The International Simutrans Forum


Author Topic: Way constraints  (Read 14644 times)

0 Members and 1 Guest are viewing this topic.

Offline Spike

  • *
  • Posts: 1361
  • First Simutrans Developer and Graphics Artist
Re: Way constraints
« Reply #35 on: December 18, 2008, 02:17:26 PM »
An heuristic could find this end "suspicious" but it might as well be an intended design. So diagnosis is possible, basically, but also imperfect.

Also the path finding must try to jump over the gap, and check if the continuing route actually leads to the desired destination. It's doable but quite a bit extra work.

I assume the feature would still be helpful, even if detecting a few false "problems". Prissi and the patchmakers must decide if it's useful enough to be tried.

Offline jamespetts

  • Simutrans-Extended project coordinator
  • Administrator
  • *
  • Posts: 20659
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Way constraints
« Reply #36 on: December 18, 2008, 06:54:36 PM »

I imagine that detecting routing problems caused by signals, signs, electrification and way constraints is orders of magnitude easier than detecting routing problems caused by missing sections of a way, since the former can just run the same pathfinding routine with the constraint checking turned off and comparing the result to that with it turned on, whereas the latter would require enormously complicated hypothetical and heuristic means of determining the route.

A system that told players when their destination wasn't reachable because of a problem with signals, signs, electrification or constraints but gave no reasons when the problem was an entirely missing connexion is vastly better than one that never gives any reasons.