You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for the roguelike tutorial, it was a great way to get started with rust. I've been playing around with it and noticed the DijkstraMap is not quite working like I would expect.
In the DijkstraMap build function although the start tiles are assigned 0 depth in the open_list queue, the depth only gets written to dm.map when iterating exits. The effect is that a start tile's own value only gets set when iterating an adjacent tile's exits, at which point new_depth is 1. The start tiles then appear to have the same depth as their adjacent tiles. This is a problem for using find_lowest_exit to path all the way to a start tile, as the gradient disappears before actually reaching it.
For example in this image the grayscale tiles are starts and the green tiles show the distance to the gray tiles. Trying to path out of the green and into the gray tiles is unreliable because the gray exits have the same depth as some of the green exits:
Looks like the other build variations have the same issue.
I think when you push the starts into open_list you should also set dm.map[*start] = 0.0; (or the provided depth for the weighted build). Trying that gives the result I'd expect:
The text was updated successfully, but these errors were encountered:
Thanks for the roguelike tutorial, it was a great way to get started with rust. I've been playing around with it and noticed the DijkstraMap is not quite working like I would expect.
In the DijkstraMap build function although the start tiles are assigned 0 depth in the
open_list
queue, the depth only gets written todm.map
when iterating exits. The effect is that a start tile's own value only gets set when iterating an adjacent tile's exits, at which pointnew_depth
is 1. The start tiles then appear to have the same depth as their adjacent tiles. This is a problem for usingfind_lowest_exit
to path all the way to a start tile, as the gradient disappears before actually reaching it.For example in this image the grayscale tiles are starts and the green tiles show the distance to the gray tiles. Trying to path out of the green and into the gray tiles is unreliable because the gray exits have the same depth as some of the green exits:
Looks like the other build variations have the same issue.
I think when you push the starts into
open_list
you should also setdm.map[*start] = 0.0;
(or the provided depth for the weighted build). Trying that gives the result I'd expect:The text was updated successfully, but these errors were encountered: