Originally an experiment to explore a cooperative thunk system to implement simple behavioral AI...

But to exercise such an AI system, we need a world to set it within...

And I'd been wanting to build a world where AI only sees as much a state as the human does...

W, w, <ArrowUp>

Move Up

A, a, <ArrowLeft>

Move Left

S, s, <ArrowDown>

Move Down

D, d, <ArrowRight>

Move Right

.

Stay

• this demo is currently little more than a redeployed copy of the offline test suite
• currently the only two thunk behaviors developed are the player proxy and a random wanderer...
• ...read that again: an interesting property that fell out of this approach is that "the player is just another minded entity, whose mind thinks about key input events, and processes view data into an output element"; this means that transference of play control should be as simple as rebinding input and swapping behavior
• a component-entity system that has a degree of ocap-inspired hardening so that no entity has access to all/raw data
• in fact each entity never even knows its own id, let alone can it truly identify other entities; instead it only observes the world through incoming event data, including view-field sensing
• side modules generated:
  • a more spartan domkit of general browser quality of life factored out of cdom
  • a shadow-casting and spatial index package reprised out of cdom
  • a bigint port of xorshiro256+ to provide each entity with an independent RNG
  • this demo itself has a cute spinner/text-bouncer that could be liberated... maybe you already noticed?