View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000492 | FSSCP | gameplay | public | 2005-07-30 14:07 | 2005-11-01 03:05 |
| Reporter | WMCoolmon | Assigned To | Goober5000 | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | closed | Resolution | open | ||
| Summary | 0000492: AI spins in circles when told to attack multiple ships | ||||
| Description | At least that's my guess, because the AI seems to be constantly switching between two targets and spinning in circles because it's unable to decide which to attack. | ||||
| Tags | No tags attached. | ||||
|
|
shouldn't it go to the most recently ordered ship? |
|
|
More information is A-1 SUPAR! |
|
|
We need more information. :p |
|
|
Create a mission, put in eight fighters and assign them to two friendly and two hostile wings. Order each wing to attack the two opposing wings, priority 80 and 50 for example. The fighters just keep spinning around. And the build's fs2_open_G-20050728.exe. edited on: 08-09-05 01:21 |
|
2005-08-10 02:04
|
|
|
|
I just uploaded a test mission set up like Fury described. I see what is meant, that the ships fly in a close circle. I can verify that this happened in the current cvs build for linux, as well as in WMC's 20050329 linux build. The mission was created usind Fred2_Open 20050630. When I have a chance I'll try ths out in vanilla fs2 and see what the results are. |
|
|
Appeared to work fine for me. The wings blew the crap out of each other and enemy fighters properly targetted and fired at me as well. They didn't seem to be switching between various targets either, just picked one and went for it. So this bug is that they don't attack or just how they attack that's at issue? |
|
|
@ taylor: i didn't see target switching either, just that they flew close together. I as attacked as well by one of the ships. I don't know if there is a bug as described, but I'll try and check a few other builds and retail for differences and report back. |
|
|
I saw that they grouped up too but that only appeared to happen when the two ships were attacking each other. Wasn't sure if that was the original behavior or not so I checked retail and it did the same thing. If there was any difference from retail it was that the enemy ships there took a bit more interest in killing me. |
|
|
I went through the AI files a few weeks ago and compared everything to retail, inserting MISSION_FLAG_NEW_AI where appropriate. Except for the turret firing code (which I still have to look at) and the missile homing code (which is in a different file) the AI behavior should be exactly the same as retail in the 0728 build. edited on: 08-10-05 02:09 |
|
|
I can tell you for certain that this behaviour is a bug and only appeared in later builds, however of those later builds I have only tested G-20050728. With build 20050103 AI still worked properly. I noticed this problem by accident, when I was testing new fighters balance in TBP. There were four wings and four fighters in each wing, two friendlies and two hostiles. Distance was about 3km. With build 20050103 all fighters would do the initial spin and engage their targets properly. With G-20050728 they never stopped doing that initial spin, they never moved, just spun around their starting location. This is not a feature, this is a confirmed bug in build G-20050728. I can test several other builds as well as soon as I can get some one shockwave related problem solved. |
|
|
Okay, in that case I may have made a mistake in reverting the AI. It would be helpful to know which build this first appears in. |
|
|
So are we sure that the AI wasn't wrong in the first place and Goober's change didn't just fix it? The AI appears to work as retail did now, I went over Goober's changes as well just in case, so if everyone says that it's broken then it looks more like the behavior you've come to expect was actually wrong. If I add the MISSION_FLAG_AI_NEW to $Flags: (just change 0 to 4096 in the mission attached to the bug) in the mission then it better gives the behavior that you are describing, but it's not retail behavior. I think Goober just unbroke the problem and everyone has just gotten used to it being wrong. |
|
|
taylor, please. You are implying that the AI has been broken since FS1 and should have never followed any orders they were given in wing editor. Fighter AI worked properly in G-20050715, was broken in G-20050722 and G-20050728, worked again in T-20050807. With builds G-20050722 and G-20050728 all fighters do is spin around their starting location using their afterburners and whatnot. |
|
|
No, I'm saying that parts of the fighter AI have been broken with SCP changes. I had retail FS2 and current CVS going and with this mission they work the same. The fighters affectively stay in a little furball and fight it out. If MISSION_FLAG_AI_NEW is set then it has pre-20050722 behavior, behavior that has been broken since the original FS2 code release until Goober made his changes. With this behavior the fighters will break off from each other and re-engange a little later. Basically, just spreading out the fight. This is with current CVS though so I may be testing changes that have been made since 20050728 and those changes would also be in 20050807. I can't test 20050728 so I'm not seeing the exact same problems. What I see works but the AI behavior is slightly different than it was in previous builds (probably since around 3.6). That's why I was curious about what people are seeing. I don't get the same problems reported with 20050722/28 but there is a difference in how it used to be. If 20050807 works then perhaps whatever was broken in 20050722/28 has worked itself out with later commits. Even so there is still going to be the AI difference since we're finally back to the original FS2 fighter behavior, at least what of it Goober has fixed already. There is obviously more than one issue here, so I just want to make sure that what people are referring to as this bug isn't in fact just fixing the fighter AI to get away from the SCP changes and back to the retail behavior. If the problem with this bug isn't clear enough then some might still consider it broken after the basic problem has been fixed but that previously broken behavior that I'm seeing doesn't get restored. [Edit: last sentence added for a little clarity on what I meant.] edited on: 08-10-05 04:08 |
|
|
Can we call this bug fixed, then? |
|
|
I'd call it closed. But first I want to find out the deal with T-20050807. |
|
|
It really needs to be checked again with T-20050823 since it's just plain CVS. I don't remember exactly what I did with 0807 but it was based on 0806 which was for DaBrain and I'm not sure how pure to CVS it was. I had posted T-20050812 as updated CVS with a few more fixes in it for TBP but no one has said how it performed at all. |
|
|
Sorry taylor, but I haven't had any time to do any work on TBP or test new fs2_open build for reported bugs. |
|
|
This needs to be updated or closed. Any time for testing, Fury? |
|
|
* BUMP * I haven't heard any other problems about this since and I assume that Fury isn't going to have a chance the test this again any time soon. I say we just resolve it as fixed after Goober's changes and reopen if it gets reported again. |
|
|
Sorry, but I do no longer play FS2/TBP for various reasons. If someone else had the problem too, hopefully he'll be able to help you. |
|
|
I think we should just close this. Closered. :) |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2005-07-30 14:07 | WMCoolmon | New Issue | |
| 2005-08-02 01:19 | Hippo | Note Added: 0002913 | |
| 2005-08-06 04:26 | Goober5000 | Note Added: 0002934 | |
| 2005-08-08 18:25 | Goober5000 | Note Added: 0002956 | |
| 2005-08-08 18:25 | Goober5000 | Assigned To | => WMCoolmon |
| 2005-08-08 18:25 | Goober5000 | Status | new => assigned |
| 2005-08-09 05:20 | Fury | Note Added: 0002958 | |
| 2005-08-09 05:21 | Fury | Note Edited: 0002958 | |
| 2005-08-10 02:04 | Fractux | File Added: AI_test.fs2 | |
| 2005-08-10 02:34 | Fractux | Note Added: 0002967 | |
| 2005-08-10 04:35 | taylor | Note Added: 0002968 | |
| 2005-08-10 04:44 | Fractux | Note Added: 0002969 | |
| 2005-08-10 06:02 | taylor | Note Added: 0002971 | |
| 2005-08-10 06:09 | Goober5000 | Note Added: 0002972 | |
| 2005-08-10 06:09 | Fury | Note Added: 0002973 | |
| 2005-08-10 06:09 | Goober5000 | Note Edited: 0002972 | |
| 2005-08-10 06:11 | Goober5000 | Note Added: 0002974 | |
| 2005-08-10 06:42 | taylor | Note Added: 0002975 | |
| 2005-08-10 07:07 | Fury | Note Added: 0002976 | |
| 2005-08-10 07:57 | taylor | Note Added: 0002977 | |
| 2005-08-10 08:08 | taylor | Note Edited: 0002977 | |
| 2005-08-12 18:33 | Goober5000 | Assigned To | WMCoolmon => Goober5000 |
| 2005-08-24 11:26 | kasperl | Note Added: 0003129 | |
| 2005-08-25 00:54 | Goober5000 | Note Added: 0003158 | |
| 2005-08-25 01:05 | taylor | Note Added: 0003161 | |
| 2005-08-25 04:28 | Fury | Note Added: 0003163 | |
| 2005-09-16 16:27 | Goober5000 | Note Added: 0003390 | |
| 2005-10-22 15:36 | taylor | Note Added: 0003631 | |
| 2005-10-23 16:50 | Fury | Note Added: 0003643 | |
| 2005-11-01 03:05 | Goober5000 | Status | assigned => closed |
| 2005-11-01 03:05 | Goober5000 | Note Added: 0003695 |
