Post by screwtape in Gamedev in Lisp. Part 1: ECS and Metalinguistic Abstraction comments

[] if I understand, ecs:make-object expects per object an alist where the cars are keywords and cdrs are plists keyed by keywords.

So instead of using symbol-name, I would have a :name property in a plist to construct the input to ecs:make-object. And I guess I would need a :package property if I wanted to refer to the password; actually, is there symbol support or sequence support (I understand sequences of indeterminate length aren't nice for fast searches).

I'll actually try the example when I'm at home.

Andrew165 days ago(+1)

ecs:make-object expects per object an alist where the cars are keywords and cdrs are plists keyed by keywords.

Not exactly, it expects a list of component specifiers, each specifier is a list with component name as its car and a plist with component data as its cdr.

You can certainly have a name component.

Sequences indeed do not fit nice into ECS model, you might want an assortment of entities instead (as entities are in fact just indices in component arrays).

screwtape164 days ago(+1)

oh yeah I got it backwards. that name referred to the defined component name, I would have to put a symbol-name property inside the component if I wanted it. Now I got it working I can actually use it and see ;p. And I would call make-object once for every object, the object having components having component properties.

itch.io

Viewing post in Gamedev in Lisp. Part 1: ECS and Metalinguistic Abstraction comments