Ticket #44355

AI hints metaticket

Open Date: 2022-04-14 08:02 Last Update: 2022-04-14 10:13

5 - Medium
5 - Medium


As more things happen in Lua script – like entering huts, inspiring partisans, as well as user actions and effects – it becomes less possible for AI to understand those things, since it has to either assign them some hard-coded worth appropriate for the old behavior (which doesn't do well with rulesets that change it), or it ignores them completely. This ticket is about alleviating that problem by adding new ruleset features that allow ruleset authors to tell AI how to estimate the utility of these things.

Ticket History (2/2 Histories)

2022-04-14 08:02 Updated by: alienvalkyrie
  • New Ticket "AI hints metaticket" created
2022-04-14 10:13 Updated by: cazfi

Implementation should try to minimize problems related to hints becoming specific to one AI type. We previously have two concepts living in ai/ main-level (and always linked to the server, regardless what AI types get built in). I guess this would be third one, to have at least some component there.

One of the existing concepts is AI traits, which are adequately agnostic to AI type - one implementing an AI type might need to scale actual values to match their exact code, but the high level values try to have a consistent range for what are the sensible values, and the traits should make sense in the context of most AI types.

The other concept are handicaps. They were part of the classic/default AI long before support for different AI types was implemented. They could not remain feature specific to classic/default, as the server expects them from any AI type it uses. They have just been moved from classic/default AI type to generic ai-level. It's not always easy to see how something like H_ASSESS_DANGER_LIMITED would be implemented in a concept of a totally different AI.

Attachment File List

No attachments


You are not logged in. I you are not logged in, your comment will be treated as an anonymous post. » Login