Concurrent AI Actions

[jackpoz]

• AIs should have pieces activated/deactivated depending on other AI decisions, maybe with a stack of states with a pop based on flags (es explore > combat > autorez > explore, popping autorez and combat at once), or the popped AI should pop itself if there's nothing to do (chain popping).
• Pushing an AI on the stack enables/disables scheduled actions by flag currently executing and to be scheduled.
• Each AI could choose to allow/disallow to be pushed/popped.
• Scheduled tasks should be cancellable and pausable&resumeable.
  • Maybe passing the scheduled action as parameter to cancel action
  • Scheduled tasks might be handled like AIs with their own state
• The base AI in the stack is empty/idle or with a passive react state.
• Each unit test could be a particular AI. The AI could have an Update() method or use only scheduled actions with greater intervals.
• ScheduleAction() should return an int as action id to be used to cancel the action.
• Most of BotGame methods should also be moved down to AutomatedGame.
• The documentation should be updated about this too.
• Maybe instead of a simple stack of AIs it could be better a priority based LinkedList that allows to insert AIs in the middle, for example a bot exploring areas (ExplorerAI) which is fighting an enemy (CombatAI) and that gets invited to a party (FollowPartyLeaderAI) should finish the fight then follow the leader till it is in a party, then go on exploring/fighting as soon as it leaves party and follow again the leader if it gets invited again.

---

Want to back this issue? Post a bounty on it! We accept bounties via Bountysource.

[jackpoz]

There could be different AI buckets all updated, like WhatToDoAIs and HowToDoAIs and TopPriorityAIs (i.e. combat, staying alive), with callbacks and status updates

[jackpoz]

CombatAI and MovementAI could be properties of AutomatedGame that always exist and called by other AIs (go there, stop, passive combat, aggressive combat)

[jackpoz]

WIP 9c4aeb9