Oh, I love the diagrams, it is even clearer and faster to understand that the introduction written by Richard Lord (Ash Framework). Good job yay !

This is concise and easy to understand. A real nice Intro to ECS!

Like this. Nice intro, look forward to seeing some sample implementation

Great article. What did you use to create the diagrams?

I agree, this is a nice intro. Are you planning to follow this up with an article on systems and design challenges? This could easily be made into a small series, especially with how often the topic comes up. (Well, maybe not often relative to "Which language?")

Great article. What did you use to create the diagrams?

I used PlantUML, a text-based UML graphing tool.

I agree, this is a nice intro. Are you planning to follow this up with an article on systems and design challenges? This could easily be made into a small series, especially with how often the topic comes up. (Well, maybe not often relative to "Which language?")

I think so, yes.  There's a lot more to CES than just the theory.

Wonderful. Looking forward to it.

I like how you presented this thing. Many forum posts I've read on component system tend to be extremely elaborated and fail to point out what really counts. It appears to me you have been very good at condensing the core notions.

Not related to your article but to Component Systems in general, how would you deal with gameplay specific behaviours?

Here's an example for the purpose of discussion. 

• A simulated entity A={collision brush, mass, graphical model, movement} hits an object 'B'. In this case there's "typical" response, we have no problem.
• Same as before but A has now got a power-up which allows it to ignore collisions with objects of type B. Or perhaps even worse, the specific object B.

I'm currently using overriden functions for that, it feels a bit too inheritance-based but I don't see how to improve that. I was thinking about some "filter" components but the concept seems to be quite vague. What do you think? Does it even qualifies for being called component-based?

Thank you

Not related to your article but to Component Systems in general, how would you deal with gameplay specific behaviours?

In at least your specific case, I would maybe have a filter component.  Your physics system (or possibly a separate system) can be in charge of registering these "collision groups" with your physics engine.

Lots of gameplay features can be implemented with dynamic components.

I do like this article, but I think it's worth pointing out that not every component-based system has to treat components as structs of data with no behaviour, moving all the behaviour into systems. Some component-based systems will encapsulate the behaviour into the component, such as with Unity's system for example. The benefit of this is that it's easier to mix and match components as they are usually self-contained. The downside is that when components do rely on each other, these dependencies are not always obvious.

I do like this article, but I think it's worth pointing out that not every component-based system has to treat components as structs of data with no behaviour, moving all the behaviour into systems. Some component-based systems will encapsulate the behaviour into the component, such as with Unity's system for example. The benefit of this is that it's easier to mix and match components as they are usually self-contained. The downside is that when components do rely on each other, these dependencies are not always obvious.

I've found, after many years of working with such a component system, that very few components tend to be entirely self-contained. It's very difficult to orthogonalize a lot of behavior in such a way that it can operate in isolation from other behavior, however the goal of a component system is to compartmentalize as much as possible. As a result, the more you try to split apart components the more they need to communicate to function, and the more you merge them together the less they are truly individual "components", rather than globs of somewhat-related data and functionality. I suppose you could say that separating the data and the behavior eases this pressure a bit, and makes it easier to split apart the data while also establishing clear dependencies between it and the behavior.

This article is pretty good at clearly explaining many things. If I have game Entities derive from an interface as they have in common a group of Components and a Name ID, should I also derive my Systems from a common interface?

This article is pretty good at clearly explaining many things. If I have game Entities derive from an interface as they have in common a group of Components and a Name ID, should I also derive my Systems from a common interface?

First of all, game entities shouldn't be deriving from an interface.  Specialization is done through adding components.  Systems should be derived from an interface to handle automatic component registering.

How would systems operate on a collection of components? And what would be a good way to store components? As I undersand from the article components derive from an interface and are stored under entities. This leads to me something like :

class IComponent{};

class Position : public IComponent
{
public:
float x,y;
};

class Velocity: public IComponent
{
public:
float x,y;
};

class Entity
{
public:
std::vector<IComponent*> components;
};
 

Afterwards a Physics system operates on the components of entities to add velocity to the position of these entities but how does that system work and decide if a component is a specific type(dynamic casts?virtual functions in the IComponent?). Would it be better to store components like so :

class Entity
{
public:
std::map<std::string,IComponent*> components;
};
 

So you can give names to components and access them with those names.

I think it could be nice to have a pt2 which goes a bit into implementation details.

I think it could be nice to have a pt2 which goes a bit into implementation details.

I'm currently working on my own implementation myself.  When I figure out the best way to do it, I'll follow up with an article on that.

I guess your implementation collides a bit with the one I followed in this article, which is why I gave Entities a common interface. But I see your point in why you may not do that, because then it leads back into an inheritance pattern of making all sorts of Entity classes.

Then if you want to really know that a given Entity could be a Monster, other than peeking inside its components, is to give it a generic name as a string. Maybe that's the job of an entity management system, it would see what components are stored in it and then determine, "this is a Monster". And similarly, if any components are replaced/removed, the system would check again.

The key thing to understand is that there is no notion of a "Monster".  An entity can behave like a Monster if it has the correct components, but beyond loading in the data and spawning the entity the systems don't care.

I guess your implementation collides a bit with the one I followed in this article, which is why I gave Entities a common interface. But I see your point in why you may not do that, because then it leads back into an inheritance pattern of making all sorts of Entity classes.

 

Then if you want to really know that a given Entity could be a Monster, other than peeking inside its components, is to give it a generic name as a string. Maybe that's the job of an entity management system, it would see what components are stored in it and then determine, "this is a Monster". And similarly, if any components are replaced/removed, the system would check again.

If I understand correctly, if you wanted the entity to be a "Monster", then you'd give that Entity a "Monster" component. That way, the Entity is only defined by the components given. If the last boss is a zombie pirate with hovers. The Entity (I think) would look like: Entity (boss, zombie, pirate, hover). 

Anyone, please feel free to correct any of my misunderstandings.

very helpful article , the usage of diagrams make it sample to understand (I found english not easy to understand and words relatide to game developpment make it harder ) If you can make other articles with diagrams for level conception I will be very grateful,
 

Very helpful article for someone relatively new to this stuff, like me. I've been trying to do an implementation of this in C++, and while I can understand the concept I've been having a fair bit of trouble. One thing that's really been bugging me keeping track of an entity's components. How would you recommend storing an entity's components, and then how would you suggest accessing them from a system?

Then if you want to really know that a given Entity could be a Monster, other than peeking inside its components, is to give it a generic name as a string. Maybe that's the job of an entity management system, it would see what components are stored in it and then determine, "this is a Monster". And similarly, if any components are replaced/removed, the system would check again.

There shouldn't ever be any reason to know whether an entity is a monster. If you have code that does "if this is a monster, then....", that means you're doing it wrong.

Of course, for debugging purposes it can be useful to identify the object somehow. The code that created the entity and its components can tag it with some string identifier in that case (or if you're using some sort of prefab system, then the name of the prefab).

How would you go about implementing a "kill all monsters" type of spell? Would you tag the entity with a monster component? Or, what is the right (/better) way to do such a thing in a component based approach?

<snip>

 

There shouldn't ever be any reason to know whether an entity is a monster. If you have code that does "if this is a monster, then....", that means you're doing it wrong.