If two entities reference the same component, then when encoded and decoded identity must be preserved

[alecmce]

In a scenario where entity A and B share a component World:

world = new World();
entityA.addComponent(world);
entityB.addComponent(world);

then you encode

encodedA = encoder.encode(entityA);
encodedB = encoder.encode(entityB);

then you decode

entityA = decoder.decode(encodedA);
entityB = decoder.decode(encodedB);

unfortunately

trace(entityA.getComponent(World) == entityB.getComponent(World)) // outputs false

boo!

[hyakugei]

Is there a possibility of using RL/DI "like" tools here? Ie. a map for where World is being used? So, when decoding an encoded entity, you check each of its components for matches to existing rules/maps, and use an existing instance, or create and cache it.

That sounds pretty complex.

[alecmce]

My plan is to create an ID for each entity and component, then to separate encoding into two parts: 1. encode the components under their IDs into a hash and 2. encode the entity-component relationships into a separate hash.

I just haven't had the time to work on XEmber for a while as I've just started a new job and because I am commuting by car. Once my job settles down a bit and I start commuting by train, I'll have more time to invest in this!