read an article on building bomberman on an ECS design, and it seemed contrived for that reason. I’d imagine an engine would supply a set of generic components and systems, letting you basically script the rest on top of that. Is that a common approach? How would that be implemented?

Share this post


Link to post
Share on other sites

Surprised no one has responded to this yet.

 


The classic “If X happens, do Y” doesn’t really seem to map well to this design. I’d be super happy if someone could lay out some basic principles or points me to an article that does so.

 

Well I'd point you to my Bomberman article, but I see you've already read it :-). But I think it shows exactly how this should be done.

 

There's nothing about entity-component-system frameworks that makes “If X happens, do Y” harder. You just might need to think a little more about where you put that logic, instead of dumping it in some global update function that makes assumptions about what entities are which. Do you have a more specific example you could provide?

 

All the game logic should be in the systems' update methods, or in scripts attached to individual entities. I don't see any reason that only "generic" components/systems should exist - I think it's fine to make components and systems for core gameplay logic, especially if that logic operates over large numbers of entities and has performance requirements. Enforcing strict ordering for executing various logic is also easier if that logic exists in a system (as opposed to a "script" attached to an individual entity).

 

In your above example, what is ArrowControlSys? How is it different from MovementSys?

 

Also, why are TextureCp and TextureSliceCp different components? Wouldn't you always need both together?

Share this post


Link to post
Share on other sites
Sign in to follow this