[http://sourceforge.net/p/ufoai/bugs/2161 Item 2161] imported from sourceforge.net tracker on 2013-01-28 19:22:20

revision 25946

04 - soldier can't walk over these planks, falls through them and then does a 90-degree turn at an empty space;
05 - soldier avoids walking on the nearest plank

for the record, sourceforge coders should be kicked. "Uploaded file must be no larger than 256k." my ass. it was 254 kb.
===== Comments Ported from Sourceforge =====

====== richlv (2009-08-30 20:10:35) ======

04
====== richlv (2009-08-30 20:13:05) ======

05
====== aduke1 (2009-09-05 20:59:16) ======

At least I can confirm both issues for R6035 ie. with most recent routing code.
The routing data looks fine, so I expect a true pathfinding problem.
====== aduke1 (2009-09-06 20:50:07) ======

Some debugging showed that routing calculates the floor of 135/124/2 as -16, and the ceiling of 135/124/1 as 32 !
In other words, it simply doesn't 'see' that plank.
Is there anything special about that plank in comparison to those other planks that you *can* walk over ??
====== richlv (2009-09-09 09:20:30) ======

looking at the planks in radiant, i can't find any difference between ones that can be walked on and ones that actors fall through
====== aduke1 (2009-10-17 16:31:30) ======

Found it :) In Grid_RecalcBoxRouting() we adjust the bounding box of func-breakable etc. to the need of pathfinding, namely actorsize.
This adjustement was a little off.
====== aduke1 (2009-10-17 16:32:45) ======

oh, it was R26658.
====== tlh2000 (2009-10-17 20:27:23) ======

nice - another one less on the pathfinding front. good job duke
====== richlv (2009-10-22 20:13:33) ======

confirming the fix in trunk rev 26712