widelands-dev team mailing list archive

[TiborB <tiborb95@xxxxxxxxx>]

so to your comments:

The main problem is that seafaring is based on one-water assumption. There is no scanning 
and analysis of map to identify individual distinct waters and making decision which one is worth seafaring. This would have to include cheating probably.
I have couple of maps (mine usually) that are one-water-only and used these for testing.

Ai builds strictly one shipyard only.

If there are 2-3 portspaces available for AI nearby, it will succed to use at least one for port.
This can be improved but there are limitations - like AI not in control of layout of roads and so on.

As unfinished ship decays the building of ship starts when shipyard is fully stocked - you mentioned this problem probably.

If AI finds a portspace it scans the vicinity in some radius - if it runs into other player - the spot is ignored - this was probably what you encountered. The idea behind is that expedition would not be succesfull anyway.

But this can be finetuned as well.

The result of this is that last expedition lasts forever. This is inefficiency, because a stock on ship and ship itself could be used for transport. But easy to fix/modify.

I used my maps spider lake, cristobals sea and soutfall islands for testing - I recommend to use one of them to see seafaring at its bests :)

-- 
https://code.launchpad.net/~widelands-dev/widelands/seafaring-ai/+merge/242271
Your team Widelands Developers is subscribed to branch lp:~widelands-dev/widelands/seafaring-ai.