I don't believe that Cannon Jerk was not actually targeting those monsters; this looks like the same bug that was shown to me recently with monsters like the Chaingun Guy and Arachnotron. I'm going to have to see if there is some kind of Actor variable I can clear to stop these "rapid-firing" types of monsters from doing this.namsan wrote:I found a bug.
I was playing The Adventure of Square with your latest(1.1.1) UEA mod in GZDOOM(3.6.0), and the Cannon Jerks attack other enemies for some reason.
Without this mod, Cannon jerks never target and attack other enemies, so this bug is probably caused by this mod.
I took a video of this bug.
Here's what I believe is technically happening in that video: The Cannon Jerk is just in range of another monster that is seeing you and broadcasting your location to all nearby monsters and creating a group search path for them to come find you. All monsters informed in such a way are given the first node in their search path at the same location as the monster that broadcast your location. So, when the monsters below the bridge area informed the Cannon Jerk of your location, he turned to "investigate" starting at their location.
The problem I need to fix here is really a few problems:
1) - The Cannon Jerk can see you and should have been able to see its target recently enough to ignore these sorts of broadcasts.
2) - The Cannon Jerk (and continuously firing monsters like the Arachnotron) will stay in their firing loops when they are given a search path, so they fire at the path node instead of trying to walk to it. This essentially freezes them in their firing states indefinitely.
I'm going to investigate and see if there is some kind of counter I can clear on these Actors when they have a path goal so that they will stop firing and go back to normal functioning.
EDIT: I'm about to release a potential fix for this problem. So far, I can't replicate the issue since making this one simple change, but it'd be better if I can get more players that have been encountering this sort of issue to test it.
EDIT2: The latest master on the Universal Enhanced AI GitHub has a potential fix implemented. If no one can replicate this issue for a several days or so, I'll make it an official release.