View Issue Details

IDProjectCategoryLast Update
0026682AI War 2Bug - GameplayMar 25, 2022 7:09 pm
ReporterTim_Fragmagnet Assigned ToChris_McElligottPark  
Status feedbackResolutionreopened 
Product Version4.002 Macrophage Live! 
Fixed in Version4.004 Easy There, Hack Response 
Summary0026682: Units sometimes not automatically targeting hostiles even when in pursuit mode.
DescriptionThis is on an Expert game with 2 difficulty 7 AI's, no extra factions, an 80 planet map.
The game had been running for about 4.5 hours
The save file is 2:25:12 in, the game had not been closed since the beginning of the save, nor had the save been unloaded and reloaded.

I had 2 separate instances of this occurring

The first instance, a singular AI pulsar tank on one of my planets making its way across the entire planet uninterrupted despite there being turrets and a fleet in pursuit mode on said planet, with a large portion of the mobile units pretty close to the enemy. Selecting units and manually targeting the pulsar tank worked to make them target it.
I figured it was a fluke.


The second instance, 2 warden fleets and 2 hunter fleets, a total fleet strength of about 137, going passive on a planet with an active sniper ring. I watched the tritium snipers pick off targets uninterrupted for a decent bit of time, plenty of time for the AI to wake up and destroy them, it didn't.
At this point, I knew something was up.
Talk with people on discord about it.
Save and quit, copy the save to a separate folder.
Reload the game.

The AI fleets immediately go active and destroy the sniper turrets.
I Have a screenshot of before the save and quit, and after reloading the save file

The save file I have included is the one copied to a separate folder and not loaded. If the enemy targeting logic is only fixed upon reloading the save, and not upon save quitting, then that broken logic is in that file.

This is on the planet Thrun.
TagsNo tags attached.

Activities

Tim_Fragmagnet

Mar 9, 2022 8:15 pm

reporter  

Before savequit.png (3,100,715 bytes)
after reload.png (3,055,702 bytes)

Chris_McElligottPark

Mar 10, 2022 11:07 am

administrator   ~0065337

I suspect the issue was in the auto-targeting stuff, since that's what is in charge of this part of the code. I've made some changes there:

* The tooltips for ships now show a lot more details for any auto-targeting targets they have for each of their systems.
** Additionally, it now shows which auto-targeting group they are a part of.

* The way that auto-targeting groups are calculated is now done centrally, and it's no longer possible for units to potentially get lost in the shuffle and not get auto-targeting orders.

Thanks!

Chris_McElligottPark

Mar 10, 2022 11:46 am

administrator   ~0065339

And some more:

* The way that "incoming shots" are tracked on ships, which is used in overkill detection, has been improved a lot.
** First of all, it's more efficient now, and tracked internally only, and has more granular cross-threading protection rather than allowing for a cross-threading error to interrupt the entire process.
** Secondly, any incoming shot now times out after 10 game seconds, so nothing can get stuck in a "it's going to die because of incoming damage" state where enemy units won't fire on it. Not sure if this was happening, but it's possible, and now it should not be.
** The "Show incoming-shot debug data in unit tooltips" debug setting has been removed, as the individual shots are no longer interesting to see in the same way.
** The tooltips now show how much damage the ship is expected to take from incoming shots, or if there are incoming shots but no damage, then it also shows that.
** It is likely this will fix some bugs, or it's possible it might just make things more clear if a bug in this area resurfaces.

Chris_McElligottPark

Mar 10, 2022 11:52 am

administrator   ~0065340

Okay, this is from our discord discussion:

* Adjusted targeting logic so that if a unit is not fully claimed, but belongs to a player, it is now valid to be shot. Previously, some metal harvesters that were half-alive were not being shot further, which looked like a bug. Now they get pasted instantly.

Hopefully this solves all the issues, but if you see something happen again, this time there's a lot more to see in the tootltips.

Tim_Fragmagnet

Mar 10, 2022 10:56 pm

reporter   ~0065376

Last edited: Mar 10, 2022 11:01 pm

Something is still up with targeting.

On the new patch I have seen an enemy unit attack one of my planets and have nothing on that planet go active on it.
The new tooltip info just shows nothing in the targeting info, the unit isn't being targeted by sniper turrets, and pursuit units aren't moving to engage it, however, if I manually move a pursuit unit so it's in firing range, it does acquire target and fire on its own.

this specific instance isn't new with the patch either, I was seeing this exact thing on the previous one as well.
I have uploaded the same save in its current state at 9.5 hours in.

Tim_Fragmagnet

Mar 11, 2022 4:06 pm

reporter   ~0065385

I'm still seeing this occur on 4.004

Zer0h1nder

Mar 12, 2022 10:58 am

reporter   ~0065392

Can confirm it's still happening. When enemy units are not in attacking range, own units in pursuit mode sometimes don't engage them. For me it happened so far only when the planet was not in focus.

Chris_McElligottPark

Mar 15, 2022 4:07 pm

administrator   ~0065459

At the moment, I don't have any save that will show this for me, or any steps to reproduce it. At the moment I am at a loss.

The fact that the orders list is empty is slightly useful, but it basically just eliminates the most obvious possibility (that the ships not targeting have an existing order that countermands it in some way).

If anyone has this happen, it would be really useful to see what the information in the right-click of the clock shows, especially regarding targeting timings. If one of those threads is stuck or something, that would make sense. That sort of thing would not survive a save and reload.

Strategic Sage

Mar 17, 2022 12:23 am

reporter   ~0065488

I just had this happen to me for the first time. I didn't think to using the data Chris suggested here, I'll try to remember doing that last time. As others have said, reloading fixed it. The one potentially useful bit of info I can mention is that it was a guard post on an enemy units, so it's not just mobile units.

I'll keep watching and try to give better data from the clock right-click if it happens again.

Strategic Sage

Mar 18, 2022 9:30 pm

reporter   ~0065503

All right, got another one. This was even weirder. First planet I attacked, ships on pursuit stopped hitting a Sabot GP and went after a guardian instead. At the time I just thought 'huh, I've never seen them do that before'. But then after the planet was cleaned, they still weren't interested in the Guard Post.

Screenshots of everything with a 'red' value is attached. If there's anything specific beyond this I can capture next time, let me know.
RedOne.png (1,698,183 bytes)
RedThree.png (1,683,162 bytes)
RedTwo.png (1,909,870 bytes)

Chris_McElligottPark

Mar 24, 2022 10:07 pm

administrator   ~0065578

Sorry for the really slow response. On the screenshots, for targeting that is slowed, that's generally what I would need. I suppose one other thing would be to see if the targeting numbers are climbing up or not in the escape menu under the NP-A or whatever section, or the player section.

From what I can see in that screenshot, it looks like the targeting thread was run about 1 second before you paused the game, so no reason to think that it wasn't running continuously.

Strange things to look at would be if you see, while the game is running, the number of calls to the targeting not going up, or it not generating anything for the player (if the player ships are the ones not firing).

If that's not the problem, then targeting is running fine -- my guess based on those screenshots is that it is indeed. At that point, that would leave two possibilities.

First, it could be that the targeting is making a wrong choice for some reason. Incomplete data? Wrong data? Hard to say.

Second, it could be that the targeting data is being calculated perfectly, but then a) it can't get to the unit for some reason; b) there's some issue blocking it from being inserted when it does get to the unit; c) the unit gets it, but is discarding it immediately for some reason (thinking erroneously that the target is invalid, for example).

These are all much harder to debug, and I think that's where we are now. I don't have any real good solutions for tracking that just yet.

Strategic Sage

Mar 25, 2022 5:13 am

reporter   ~0065582

More information - I had another happen where the melee ships, and only the melee ships, were willing to attack. Other ships in the system ignored the targets, and the melee ones did briefly as well, but then they snapped out of it. So based on that and what I saw in a brief look at the numbers, I would definitely say it didn't look like targeting.

Ofc that doesn't say why it *is* happening, but hey another piece of info.

Strategic Sage

Mar 25, 2022 7:09 pm

reporter   ~0065594

I think I have a reliable save for this! At least, I'm seeing it consistently on my end.

Replication Steps:

** Go to the 2e24 Cats In Trenchcoat system. There is an ongoing battle there with some enemy ships and three remaining Stealth Guard Posts. Set fleets 1 and 3 to Pursuit. Do nothing else but wait.
** After almost two minutes, they usually stop firing at the lower-right GP just before it would actually be destroyed. They do not return to it.
** Use your magical powers of code prowess to figure out why :).
TargetingTest.save (280,356 bytes)

Issue History

Date Modified Username Field Change
Mar 9, 2022 8:15 pm Tim_Fragmagnet New Issue
Mar 9, 2022 8:15 pm Tim_Fragmagnet File Added: Before savequit.png
Mar 9, 2022 8:15 pm Tim_Fragmagnet File Added: after reload.png
Mar 9, 2022 8:15 pm Tim_Fragmagnet File Added: Ironman Save at 2h 25m 12s.save
Mar 10, 2022 11:07 am Chris_McElligottPark Note Added: 0065337
Mar 10, 2022 11:46 am Chris_McElligottPark Note Added: 0065339
Mar 10, 2022 11:52 am Chris_McElligottPark Assigned To => Chris_McElligottPark
Mar 10, 2022 11:52 am Chris_McElligottPark Status new => resolved
Mar 10, 2022 11:52 am Chris_McElligottPark Resolution open => fixed
Mar 10, 2022 11:52 am Chris_McElligottPark Fixed in Version => 4.004 Easy There, Hack Response
Mar 10, 2022 11:52 am Chris_McElligottPark Note Added: 0065340
Mar 10, 2022 10:56 pm Tim_Fragmagnet Note Added: 0065376
Mar 10, 2022 10:56 pm Tim_Fragmagnet File Added: Ironman Save at 9h 30m 11s.save
Mar 10, 2022 11:01 pm Tim_Fragmagnet Note Edited: 0065376
Mar 11, 2022 4:06 pm Tim_Fragmagnet Status resolved => feedback
Mar 11, 2022 4:06 pm Tim_Fragmagnet Resolution fixed => reopened
Mar 11, 2022 4:06 pm Tim_Fragmagnet Note Added: 0065385
Mar 12, 2022 10:58 am Zer0h1nder Note Added: 0065392
Mar 15, 2022 4:07 pm Chris_McElligottPark Note Added: 0065459
Mar 17, 2022 12:23 am Strategic Sage Note Added: 0065488
Mar 18, 2022 9:30 pm Strategic Sage Note Added: 0065503
Mar 18, 2022 9:30 pm Strategic Sage File Added: RedOne.png
Mar 18, 2022 9:30 pm Strategic Sage File Added: RedThree.png
Mar 18, 2022 9:30 pm Strategic Sage File Added: RedTwo.png
Mar 24, 2022 10:07 pm Chris_McElligottPark Note Added: 0065578
Mar 25, 2022 5:13 am Strategic Sage Note Added: 0065582
Mar 25, 2022 7:09 pm Strategic Sage Note Added: 0065594
Mar 25, 2022 7:09 pm Strategic Sage File Added: TargetingTest.save