i did some pathing tests in the current spring release (0.82.3.0) and the test version (0.82.3-152). Used BA 7.1. Spawned mixes of peewees, cans, sumos and heavy tanks and moved them around.
imo 0.82.3-152 fixed the main problem with 0.82.3 : if you send a tightly packed bunch of 20 peewees to a location, they remain tightly packed and move efficiently instead of spreading out and wandering around in circles.
spring_test (0.82.3-152 , Ba 7.1)
--- 2 groups of units moving in opposite directions through each other :
- sometimes get stuck in a "vibrating" mode, colliding with each other. Imo what should happen is, as repeated collision is detected, the slowest units should keep trying the trajectory and the fastest should move aside to where there is free space and only then resume course.
--- big group of units moving into position (converging to a point):
- mostly ok, but sometimes a fastmoving unit constantly bumps into slow moving unit moving in the same direction, instead of going around.
spring (0.82.3, ba 7.1)
--- 2 groups of units moving in opposite directions through each other :
- they avoid each other, but move away from the direction they should be moving a bit too much, sometimes in the opposite direction they should be moving!
--- big group of units moving into position (converging to a point):
- they avoid each other, but turn away from the optimal direction a bit too much, some times in the opposite direction!
- sometimes a fastmoving unit constantly bumping into slow moving unit moving in the same direction, instead of going around
common bugs:
- sometimes it seems units start showing move animations while stopped, when i click shift on them, they don't show any move destination, as if they stopped but the cob script's StartMoving() was called but StopMoving() wasn't. I think it is related to aiming. Selecting them and pressing stop doesn't stop the move animation.
spawned an enemy bulldog in the middle of 3 cans, when they killed the bulldog, they all started moving in place. Did that again, two of them moved. Spawned a reaper in the middle of 10 peewees, one started to "move" in place after the reaper was killed.
some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
Moderator: Moderators
Re: some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
I Support what he says
Pathfinding of huge unit composition is worse in the new version,
units try to avoid each other instead of actually moving.
Ordering to line up units clank up more.
Pathfinding of huge unit composition is worse in the new version,
units try to avoid each other instead of actually moving.
Ordering to line up units clank up more.

-
- Moderator
- Posts: 2464
- Joined: 12 Oct 2007, 09:24
Re: some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
Sounds good. Did you try toggling heatmaps as and turnInPlace?
Re: some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
is that toggleable?
(i don't tamper with the game engine, just play it and do mod stuff)
(i don't tamper with the game engine, just play it and do mod stuff)
-
- Moderator
- Posts: 2464
- Joined: 12 Oct 2007, 09:24
Re: some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
Yes both are toggleable. Heatmaps can be disabled in movedata and turnInPlace is a unit attribute.
Re: some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
Is it toggleable client-side? So I can toggle it off on any host?
-
- Moderator
- Posts: 2464
- Joined: 12 Oct 2007, 09:24
Re: some tests, mostly pathfinding (0.82.3 vs 0.82.3-152)
TurnInPlace is potentially toggleable client side, currently no games implement a toggle.Hobo Joe wrote:Is it toggleable client-side? So I can toggle it off on any host?
I am pretty sure heatmapping can be toggled client side by switching around which movetype is used by units. I am not sure if the movedata can be changed directly. Again noone has implemented this.