Pathfinding Lag Fix | Pathfinding Lib | Smart Enemy Pathfinding | Lethal Company Modding | Page 4

crimson jolt Feb 4, 2025, 4:45 AM

#

my friend

#

i do not envy you LOL

keen ruin Feb 4, 2025, 4:45 AM

#

I wouldn't tend to say you should go and rewrite it, since I'll end up having to do that too whenever I get around to making the elevator API

crimson jolt Feb 4, 2025, 4:46 AM

#

well it's so heavy handed it would absolutely come with a toggle

green gust Feb 4, 2025, 4:46 AM

#

There's already so many mods that touch the masked, this should be fun

crimson jolt Feb 4, 2025, 4:46 AM

#

and when that future comes i can design it to work in tandem or just dump it completely if your patch is sufficient

crimson jolt Feb 4, 2025, 4:46 AM

#

green gust There's already so many mods that touch the masked, this should be fun

this is the other problem too yeah

keen ruin Feb 4, 2025, 4:46 AM

#

green gust There's already so many mods that touch the masked, this should be fun

yeah I don't really care lmao

they can use the API themselves if they want to

#

it definitely will solve the problem better than the things I've seen for these types of problems in the past

#

it's just a lot of work and I have to make sure the API is solid enough to actually not need a rework later

#

that's the most daunting part

#

probably should run my idea by you once I revise it actually @crimson jolt

#

I gotta figure out how to communicate a path to an arbitrary transporter device while giving the API user enough control to animate things how they want

#

I had talked this through with Matty before and come up with something decent, but it didn't account for entrance teleports, only elevators

crimson jolt Feb 4, 2025, 4:51 AM

#

hahahahaaa yeeess my chidlren

keen ruin Feb 4, 2025, 4:51 AM

#

oh god

#

also buttery I give up

crimson jolt Feb 4, 2025, 4:51 AM

#

with the elevator?

keen ruin Feb 4, 2025, 4:51 AM

#

feel free to do a prefix cancel shrug making a transpiler for this would be too much of a pain

crimson jolt Feb 4, 2025, 4:51 AM

#

lol

#

yep

#

if even you don't want to handle it

ashen harbor Feb 4, 2025, 4:51 AM

#

keen ruin feel free to do a prefix cancel <:shrug:976243911922167899> making a transpiler ...

thanks Zeekerss

crimson jolt Feb 4, 2025, 4:51 AM

#

i know there is absolutely no way i can figure it out

keen ruin Feb 4, 2025, 4:52 AM

#

you can probably change the flag variable to be whether someone is in the top area of the elevator, but then you also have to hijack isInElevatorStartRoom or create a new variable for whether the masked is also in that top area

#

but it's a pain

crimson jolt Feb 4, 2025, 4:52 AM

#

well

#

im thinking if i do a bounds check

#

for both the players and the masked

keen ruin Feb 4, 2025, 4:52 AM

#

if you wanna set up bounds for that then patching it may be workable

crimson jolt Feb 4, 2025, 4:53 AM

#

i can get a pretty clean and solid answer for "are they on the same side of the elevator or not"

#

their wandering behavior is really simple at the heart of it all

#

if there's a player on the same "layer" as them, they wander randomly

#

otherwise they travel to where they can find players

crimson jolt Feb 4, 2025, 4:54 AM

#

crimson jolt otherwise they travel to where they can find players

all i need to do is probably a version of this function that works for the factories and manors

keen ruin Feb 4, 2025, 4:54 AM

#

ok then maybe you could transpile

#

up to you

crimson jolt Feb 4, 2025, 4:54 AM

#

then write a special version for the mineshafts that considers the elevators and does the bound checks i mentioned, etc.

#

and then i need to force the masked to not be able to use the elevator unless it is guaranteed to take them closer to where the players are

keen ruin Feb 4, 2025, 4:54 AM

#

basically I think the intention was that flag is whether the player is in that top room of the elevator, and same for isInElevatorStartRoom (obviously), so you could just change both of those to bounds checks and it should theoretically work

crimson jolt Feb 4, 2025, 4:54 AM

#

so they stop going up and down forever

crimson jolt Feb 4, 2025, 4:55 AM

#

crimson jolt and then i need to force the masked to not be able to use the elevator unless it...

i thought this might be heavy-handed at first, but this is already pretty much how they work in vanilla

#

if no players are in the building, they "just know" and automatically path straight to the main entrance with no meandering

#

so zeekerss probably intended the same for the elevator and it just got messy to work out a way to do that

#

so i pretty much have no qualms just boiling it down to that to absolutely ensure they won't get stuck

#

anyways all that being said

#

i definitely am not waiting to do all of this before i release the first version of this plugin

#

this will be v1.1 work probably

keen ruin Feb 4, 2025, 4:57 AM

#

yeah

#

sounds good

#

maybe I should beat you to it with the API lol but god I don't look forward to maintaining two APIs for something that might be used so heavily

#

someone is undoubtedly gonna find holes in it and make me have to make breaking changes

#

(not that I'm afraid to do that if I have to teehee )

crimson jolt Feb 4, 2025, 4:58 AM

#

lol well im maybe not your best consumer anyway

keen ruin Feb 4, 2025, 4:58 AM

#

I think I have to be my own consumer

#

for two reasons

crimson jolt Feb 4, 2025, 4:58 AM

#

i would be most interested in utilizing it just to ensure compatibility

#

but i have close-to-zero interest in custom content for lethal company and almost all of my stuff is just designed with fixing/polishing/restoring vanilla concepts

keen ruin Feb 4, 2025, 4:58 AM

#

I get to dogfood my API, and I can ensure that people have a single mod to depend on that gives them a way to create any number of elevators

crimson jolt Feb 4, 2025, 4:59 AM

#

so i would solely be using it to make masked work solely in mineshaft elevators and would probably not be able to provide much testing in other departments

keen ruin Feb 4, 2025, 4:59 AM

#

yeah I wouldn't be inclined to do this if I wasn't already in the business of async pathfinding and also creating an elevator in Black Mesa

#

it's not my idea of a great time but I don't think anyone else is ever gonna do it

#

and yet everyone wants to make elevators

#

Despairge

crimson jolt Feb 4, 2025, 5:00 AM

#

well if there's anybody right for the job it would be you

#

i will say

keen ruin Feb 4, 2025, 5:00 AM

#

yeah

#

wish I wasn't lol

#

I have so many other things that I've put off for a long while already

crimson jolt Feb 4, 2025, 5:02 AM

#

ok ive gotta go pull the trigger on 3 separate releases

#

if im lucky maybe i wont forget to ship the DLL in one of them

#

this is about to be sort of stressful

#

be back

#

probably

keen ruin Feb 4, 2025, 5:02 AM

#

good luck

ashen harbor Feb 4, 2025, 5:05 AM

#

crimson jolt if im lucky maybe i wont forget to ship the DLL in one of them

You got this

crimson jolt Feb 4, 2025, 6:00 AM

#

i can check really quick if nobody has the answer on hand

#

but is the entrance room always rotated the same way

#

for a particular interior type

ashen harbor Feb 4, 2025, 6:13 AM

#

Not sure

keen ruin Feb 4, 2025, 6:17 AM

#

yeah I don't actually know either

#

but whatever the case may be, I would recommend making a list of vertices instead and then transform them to world space by the tile transform, then encompass those in an axis-aligned bounding box

crimson jolt Feb 4, 2025, 6:24 AM

#

lol i was just going to make a giant boxcollider as a child of the tile, apply a pre-calculated position/scale, calculate bounds, and then delete the object

keen ruin Feb 4, 2025, 6:24 AM

#

oh don't do that

#

that's a waste of work, it shouldn't be hard to transform vertices

crimson jolt Feb 4, 2025, 6:25 AM

#

then maybe i ought to learn me a thing

keen ruin Feb 4, 2025, 6:25 AM

#

https://github.com/2394425147/LC_CullFactory/blob/master/CullFactory/Data/Portal.cs#L16-L61

this is kinda in the vein of what you need

#

only 4 corners here, but you can expand it to 8 no problem

#

and you can use this to transform them https://docs.unity3d.com/6000.0/Documentation/ScriptReference/Transform.TransformPoints.html

crimson jolt Feb 4, 2025, 6:29 AM

#

oh i see

#

so i guess im just defining like

#

the shape i'd want my box collider to be

#

basically

#

and then i am transforming them by the start tile's transform

#

to get where they'd be in actual world space

crimson jolt Feb 4, 2025, 6:30 AM

#

crimson jolt lol i was just going to make a giant boxcollider as a child of the tile, apply a...

if i had gone through all of this

keen ruin Feb 4, 2025, 6:31 AM

#

yeah

#

each vertex would be a corner of the cube, that gets transformed to world space

crimson jolt Feb 4, 2025, 6:31 AM

#

ya

#

well ok that seems pretty easy

#

as long as transformpoints accounts for rotation

#

which i see no reason to assume it wouldnt

keen ruin Feb 4, 2025, 6:32 AM

#

it does

crimson jolt Feb 4, 2025, 6:32 AM

#

then yeah i'll probably do that

#

it's getting to be pretty late and im probably about to head off

#

but i already have a skeleton of the new roaming behavior basically done

#

bool wantsToBeInside = closestInsidePlayer != null && (!maskedAI.isOutside || closestOutsidePlayer == null);

if (wantsToBeInside == maskedAI.isOutside && Time.realtimeSinceStartup - maskedAI.timeAtLastUsingEntrance > 3f)
    UseEntrance(maskedAI);
else if (!maskedAI.searchForPlayers.inProgress)
    maskedAI.StartSearch(maskedAI.transform.position, maskedAI.searchForPlayers);

this is a snippet

#

it is so much easier boiled down like this

#

but tomorrow i need to write the actual meat of UseEntrance which is going to need to sample bounds in mineshafts

#

and have all the logic to make masked use the elevator

#

unless i can actually reuse zeekerss' function without issue

#

we'll have to see

keen ruin Feb 4, 2025, 6:34 AM

#

I think UseEntrance just sends the AI to the entrance if it's pathable, so yeah it should work

#

also sounds good, good night!

crimson jolt Feb 4, 2025, 6:35 AM

#

keen ruin I think UseEntrance just sends the AI to the entrance if it's pathable, so yeah ...

oh to clarify UseEntrance is a custom function im going to write with all the pathfinding stuff handled

#

although i might be able to reuse whatever the vanilla function is, depending

#

GoEntrance i think it's called?

keen ruin Feb 4, 2025, 6:35 AM

#

oh right yeah

crimson jolt Feb 4, 2025, 6:35 AM

#

but yeah

keen ruin Feb 4, 2025, 6:35 AM

#

GoToEntrance

#

I wasn't looking at the code so I assumed you were talking about duplicating it lol

crimson jolt Feb 4, 2025, 6:36 AM

#

anyways fun stuff happening but it can wait until the morning

#

i appreciate the chatter

#

hope it didnt distract you too much

keen ruin Feb 4, 2025, 6:36 AM

#

NODDERS

#

no I wasn't really doin that much lol

tardy pivot Feb 4, 2025, 7:41 AM

#

keen ruin I had talked this through with Matty before and come up with something decent, b...

from this weekend i can probably help you again with that, in case you need :D

keen ruin Feb 4, 2025, 11:58 AM

#

yeah, if I have time that definitely wouldn't hurt to run some ideas by you

crimson jolt Feb 4, 2025, 2:52 PM

#

i had to grow the scope of my AI a bit this morning

#

because i realized i was making some bad assumptions

#

but i think this is a really easy way to integrate the elevator into it

#

#

UseElevatorIfNeeded immediately returns true if the masked is outside or the elevator does not exist

#

and then it goes straight into its behavior of "path straight to main entrance, or search for players on this layer"

#

the idea is that, in a mineshaft, UseElevatorIfNeeded would check:

where the masked is in the interior
where the closest player is in the interior (if players are inside)
if the masked wants to go outside

#

and then there would be a boolean flag like "needsToUseElevator" that would be set to true if the masked can't reach its goal without using the elevator

#

if that is the case, it would path straight to the elevator, call it, ride it, etc. do whatever is needed to get it to the other side of the elevator

#

but it'll be blocked from using the elevator otherwise

#

@keen ruin i did want to check with you on this since i believe this is what you warned against

#

in vanilla, masked used to do CalculatePath in v56 before they would path to the main entrance

#

and if main entrance wasn't reachable, they would wander the interior randomly

#

i'd like to restore that behavior if it is possible, since i think it's a little less ugly than them getting stuck in place

#

but iirc CalculatePath was extremely expensive to be calling every frame

#

or, well, every AI interval

#

and i thought that's what pathfindinglib was supposed to offer a better alternative for

keen ruin Feb 4, 2025, 3:00 PM

#

yeah it is, and I would definitely say it would be nice to make use of it if you have a good place to do so

#

but it is a little involved to set up

#

the readme should give you an idea of how to, but there maybe missing info

crimson jolt Feb 4, 2025, 3:01 PM

#

well im between two places right now

#

1 is "dont make them calculate path" and leave it as-is, since they dont do it in v60+ and it works mostly fine

#

2 is "use pathfindinglib to do it the right way"

#

so im definitely interested

crimson jolt Feb 4, 2025, 3:01 PM

#

keen ruin the readme should give you an idea of how to, but there maybe missing info

i was just reading that in the meantime

green gust Feb 4, 2025, 3:01 PM

#

Well what do you consider "the right way"?

crimson jolt Feb 4, 2025, 3:02 PM

#

not using CalculatePath every AI interval

green gust Feb 4, 2025, 3:02 PM

#

Well that's definitely the right way yeah lol

crimson jolt Feb 4, 2025, 3:02 PM

#

if it's expensive, and vanilla masked get away with not doing it, i dont want to add a high performance cost just for this minor feature

#

hence pathfindinglib or leaving it out

keen ruin Feb 4, 2025, 3:02 PM

#

one CalculatePath per interval is far from the worst thing you can do but it's also not ideal

crimson jolt Feb 4, 2025, 3:02 PM

#

yeah

keen ruin Feb 4, 2025, 3:03 PM

#

basically you would have to have a class where you track the status of all jobs you care about, both so you can query them and so you can dispose of them so that PathfindingLib can reclaim memory

#

in PathfindingLagFix I just have a class wrapping an array that expands to fit all thisEnemyIndex

green gust Feb 4, 2025, 3:04 PM

#

Depending on how you do it, you can expand on the masked by allowing them to use fire exits too greed , but idk if you consider this expansion "bad" since zeekerss just uses main entrance

#

Though I kinda remember they used fire exits once upon a time

#

https://github.com/XuuXiaolan/CodeRebirth/tree/main/Plugin/CodeRebirth/src/MiscScripts/PathFinding I think I'm currently the only mod that uses pathfindinglib too, if it helps this is my code, most of it was just me following what zaggy said lol

keen ruin Feb 4, 2025, 3:11 PM

#

in other words if something looks stupid it's probably my fault

#

teehee

green gust Feb 4, 2025, 3:12 PM

#

Si

#

Though I really need to make my own DirectPathfindingOperation at some point lol

keen ruin Feb 4, 2025, 3:13 PM

#

yeah lol

#

also that code is set up more for something where you have full control over the AI pathfinding code, so it might not totally fit the use case

#

but the disposal stuff might be useful if you want to ensure you don't keep data around when it shouldn't be kept

#

PathfindingLagFix doesn't use any PathfindingLib pools and just maintains its own jobs instead so I can't necessarily say what is best in that regard

crimson jolt Feb 4, 2025, 3:19 PM

#

anyways i will probably do the pathfindinglib integration in a v1.2 or such

#

for now i just want to get to the core of the issue with v1.1, replacing the roam state with one that functions 99% of the time

crimson jolt Feb 4, 2025, 3:20 PM

#

green gust Depending on how you do it, you can expand on the masked by allowing them to use...

this is something i'd like to allow in the future

#

probably as a disabled-by-default config

#

it is going to make my mineshaft behavior a bit tricky though

#

since right now im making the assumption masked have to use the main entrance, and it helps me keep the "expected elevator behavior" really straightforward and easy

#

if they can skip the elevator by taking fire exits i need to account for that possibility

#

and then i would need to decide if i want the masked to do some sort of consideration like "is the closest inside player closer to the fire exit or closer to the elevator"

#

or if i just want it to be random

#

etc.

#

lots of stuff to consider and im not in a hurry to tackle that immediately

green gust Feb 4, 2025, 3:27 PM

#

Makes sense

#

For me in my own logic, if I wanna go outside and I'm in mineshaft below elevator, I just prioritise fire exits, and if there's no route with those exits I try to check if there's a route with elevator etc

keen ruin Feb 4, 2025, 3:28 PM

#

the pathfinding through fire exits thing is probably gonna be an option in whatever solution I come up with for elevators

crimson jolt Feb 4, 2025, 3:29 PM

#

would certainly be nice

keen ruin Feb 4, 2025, 3:29 PM

#

I'll have to make it so that it can reject certain paths though

crimson jolt Feb 4, 2025, 3:55 PM

#

wait nvm

crimson jolt Feb 4, 2025, 3:55 PM

#

crimson jolt 1 is "dont make them calculate path" and leave it as-is, since they dont do it i...

they do have it in v60

#

it was just moved

#

so i guess i will just make use of calculate path in the meantime

crimson jolt Feb 4, 2025, 4:13 PM

#

@keen ruin how would you suggest creating the bounds?

#

should I just create the bounds using the first vertex and a size of 0,0,0

#

and then feed it all the other vertices using https://docs.unity3d.com/6000.0/Documentation/ScriptReference/Bounds.Encapsulate.html

keen ruin Feb 4, 2025, 4:19 PM

#

the CullFactory code I sent has what you should do in it

#

the portal constructor

crimson jolt Feb 4, 2025, 4:20 PM

#

ah right

keen ruin Feb 4, 2025, 4:20 PM

#

it's more efficient to work on the min/max yourself than feed it to the bounds struct, since the underlying fields for that are not min/max but rather center and extents

crimson jolt Feb 4, 2025, 4:20 PM

#

yea

keen ruin Feb 4, 2025, 4:21 PM

#

so if you repeatedly read and set min/max on bounds, it is recalculating it in both directions

crimson jolt Feb 4, 2025, 4:21 PM

#

yeah

#

it doesn't have to be super optimized because im just calculating the bounds after the level is done generating and then caching it

#

but like

#

i see no reason to do it wrong if i can do it right

#

basically

keen ruin Feb 4, 2025, 4:27 PM

#

yeah

#

not gonna say you have to do it that way since it will be imperceptible, but the example was there lol

crimson jolt Feb 4, 2025, 6:45 PM

#

i got it to work

#

unfortunately

#

vanilla's behavior uses a lot more PathIntersectedByLineOfSight than i thought

#

so I had like

#

4 total CalculatePath calls

#

i think it only does one per frame, IIRC

#

with how the logic is laid out

#

so it's not the worst

#

but it does seem like it'll be a puzzle to implement pathfindinglib when the time comes

maiden sedge Feb 4, 2025, 7:22 PM

#

I know this thread is for the path finding mod but is anyone here up to help me debug something?

green gust Feb 4, 2025, 7:28 PM

#

If its unrelated then #help-and-troubleshooting

maiden sedge Feb 4, 2025, 7:28 PM

#

green gust If its unrelated then <#1169069074572116041>

Ok thanks

green gust Feb 4, 2025, 7:29 PM

#

Like you said, this thread is for the pathfinding mod, if you notice vanilla enemies or enemies have been acting super weird with pathing, then you'd ask here, otherwise won't be of much help

crimson jolt Feb 4, 2025, 10:23 PM

#

https://github.com/ButteryStancakes/MaskFixes/blob/1de05ae28cf747eaa73ac21c7f146bbd0f634d7f/NewMaskedAI.cs#L50

GitHub

MaskFixes/NewMaskedAI.cs at 1de05ae28cf747eaa73ac21c7f146bbd0f634d7...

Lethal Company plugin. Fixes several problems with mask items and masked enemies. - ButteryStancakes/MaskFixes

#

this is what i ultimately settled on

#

i think this is at least clean and easy to read through and understand

#

there are some performance concerns with CalculatePath() but i got full frames on my computer with 10+ masked

#

i really only noticed slowdown in stress test scenarios (like the 50 masked elevator queue test)

#

definitely some room for improvement in the future but for now I'm pretty happy

steep tartan Feb 4, 2025, 10:32 PM

#

crimson jolt there are some performance concerns with CalculatePath() but i got full frames o...

well, this made it possible for this mod to run smoothly hehe
https://thunderstore.io/c/lethal-company/p/SimonTendo/LCMaskedInfestationMod/

Thunderstore

LCMaskedInfestationMod v1.0.0

Add the possibility for infestations to spawn Maskeds!

keen ruin Feb 4, 2025, 10:35 PM

#

crimson jolt there are some performance concerns with CalculatePath() but i got full frames o...

did you check what the frame time consistency was like?

#

also, using PathfindingLib shouldn't be so bad, you'll probably end up wanting to implement it similarly to PathfindingLagFix

#

assuming you always only need the status of one path anyway

#

it just stores an integer representing the current pathfinding operation, and if that changes, it resets the status and keeps whatever destination it had before until the next result is ready

crimson jolt Feb 4, 2025, 10:41 PM

#

i'll have to take a look at what your original patch is doing

#

but it calculates paths to several different positions

#

so it'll probably be a bit more involved than that

#

each different position happens with a different set of criteria, and iirc im only calculating 1 (maybe 2 sometimes?) paths in the same ai interval

keen ruin Feb 4, 2025, 10:42 PM

#

it calculates multiple in one frame?

crimson jolt Feb 4, 2025, 10:42 PM

#

at least, separated by destination, i mean

#

let me check and see

keen ruin Feb 4, 2025, 10:43 PM

#

multiple in a frame is the only reason why it might be an issue

#

and that is still manageable, it just requires a bit more setup

crimson jolt Feb 4, 2025, 10:43 PM

#

oh oops

#

i forgot to add a shortcut for indoor players

#

or wait no i didnt

#

okay yeah, currently it calculates a path to each indoor player if the masked itself is also indoors

#

the path calculation happens last after all the other targetable criteria is checked to try and avoid that as much as possible, but it will still happen in circumstances where the masked is wandering inside and multiple players are inside to target

#

it looks like under any other circumstances, the masked only calculates one path per frame

#

if i remember how i set this all up

#

and didn't make any mistakes

keen ruin Feb 4, 2025, 10:45 PM

#

you mean vanilla code calls TargetClosestPlayer?

#

or something else does this?

crimson jolt Feb 4, 2025, 10:45 PM

#

vanilla does GetClosestPlayer

keen ruin Feb 4, 2025, 10:46 PM

#

that does path checks?

crimson jolt Feb 4, 2025, 10:46 PM

#

which, if PlayerIsTargetable is successful, does PathIsIntersectedByLineOfSight

keen ruin Feb 4, 2025, 10:46 PM

#

ew

crimson jolt Feb 4, 2025, 10:46 PM

#

it doesnt actually check the line of sight but it does use that function for CalculatePath

keen ruin Feb 4, 2025, 10:46 PM

#

okay well I should be able to patch that

#

I could've sworn that one omitted the path check but I guess I'll have to look at it

crimson jolt Feb 4, 2025, 10:47 PM

#

it runs all of this

#

everything below this is skipped

#

if im reading this correctly

#

in my case, i didnt even check the path corners or do the distance check

#

i just did CalculatePath

#

so im probably returning slightly different results than vanilla would be

#

and maybe i ought to patch that

keen ruin Feb 4, 2025, 10:52 PM

#

crimson jolt it runs all of this

I didn't mean that method, I meant GetClosestPlayer

that method is used all over the place, it's what I'm replacing with my own code wherever possible

crimson jolt Feb 4, 2025, 10:53 PM

#

crimson jolt it runs all of this

i think i must be crazy because i thought i remembered GetClosestPlayer called this function

#

but i just checked again

#

and no it doesn't

keen ruin Feb 4, 2025, 10:53 PM

#

crimson jolt in my case, i didnt even check the path corners or do the distance check

you will want to do that or you might get some stuttery movement if they try to path too far

#

I think you must have looked at TargetClosestPlayer then

#

that is the method that does that

crimson jolt Feb 4, 2025, 10:54 PM

#

oh yep you're right

#

oops

crimson jolt Feb 4, 2025, 10:55 PM

#

keen ruin you will want to do that or you might get some stuttery movement if they try to ...

i didn't notice any issues with this while i was testing but i will go ahead and fix it pre-emptively

#

just in case

#

anyway

#

i can actually remove the path check for indoor players then

#

since vanilla doesn't do it

#

it just cares if a targetable player is in the building at all, not if the current masked can reach it

#

and at that point im pretty sure im back to only calculating one path every frame again

crimson jolt Feb 5, 2025, 12:22 AM

#

im getting a weird lag spike whenever i enter a mineshaft fire exit with an insane amount of masked spawned (~50)

#

i think it's related to some of the changes i made after publishing the github files because i didn't have this issue before

crimson jolt Feb 5, 2025, 12:22 AM

#

crimson jolt im getting a weird lag spike whenever i enter a mineshaft fire exit with an insa...

it was slowing my game to a crawl for about ~3 seconds and then clearing back up

#

profiler showed a ton of log callbacks, and outside of that, it seems like maskedplayerenemy.update (presumably DoAIInterval) and occludeaudio.update (which is probably also related to the masked, since they have 3 per)

#

sort of bizarre

#

by forcing the unity log level not to print anything it cleared up most of the lag but not all of it

#

it doesn't cause performance issues for me with a reasonable amount of masked (10) but im wondering where this suddenly came up from

keen ruin Feb 5, 2025, 12:26 AM

#

if the profiler is just showing LogToConsole you might want to install AsyncLoggers to reduce that

crimson jolt Feb 5, 2025, 12:26 AM

#

yeah i normally have it disabled on my test profile because it causes logs to get out of order

keen ruin Feb 5, 2025, 12:26 AM

#

it collects stack traces always when attached I think

crimson jolt Feb 5, 2025, 12:26 AM

#

but i do use it in normal games

keen ruin Feb 5, 2025, 12:26 AM

#

(unless you change a setting)

crimson jolt Feb 5, 2025, 12:26 AM

#

im gonna test it again with async to see how much that helps

#

it's a super weird issue

keen ruin Feb 5, 2025, 12:27 AM

#

I think the out of order thing is either less of an issue or not an issue anymore, but Matty would have to confirm

#

but yeah, I tend not to run it because I like being able to pause the game

crimson jolt Feb 5, 2025, 12:27 AM

#

it is a terrible lag spike specifically when entering fire exits, only when it's mineshaft, only with masked on the map

keen ruin Feb 5, 2025, 12:28 AM

#

huh that's really curious

crimson jolt Feb 5, 2025, 12:29 AM

#

yeah it mostly goes away with asyncloggers

#

it's definitely not invisible, but i could stomach a 1 second blip

#

i assume it has to be worse for anyone running lower spec hardware though

#

what on earth is happening

keen ruin Feb 5, 2025, 12:29 AM

#

definitely shouldn't need to have any delay as far as I'm aware

crimson jolt Feb 5, 2025, 12:32 AM

#

this is my first time ever using a profiler so i'm maybe not aware of the best practice

#

huge spike, most of it coming from scripts, which checks out

#

pretty unsurprisingly, it gets way worse with pathfindinglagfix disabled

#

oh huh this happens in vanilla too

#

i disabled my patch and im still getting it

#

that's a relief but still an oddity

gentle charm Feb 5, 2025, 1:25 AM

#

@keen ruin can you publish the PathFindingLib mod on NuGet?

keen ruin Feb 5, 2025, 1:38 AM

#

crimson jolt i disabled my patch and im still getting it

did you manage to see what behaviors are doing it?

crimson jolt Feb 5, 2025, 1:39 AM

#

no, it didn't go any deeper than MaskedPlayerEnemy.Update

#

taking up about half as much of that was OccludeAudio.Update

#

that's what i would need deep profiling or profiling markers for right?

#

to get more specific information

keen ruin Feb 5, 2025, 1:40 AM

#

gentle charm <@193574164639121409> can you publish the PathFindingLib mod on NuGet?

uhhh I could look into it.. I would have to set up a build system to be at all consistent with that though, and I haven't ever looked at how others have done that

#

there's a thunderstore source that I just use for everything though

#

not sure if I can bundle docs into a NuGet package, that feels like the only big benefit over using that source

keen ruin Feb 5, 2025, 2:10 AM

#

if anyone has any repos where they have github actions set up to upload to NuGet then I could just adapt that though NODDERS

#

otherwise, I just sent this to giosuel today and you could reference it to get the lib assembly from thunderstore as well
an example of the thunderstore nuget setup is here
https://github.com/Zaggy1024/LC_OpenBodyCams/blob/master/NuGet.Config#L5
and here
https://github.com/Zaggy1024/LC_OpenBodyCams/blob/master/OpenBodyCams/OpenBodyCams.csproj#L33

tardy pivot Feb 5, 2025, 7:18 AM

#

that nuget from windows10ce is really handy

tardy pivot Feb 5, 2025, 7:28 AM

#

keen ruin I think the out of order thing is either less of an issue or not an issue anymor...

it should habe been solved in V2. it was one of the issues reported before so i tried to ensure that, at least in the managed listeners ( eg: BepInEx LogOutput.log ) they are printed in the order they are received

and for unity logs i made so asyncloggers doesn't touch it by default, so should be all in order there too

ashen harbor Feb 5, 2025, 11:54 PM

#

@keen ruin @tardy pivot any idea if you guys can look into getting this error spam fixed

📎 message.txt

#

Baboon hawks throwing this with Mimics all the time is so annoying

keen ruin Feb 6, 2025, 12:10 AM

#

huh ??

#

did a masked get wormed or something??

crimson jolt Feb 6, 2025, 12:11 AM

#

this isn't a pathfinding thing

keen ruin Feb 6, 2025, 12:11 AM

#

yea definitely not

ashen harbor Feb 6, 2025, 12:11 AM

#

True it's just an error that I see all the time in my logs

keen ruin Feb 6, 2025, 12:11 AM

#

and it also would change behavior to fix it, so I am wary of this

crimson jolt Feb 6, 2025, 12:12 AM

#

keen ruin and it also would change behavior to fix it, so I am wary of this

would it meaningfully change behavior?

#

im not even sure how this error is happening in the first place

keen ruin Feb 6, 2025, 12:13 AM

#

the masked is probably getting destroyed by some other enemy

crimson jolt Feb 6, 2025, 12:13 AM

#

i dont understand how the masked can exist where the baboon pings it

#

but the navmeshagent is null

#

.velocity doesn't throw nullref if the agent is disabled or something, does it?

keen ruin Feb 6, 2025, 12:13 AM

#

the IVisibleThreat is a destroyed behavior

#

that's how that happens

crimson jolt Feb 6, 2025, 12:14 AM

#

one second

keen ruin Feb 6, 2025, 12:14 AM

#

Unity's NullReferenceExceptions are very deceptive

#

they have a MissingReferenceException in editor that I really wish they had hooked up in release

#

or maybe it's two, MissingReference and something to indicate destruction, I forget

crimson jolt Feb 6, 2025, 12:15 AM

#

ahh i see, the baboons are caching a threat, i thought this would only apply if they had grabbed the masked on the same frame

crimson jolt Feb 6, 2025, 12:15 AM

#

keen ruin Unity's NullReferenceExceptions are very deceptive

no kidding

#

weird that it throws a nullref on get_velocity if it shouldn't even be able to get a reference to the agent in the first place

#

is it just weird with interfaces? or does it do this with everything? i've never seen this before

keen ruin Feb 6, 2025, 12:16 AM

#

the agent is a managed field, not a native one

#

Unity can only throw NullReferenceExceptions when you call one of the runtime functions

#

you can do anything with a destroyed behavior as long as you're not interacting with the runtime

#

that's why you can still access stuff on your plugins with the BepInEx setting to destroy them after chainloader

crimson jolt Feb 6, 2025, 12:18 AM

#

i see

#

that is a horribly obfuscative behavior

keen ruin Feb 6, 2025, 12:19 AM

#

it's only confusing because the Unity devs decided it should reuse an exception with a very specific meaning lol

#

classic Unity

#

also the behavior change would be that the code could continue execution

crimson jolt Feb 6, 2025, 12:20 AM

#

yeah im seeing it now

keen ruin Feb 6, 2025, 12:20 AM

#

not sure if this is possible to happen in vanilla though, I wouldn't doubt that someone is screwing with something to make this happen

crimson jolt Feb 6, 2025, 12:21 AM

#

in my mind i thought this was just causing an exception for one frame, then it would resolve itself the next

#

but since it's a cached value it is going to be persistent yeah

crimson jolt Feb 6, 2025, 12:21 AM

#

keen ruin not sure if this is possible to happen in vanilla though, I wouldn't doubt that ...

it is possibly something to do with mirage or one of the hundreds of other masked mods

ashen harbor Feb 6, 2025, 12:22 AM

#

Probably Mirage

crimson jolt Feb 6, 2025, 12:22 AM

#

just the other day i was talking about a weird bug that is caused by vanilla behavior but only occurs in modded circumstances

ashen harbor Feb 6, 2025, 12:22 AM

#

I've seen this error for a long time

crimson jolt Feb 6, 2025, 12:22 AM

#

zeekerss added some code to the masked to make them turn invisible if they are mimicking a player and that player has a corpse spawned

#

since being converted into a masked first spawns a ragdoll, then spawns the masked mimicking you, then deactivates your ragdoll

#

in vanilla that works fine, and it prevents 2 bodies of the same player being displayed at the same time

#

but then if you have a mod like mirage that lets masked mimic living players, it causes those mimics to turn permanently invisible if that player dies to any means that spawns a corpse

crimson jolt Feb 6, 2025, 12:24 AM

#

crimson jolt since being converted into a masked first spawns a ragdoll, then spawns the mask...

because that modded behavior breaks this assumption that works fine in vanilla

#

it's possible there could be something similar happening here but TBH i don't think the worm is a bad guess either

keen ruin Feb 6, 2025, 12:25 AM

#

huh, I think the mimickingPlayer field is intended to be used in conjunction with PlayerControllerB.redirectToEnemy, so that seems a bit sus to me

ashen harbor Feb 6, 2025, 12:25 AM

#

I can say a mimic didn't die to a worm, we didn't have a worm that game. Not saying it couldn't happen though if a masked died to one

keen ruin Feb 6, 2025, 12:25 AM

#

but I'm unsure how it should be set up

crimson jolt Feb 6, 2025, 12:25 AM

#

ashen harbor I can say a mimic didn't die to a worm, we didn't have a worm that game. Not say...

if it wasn't destroyed by a worm then im thinking it would have to be caused by some mod changing vanilla behavior

green gust Feb 6, 2025, 12:26 AM

#

why was the stack trace part of baboonhawk's doaiinterval?

crimson jolt Feb 6, 2025, 12:26 AM

#

baboon hawks call IVisibleThreat.GetThreatVelocity() on the enemy they are focusing

ashen harbor Feb 6, 2025, 12:26 AM

#

Full log if it helps

📎 LogOutput_2.log

green gust Feb 6, 2025, 12:26 AM

#

ah right

#

nvm then that makes sense

crimson jolt Feb 6, 2025, 12:26 AM

#

all outdoor enemies (or enemies that can go outdoors) implement this *interface so they can be targeted by baboons (and old birds)

#

i noticed another weird bug with masked the other day while testing my mineshaft patch

green gust Feb 6, 2025, 12:27 AM

#

ye i messed with that interface before, StarlancerEnemyEscape also implements this interface on enemies that dont have iirc

crimson jolt Feb 6, 2025, 12:27 AM

#

masked cause nullreferenceexceptions in quicksand triggers because zeekerss tagged their body as Player

green gust Feb 6, 2025, 12:27 AM

#

this is kinda funny though, radmech threattype lol

keen ruin Feb 6, 2025, 12:28 AM

#

I'm guessing this was meant to have a null check for the threat interface's object as well perhaps

crimson jolt Feb 6, 2025, 12:28 AM

#

crimson jolt masked cause nullreferenceexceptions in quicksand triggers because zeekerss tagg...

the quicksand triggers are reused for mineshaft water tunnels so if a masked walks in/out of them they spam nullreferenceexceptions and it is super ugly

#

but i doubt the player tag is contributing directly to this bug

keen ruin Feb 6, 2025, 12:29 AM

#

green gust this is kinda funny though, radmech threattype lol

lol what the hell, I wonder why it's not player

crimson jolt Feb 6, 2025, 12:30 AM

#

keen ruin lol what the hell, I wonder why it's not player

i was gonna guess that it was to make old birds not target masked

#

but that is done with a typecheck for RadMechAI instead

#

unless im missing it, it looks like baboons never check the type either

#

so im not sure what's up

#

but it seems like it shouldn't have any effect on anything, in that case

keen ruin Feb 6, 2025, 12:37 AM

#

huh, I see

#

curious indeed

ashen harbor Feb 7, 2025, 10:06 AM

#

keen ruin curious indeed

I have an odd question, this mod doesn't replace the Bracken's animations at all does it? I've just noticed when we have the Nymph Bracken recently when she's going after a player her animations are really weird and she doesn't do her normal skipping animation

keen ruin Feb 7, 2025, 10:06 AM

#

only thing it could affect is the conditions where it switches states if they depend on pathing

#

or if they are using transpilers to change animation parameters and they are failing

#

I don't know anything about how they set up the mod though so who knows what it would be

ashen harbor Feb 7, 2025, 10:08 AM

#

https://thunderstore.io/c/lethal-company/p/AntlerShed/NymphBrackenSFW/source/ Lucky for you it has source

Thunderstore

Decompiled source of NymphBrackenSFW v1.2.1

Replaces the Bracken on certain maps with an overly-friendly nymph

keen ruin Feb 7, 2025, 10:11 AM

#

das decomp but sure

#

wait

#

why does it have animancer bundled ?????

#

I don't even know what the legal way to use animancer in a mod here is but bundling it like that doesn't seem right to me lol

#

and also if they are using animancer for this then it's almost undoubtedly some issue with their animator and/or its syncing to the bracken's AI state

#

not much I can do if they screw that up

#

I also have no clue what BitoSkin is

ashen harbor Feb 7, 2025, 10:17 AM

#

Bito is the name of the skin lol

#

What does animancer do?

#

🤔

keen ruin Feb 7, 2025, 10:17 AM

#

it's a replacement for the animation system that LC uses

ashen harbor Feb 7, 2025, 10:18 AM

#

Oh yeah that would make sense

#

She has entirely custom animations

keen ruin Feb 7, 2025, 10:18 AM

#

yeah then it doesn't surprise me at all if things get out of sync

#

the amount of code in here is insane

#

would be nice if there was an actual GitHub link instead of it linking to the skin registry mod lol

steep tartan Feb 7, 2025, 10:10 PM

#

idk how to explains this

#

Maneater AI just gave up chasing us

#

i don't have any logs with pathfinding trowing errors, but it was chasing us and just decided to stop in the middle of the corridor

#

we look at him and he was just stading there looking at us

#

we ran away and he didn't chase us

#

maneater has something in that AI man idk

tardy pivot Feb 7, 2025, 10:18 PM

#

were you on the other side of a powered door?

keen ruin Feb 7, 2025, 10:51 PM

#

steep tartan i don't have any logs with pathfinding trowing errors, but it was chasing us and...

chasing as in he screamed at you and charged?

#

the maneater is pretty complicated, I don't even fully understand how it's supposed to work yet

crimson jolt Feb 7, 2025, 10:54 PM

#

keen ruin chasing as in he screamed at you and charged?

IIRC it stalks people within a certain radius of its "hiding spot" up until it gets close enough to someone to lunge

#

and then once it lunges for the first time, it starts chasing the player with no more emphasis on stealth up until that player escapes the radius

#

i havent seen how the maneater behaves with pathfindinglagfix because personally i avoid it like the plague once it's transformed

#

and it's also pretty difficult to survive that first lunge even if you tried

steep tartan Feb 7, 2025, 10:57 PM

#

tardy pivot were you on the other side of a powered door?

Nope

steep tartan Feb 7, 2025, 10:58 PM

#

keen ruin chasing as in he screamed at you and charged?

Nope

#

My friend left the maneater alone and when we're getting back we saw him trying to approach, we freaked out and started running but he suddenly stopped chasing us and stood still on the corridor just looking at us

#

We didn't try to approach it but he wasn't moving

crimson jolt Feb 7, 2025, 10:59 PM

#

oh then it would still be trying to do the stealth behavior

keen ruin Feb 7, 2025, 10:59 PM

#

yeah probably

crimson jolt Feb 7, 2025, 10:59 PM

#

i think for the most part it's intended to work like how the bracken approaches players, moving towards them but only in a way where it can avoid crossing anybody's line of sight

keen ruin Feb 7, 2025, 11:00 PM

#

it's similar but it switches nodes more often

crimson jolt Feb 7, 2025, 11:00 PM

#

except it doesn't retreat when it is spotted, so it might be locking up instead?

#

idk

keen ruin Feb 7, 2025, 11:00 PM

#

the line of sight cubes are not very long

steep tartan Feb 7, 2025, 11:00 PM

#

The only mods I'm using that may alter maneater in some way is pathfinding lag fix and ButteryBalance

keen ruin Feb 7, 2025, 11:00 PM

#

if it was a long hall it makes sense it would stop in line of sight

crimson jolt Feb 7, 2025, 11:01 PM

#

steep tartan The only mods I'm using that may alter maneater in some way is pathfinding lag f...

i dont alter maneater pathfinding, just override its agent's angular speed and acceleration during a lunge

#

so i don't think i'm causing it here

keen ruin Feb 7, 2025, 11:01 PM

#

only thing I change related to the maneater is when it is selecting a target player

crimson jolt Feb 7, 2025, 11:01 PM

#

tbf it might not be pathfinding lag fix either, zaggy discovered a vanilla bug with the bracken while he was writing PLF patches, so it's possible some of his behavioral changes might be exposing something similar for vanilla maneater

keen ruin Feb 7, 2025, 11:02 PM

#

I don't think so in this case

#

at least not a related issue

steep tartan Feb 7, 2025, 11:03 PM

#

Ok then

#

The best way I can describe what happened to the maneater was the same thing as that video I sended a long time ago

#

He gave up

#

Really strange why he does that sometimes

keen ruin Feb 7, 2025, 11:05 PM

#

like buttery mentioned, it has a certain sphere around a cave node (or a random node if not mineshaft) that it wants to "protect" by sneaking around near the player

#

if you leave that sphere, then it will stop following you afaik

cursive sonnet Feb 7, 2025, 11:08 PM

#

Does it's nest get bigger when at least 1 person is inside

keen ruin Feb 7, 2025, 11:08 PM

#

something like that, yea

steep tartan Feb 7, 2025, 11:09 PM

#

Well, that must been the case

crimson jolt Feb 9, 2025, 2:12 PM

#

@keen ruin something i meant to ask but maybe didnt phrase very well

#

how does pathfindinglib handle changing destination each frame?

#

i optimized my masked patch a bit and now im pretty confident masked never CalculatePath more than once a frame (and if they do it's a bug i can probably fix)

#

but their destination changes depending on a couple of factors

#

my understanding is that pathfindinglib doesnt guarantee a path will be found in a single frame, since it's designed to be async, so what happens if it's still looking for a path to an old destination and i need to feed it a new one

keen ruin Feb 9, 2025, 2:28 PM

#

PathfindingLagFix doesn't handle destinations, those are all done by Unity

#

they are already done in slices, so they don't complete in one frame normally unless you modify the budget

#

the agents stop moving while they are calculating a path

green gust Feb 9, 2025, 2:30 PM

#

I assume its not noticeable because of a couple things like it being a small amount of time and acceleration carrying an already moving agent

crimson jolt Feb 9, 2025, 2:34 PM

#

keen ruin they are already done in slices, so they don't complete in one frame normally un...

ah okay

#

so then there should be minimal behavioral *difference anyway, then

keen ruin Feb 9, 2025, 2:38 PM

#

yeah it doesn't sound any different than what vanilla enemies do

crimson jolt Feb 10, 2025, 4:13 PM

#

out of curiosity

#

do you know why the bracken acts weird when it first enters anger mode

#

and does that need fixing + is it within scope of fixing in pathfindinglagfix

crimson jolt Feb 10, 2025, 4:14 PM

#

crimson jolt do you know why the bracken acts weird when it first enters anger mode

for some reason it kind of turns away from players and runs away for just a brief moment, and then immediately starts chasing

#

it's always felt kind of awkward and i doubt it's intentional

#

in the absence of having clips recorded, i tried to find some instances of it online, if you want to see what i mean:
https://youtu.be/A7A36i0v9yA?t=213
https://youtu.be/A7A36i0v9yA?t=285
https://youtu.be/A7A36i0v9yA?t=330
https://youtu.be/A7A36i0v9yA?t=358

#

turns out alter ego's video had 4 separate instances that were pretty clear

green gust Feb 10, 2025, 4:34 PM

#

what im guessing is happening is this:

Bracken has a spot it's moving towards while facing the player
Player forces it to switch to Threaten state.
Bracken gets the data a bit delayed so it's still moving, but no longer looking at player.
This makes it look like it moves away for a bit since it's destination hasn't been set to targetPlayer yet.

crimson jolt Feb 10, 2025, 4:36 PM

#

yeah that's what i had in mind too

#

i think it would make more sense if it were to freeze until a path gets calculated

#

because right now it looks goofy

#

and it ruins the tension a bit

green gust Feb 10, 2025, 4:37 PM

#

yeah that makes sense, im just trying to figure out where he sets the destination rn to confirm this lol

#

probably part of AvoidClosestPlayer

keen ruin Feb 10, 2025, 4:58 PM

#

my guess is that it avoids the player, but when it wants to do the staredown, it sets agent speed to 0 and doesn't reset the destination, then when staredown transitions to aggro, it sets agent speed to non-zero and then sets the agent's destination, but setting the destination must not be resetting the previous path in time to stop it moving

#

I'm not sure if that's what Xu was trying to describe, but it seems to make the most sense to me that that is the reason

#

that said, I'm not sure why it wouldn't be resetting the path when setting the destination to the player, since it does reset the path when setting destination to a new path while a partial path is calculated

#

and it also seems to reset it when the bracken is pathing away from the player and is "discovering" new pathable nodes along the way, due to its distance cap

#

but I don't know what else it would be

crimson jolt Feb 10, 2025, 5:00 PM

#

if (!this.carryingPlayerBody && this.evadeModeStareDown && this.evadeStealthTimer < 1.25f)
{
    this.AddToAngerMeter(Time.deltaTime * 1.5f);
    this.agent.speed = 0f;
}

it looks like when it's staring down players (the part that immediately precedes its anger) it does just set speed to 0 yeah

#

doesn't change destination

keen ruin Feb 10, 2025, 5:00 PM

#

yeah, I didn't figure it would

crimson jolt Feb 10, 2025, 5:01 PM

#

i doubt it sets its destination immediately when entering phase 2

keen ruin Feb 10, 2025, 5:01 PM

#

two solutions are possible:

calculate path ahead of time, set it in the agent (I haven't tested if this works effectively)
set the speed to 0, then wait until the agent's path is complete, then set it to non-0

crimson jolt Feb 10, 2025, 5:01 PM

#

so i assume it's just the delay xu mentioned, where it's moving with "anger mode" speed towards its retreat point until it can set its target to the player

keen ruin Feb 10, 2025, 5:01 PM

#

oh you may be right

#

where does it set the speed back to non-zero?

crimson jolt Feb 10, 2025, 5:02 PM

#

look like it's here

#

if (this.stunNormalizedTimer > 0f)
{
    this.creatureAnimator.SetLayerWeight(2, 1f);
    this.agent.speed = 0f;
    this.angerMeter = 6f;
}
else
{
    this.creatureAnimator.SetLayerWeight(2, 0f);
    this.agent.speed = Mathf.Clamp(this.agent.speed + Time.deltaTime * 1.2f, 3f, 12f);
}

#

this is phase 2 in Update()

#

if it's not stunned, it starts accelerating, and is forced to be 3+

keen ruin Feb 10, 2025, 5:03 PM

#

hm, looks like entering anger mode also immediately jumps it to 9?

#

in EnterAngerModeClientRpc()

crimson jolt Feb 10, 2025, 5:03 PM

#

oh yeah

#

weird

#

maybe the acceleration was too slow

#

not sure why he put that in the client RPC, but the bracken was the first AI he wrote for the game I believe

#

so maybe he hadn't entirely worked out the navmeshagent owner relationship yet

keen ruin Feb 10, 2025, 5:04 PM

#

hmm, yea, that is a little funny

#

who knows

crimson jolt Feb 10, 2025, 5:06 PM

#

well

#

i was going to say

#

i think what's happening is Update() is changing the bracken's speed immediately

#

but it looks like the player isn't targeted until DoAIInterval()

#

so if the bracken had an extremely slow interval, it would obviously cause some weirdness there

keen ruin Feb 10, 2025, 5:06 PM

#

this block of code should run every time it switches to anger state, so you could possibly use this to either

set the destination to transform.position
set the destination to the player that DoAIInterval() will eventually set it to

crimson jolt Feb 10, 2025, 5:07 PM

#

but the bracken uses an AI interval of 0.2

keen ruin Feb 10, 2025, 5:07 PM

#

yeah

crimson jolt Feb 10, 2025, 5:07 PM

#

and it looks like it "wanders" a bit longer than 0.2s

#

at least to me

keen ruin Feb 10, 2025, 5:07 PM

#

it is undoubtedly the delay from the interval

#

well

#

do you mean in your experience with PathfindingLagFix or generally?

crimson jolt Feb 10, 2025, 5:08 PM

#

well, i suppose i haven't seen it happen since the beta

#

just because im not usually in a situation where i could see it or paying attention to it

keen ruin Feb 10, 2025, 5:08 PM

#

doubtful that it would've changed, if anything it would be more delayed, but unlikely

crimson jolt Feb 10, 2025, 5:08 PM

#

but it defo happens in vanilla and happened in the previous PLF (before v2)

keen ruin Feb 10, 2025, 5:09 PM

#

200ms can feel pretty long though, idk if maybe you're underestimating how long that is

green gust Feb 10, 2025, 5:09 PM

#

crimson jolt and it looks like it "wanders" a bit longer than 0.2s

i'd take acceleration into account with that 0.2 seconds

keen ruin Feb 10, 2025, 5:09 PM

#

oh you mean because the agent has inertia?

green gust Feb 10, 2025, 5:09 PM

#

which would probably extend it by a bit

crimson jolt Feb 10, 2025, 5:09 PM

#

oh that is fair

green gust Feb 10, 2025, 5:09 PM

#

i mean the agent just doesnt brake on its own

crimson jolt Feb 10, 2025, 5:09 PM

#

bracken has really slow angular speed and acceleration

green gust Feb 10, 2025, 5:09 PM

#

keen ruin Feb 10, 2025, 5:10 PM

#

huh

green gust Feb 10, 2025, 5:10 PM

#

acceleration is actually high, (i havent seen if he changes it via code), but yeah angular speed isnt too high and it doesnt brake

#

default values ignoring any code changes since i know he does a bit

crimson jolt Feb 10, 2025, 5:10 PM

#

angular speed is decreased to 50 when he's dragging a body

#

acceleration is static

green gust Feb 10, 2025, 5:11 PM

#

i would say to turn acceleration really high, put auto braking to on, and see what happens

keen ruin Feb 10, 2025, 5:11 PM

#

I feel like that might only result in it being slightly quicker

#

also to answer your question about whether PathfindingLagFix might do anything to that, it's unlikely

#

if it had been the sliced path calculation delaying the change in direction, possibly, but I doubt it is actually

green gust Feb 10, 2025, 5:12 PM

#

seems like a logical issue rather than a pathfinding one, if its even related to the navmeshagent

keen ruin Feb 10, 2025, 5:12 PM

#

yeah

crimson jolt Feb 10, 2025, 5:12 PM

#

i think maybe it could be "fixed" by checking if evadeModeStareDown is true when entering state 2 from other states, and if it is, immediately set destination to the agent's position

#

and you'd need to also check within ~0.2s of the bracken taking damage, since that maxes out its anger immediately

keen ruin Feb 10, 2025, 5:13 PM

#

I mean if it's entering the anger behavior you know that it should have a player destination regardless of whether it came from staredown

#

if you want to make it somewhat retain the vanilla "reaction time" that the interval causes, though, you could check that and set destination to self

crimson jolt Feb 10, 2025, 5:13 PM

#

true, but i just figured the least intrusive fix would be to freeze it in place so it still has a delay before acquiring target

crimson jolt Feb 10, 2025, 5:14 PM

#

keen ruin I mean if it's entering the anger behavior you know that it should have a player...

although this is probably what's intended

keen ruin Feb 10, 2025, 5:14 PM

#

oh oopsie that's what you said

crimson jolt Feb 10, 2025, 5:14 PM

#

i suppose

keen ruin Feb 10, 2025, 5:14 PM

#

I think

green gust Feb 10, 2025, 5:14 PM

#

since its the first AI i would also just look into EnemyAI itself, i know there's some fields that if he populated would just mess with the destination on it own separate from what he's doing/included

keen ruin Feb 10, 2025, 5:14 PM

#

crimson jolt and you'd need to also check within ~0.2s of the bracken taking damage, since th...

not sure what you mean by this though?

green gust Feb 10, 2025, 5:14 PM

#

there's like a movingTowardsDestination thing and similar

keen ruin Feb 10, 2025, 5:15 PM

#

I'm not sure if setting movingTowardsDestination to false causes it to stop moving though?

#

hmm

crimson jolt Feb 10, 2025, 5:15 PM

#

keen ruin not sure what you mean by this though?

in the last clip i sent, you can see that whacking the bracken while it's actively retreating doesn't make it stop retreating

green gust Feb 10, 2025, 5:15 PM

#

i dont even know if the bracken ever turns that on to true, since he has his own movingTowardsTargetPlayer thing iirc

crimson jolt Feb 10, 2025, 5:15 PM

#

crimson jolt in the last clip i sent, you can see that whacking the bracken while it's active...

it keeps moving in the same direction, but it speeds up and stops facing the player, up until it targets the player and starts chase

keen ruin Feb 10, 2025, 5:16 PM

#

hmm, I see, so the speeding up is what's weird about that

crimson jolt Feb 10, 2025, 5:16 PM

#

i guess now that i think about it, you don't even need to check what happened to the bracken

#

if it entered anger mode you can just freeze it or immediately set a target to a player

keen ruin Feb 10, 2025, 5:16 PM

#

it doesn't sound that weird but I suppose it's a little odd if it accelerates

crimson jolt Feb 10, 2025, 5:16 PM

#

doesn't matter if it was from staredown or being attacked

keen ruin Feb 10, 2025, 5:16 PM

#

yeah, in theory that could look okay

crimson jolt Feb 10, 2025, 5:16 PM

#

keen ruin it doesn't sound _that_ weird but I suppose it's a little odd if it accelerates

is it weird?

keen ruin Feb 10, 2025, 5:16 PM

#

especially if it retains its momentum

crimson jolt Feb 10, 2025, 5:16 PM

#

once it enters anger mode it immediately sets its speed to 9

keen ruin Feb 10, 2025, 5:17 PM

#

I mean if it accelerates along its previous path and then turns afterwards

crimson jolt Feb 10, 2025, 5:17 PM

#

oh

keen ruin Feb 10, 2025, 5:17 PM

#

no I don't mean implementation-wise, I mean from the player's perspective

crimson jolt Feb 10, 2025, 5:17 PM

#

ahh

#

yeah

#

that's why i was thinking it'd feel better if the bracken stopped in place

keen ruin Feb 10, 2025, 5:17 PM

#

as long as it decelerates naturally, yeah

#

if it stops dead then I don't think that would be better

crimson jolt Feb 10, 2025, 5:18 PM

#

well, i could be wrong, but i think setting its destination to its own position wouldn't stop it from "overshooting"

green gust Feb 10, 2025, 5:18 PM

#

crimson jolt well, i could be wrong, but i think setting its destination to its own position ...

thats why i was thinking of just amping acceleration and auto bracking momentarily

crimson jolt Feb 10, 2025, 5:18 PM

#

it would still decelerate and then move back towards its original position if it overshot too far

#

but by the time it thinks about doing that it should've already acquired a player target probably

green gust Feb 10, 2025, 5:18 PM

#

i dont know if bracken uses it but zeekerss also has a thing to overshoot destination

crimson jolt Feb 10, 2025, 5:19 PM

#

but also, now that i think about it, if it were to decelerate naturally, it's probably not an issue to just set it to target players immediately

green gust Feb 10, 2025, 5:19 PM

#

wacky stuff

keen ruin Feb 10, 2025, 5:20 PM

#

crimson jolt but also, now that i think about it, if it were to decelerate naturally, it's pr...

yeah, I think if it slows and then turns around it makes sense for it to switch immediately

#

especially since the path may take a few frames to calculate

green gust Feb 10, 2025, 5:21 PM

#

green gust wacky stuff

i wonder when MoveTowardsDestination is set to false, if ever

#

i dislike this part of EnemyAI's movement logic

keen ruin Feb 10, 2025, 5:24 PM

#

green gust wacky stuff

oh is the add player velocity to destination thing how the masked strafes?? I figured it was some offset of the position in Update() or something

#

I guess it makes sense that a destination offset could work though

green gust Feb 10, 2025, 5:25 PM

#

no idea, i dislike the movement logic, i stopped using SetDestinationToPosition on my newer enemies and it caused them to have a bug where they exclusively moved to Vector3.zero because he sets teh destination in update if MoveTowardsDestination is true, which it is on start

keen ruin Feb 10, 2025, 5:25 PM

#

huh

#

was there a reason you needed to stop using SetDestinationToPosition?

green gust Feb 10, 2025, 5:27 PM

#

to use my agent script with my enemies and non-enemies without making it rely on enemyai

crimson jolt Feb 10, 2025, 5:27 PM

#

keen ruin oh is the add player velocity to destination thing how the masked strafes?? I fi...

i think the masked just always moves directly towards the player

#

but is forced to face their direction

#

though i didn't actually look into it

#

sort of like how the bracken always faces players while retreating, so it'll sidestep and stuff

keen ruin Feb 10, 2025, 5:29 PM

#

crimson jolt i think the masked just always moves directly towards the player

no they don't, they will sometimes follow the player's movement when they strafe to prevent them from dodging the masked

#

you can test this by strafing side to side, the masked will do so as well

#

okay, apparently that function is always on, it's sometimes felt like the masked can't do it but now I'm not sure why that is

#

Hmm

crimson jolt Feb 10, 2025, 5:31 PM

#

hmm i will check

keen ruin Feb 10, 2025, 5:39 PM

#

green gust to use my agent script with my enemies and non-enemies without making it rely on...

oh I see, that makes sense

keen ruin Feb 12, 2025, 8:41 PM

#

looking at optimizing the blob physics and I just found out that the blob changing shapes just straight up doesn't work properly when your fps is too high

#

Wowhat

keen ruin Feb 12, 2025, 11:36 PM

#

@crimson jolt free patch for you

    [HarmonyPrefix]
    [HarmonyPatch(nameof(BlobAI.FixedUpdate))]
    private static bool FixedUpdatePrefix(BlobAI __instance)
    {
        return false;
    }

    [HarmonyPostfix]
    [HarmonyPatch(nameof(BlobAI.Update))]
    private static void UpdatePostfix(BlobAI __instance)
    {
        if (!__instance.ventAnimationFinished)
            return;
        for (int i = 0; i < __instance.SlimeBonePositions.Length; i++)
        {
            if (Vector3.Distance(__instance.centerPoint.position, __instance.SlimeBonePositions[i]) > __instance.distanceOfRaysLastFrame[i])
                __instance.SlimeBones[i].transform.position = Vector3.Lerp(__instance.SlimeBones[i].transform.position, __instance.SlimeBonePositions[i], 10f * Time.deltaTime);
            else
                __instance.SlimeBones[i].transform.position = Vector3.Lerp(__instance.SlimeBones[i].transform.position, __instance.SlimeBonePositions[i], 5f * Time.deltaTime);
        }
    }

#

as an explanation

#

It turns out that the blob's rigidbodies for the edges of the goop never move unless there are multiple fixed updates in a frame, apparently because the blob needs the transforms to be synced before it will make any movement happen

Stack Overflow

Unity, Rigidbody.MovePosition() is not working if transform.positio...

Question
This code works. ↓
private void FixedUpdate()
{
Vector3 newPosition = new Vector3(0, 1, 0);
GetComponent<Rigidbody>().MovePosition(newPosition);
}

This code doesn't work. ↓

#

it's better for this to happen in Update() anyway, since that results in much smoother movement, given that the rigidbodies aren't set to interpolate

#

looks like the blob dealing and receiving damage works correctly as well, so no worries about triggers being weird or anything it seems like

crimson jolt Feb 12, 2025, 11:53 PM

#

keen ruin It turns out that the blob's rigidbodies for the edges of the goop never move un...

lol

#

so does that mean the blob is just completely broken at 51+ fps

keen ruin Feb 12, 2025, 11:54 PM

#

yeah basically

#

gets progressively less responsive to terrain as fps approaches 50

#

I just thought it was normal for blobs to clip through walls all the time lmao

#

then I realized just how bad it was sometimes today

#

oh lemme edit, I should've made it lerp from Transform.position instead of Rigidbody.position

#

didn't seem to make a difference in practice, presumably because the transforms are synced from the previous frame fairly consistently, but it might make it move more consistently each frame to do that

#

it's obviously slightly unfortunate not to use the existing code from FixedUpdate in this patch, but in theory nobody's gonna be transpiling FixedUpdate unless they're trying to fix the same thing this is fixing

#

and the call to MovePosition needs to be changed to an assignment to transform.position, I've never tried to transpile a reverse patch

crimson jolt Feb 13, 2025, 2:29 AM

#

keen ruin <@235439745486815233> free patch for you ```cs [HarmonyPrefix] [Harmony...

noticing a huge improvement after implementing this

#

neat find

#

i assume you are cool with me releasing it? given the context of the message

#

credited, of course

keen ruin Feb 13, 2025, 2:33 AM

#

yeah go for it

#

it's not something I can put in PathfindingLagFix since it changes the damage area

crimson jolt Feb 13, 2025, 2:36 AM

#

yea

#

super obvious on catwalks especially

#

it's a cool effect

#

before/after

tardy pivot Feb 13, 2025, 10:14 AM

#

should this blob thing have a post in https://discord.com/channels/750645598293590077/1258564990533636186 too maybe?. this seems something that zeekers could fix relatively easily if he wanted.

keen ruin Feb 13, 2025, 2:48 PM

#

I could probably post about it at some point yeah

crimson jolt Feb 13, 2025, 8:16 PM

#

i think it'd be a good post

#

it's a relatively simple fix and it is a huge improvement

keen ruin Feb 16, 2025, 5:17 AM

#

lil preview of a blob optimization

#

basically just doing one edge distance calculation per frame instead of 8

#

doesn't seem noticeable in my testing, despite the fact that it includes the player collision test

lone bronze Feb 25, 2025, 8:16 PM

#

i dont know if this a bit off topic but why does SOMETIMES when i play with friend the enemies do not see players?? like nutcrackers, barbers, butlers spiders, and bugs going into 1 random place in the map and sitting there also coilhead and brackens are stuck near the vents and do not move, i have around 230 mods so i dont know WHAT can cause it(basically enemies does not know we are inside) maybe anyone had this issue too or know what may cause it?

keen ruin Feb 26, 2025, 2:22 AM

#

hmm, not really sure what would cause that either

PathfindingLagFix does patch how targeting is done, but I haven't observed any issues with it nor gotten any reports about issues from others

#

I don't know what types of mods might cause that to start happening, unless it's failure to target players that join after the 4th player with MoreCompany, though I haven't seen issues like that in a while either

keen ruin Feb 26, 2025, 2:39 AM

#

if you do get a situation like that though, it would be good to grab logs from the host and whatever client was not being targeted, and feel free to send them my way in case anything interesting shows up

#

in fact, preferably grab Player.log from %appdata%\..\LocalLow\ZeekerssRBLX\Lethal Company\ after such a bug happens, then it will have logs from the pathfinding threads if there are any errors there

coarse hawk Feb 26, 2025, 5:18 AM

#

lone bronze i dont know if this a bit off topic but why does SOMETIMES when i play with frie...

you basically telling us to guess what 230mods you have Fish

median patrol Mar 1, 2025, 4:14 AM

#

not sure if this is a vanilla bug or a Pathfinding Lag Fix bug but whenever you have an Snare Flea chasing you and you jump it simply forgets about you

keen ruin Mar 1, 2025, 5:06 AM

#

that's vanilla, if you are the only one inside

#

it will give up the chase if it can't target any players inside, and when you jump it is unable to path to you

crimson jolt Mar 1, 2025, 5:14 AM

#

it affects a few other enemies although not quite as majorly

#

bracken locks up when you jump during a chase, but doesn't immediately deaggro

#

maneaters get hella confused mid-lunge

spring prairie Mar 1, 2025, 9:52 AM

#

Alien life in LC’s universe all live in an ecosystem where nobody jumps. If you jump they get confused and/or scared.

keen ruin Mar 1, 2025, 6:59 PM

#

factual information actually

green gust Mar 1, 2025, 7:07 PM

#

yeah because then you're not routeable while jumping so they cant go to your destination DogeKEK

keen ruin Mar 3, 2025, 11:02 PM

#

about time I released this lol

Version 2.2.0 (Beta)

Patched masked to use the cached elevator controller instead of finding it fresh every AI interval.
Patched the method to find the main entrance to use a cache and avoid a FindObjectsOfType call when masked are pathing to it.
Patched hygroderes to reduce the physics casts per frame from 8 to 1.
Patched the maneater AI to avoid synchronous pathfinding when the adult is sneaking and when it checks if it is stuck.
Patched coilheads' player targeting pathfinding to run asynchronously.

#

surely I have not broken anything

crimson jolt Mar 3, 2025, 11:12 PM

#

alright

ashen harbor Mar 4, 2025, 5:30 AM

#

keen ruin about time I released this lol ### Version 2.2.0 (Beta) - Patched masked to use ...

I'm sure it'll be fine, regardless I'll be running this for the next stream so if things explode you'll be sure to find out x)

keen ruin Mar 4, 2025, 5:49 AM

#

sounds good teehee

ashen harbor Mar 5, 2025, 5:13 AM

#

keen ruin sounds good<:teehee:1223888883201343509>

So I ran into a freeze once today but dunno if it relates cus I never reprod it, however I did notice a bug with Masked when they were around on the Galetry, they would be missing the MoreCompany cosmetics of people they were mimicing for some reason

#

Which 100% would be something from the PathdinfindingLagFix Beta somehow

keen ruin Mar 5, 2025, 6:08 AM

#

kinda seems unlikely that cosmetics not spawning could be caused by the patches I made, since they only affect DoAIInterval

#

if they're patching that to spawn cosmetics I would be shocked

#

freeze is curious, but I think unlikely to be anything I've done here? I don't recall any suspicious loops but maybe I goofed something subtle

#

if it happens multiple times I would wonder if downgrading removes the problem, but if it only froze once then you probably need to use it at least 5 times with beta and 5 times with stable to ensure it wasn't a coincidence for it to stop when you downgrade

#

something happening only once during a play session and then switching something out isn't gonna be conclusive

ashen harbor Mar 5, 2025, 6:38 AM

#

keen ruin something happening only once during a play session and then switching something...

We never did swap from the beta back to stable after it happened

#

just rebooted

keen ruin Mar 5, 2025, 6:38 AM

#

right, I kinda figured that's what you meant

#

I just meant that if you end up encountering it again and want to test without beta, would be good to make sure you play a good number of sessions to confirm that it's not something else

ashen harbor Mar 5, 2025, 6:39 AM

#

Yeah

#

I'm gonna try to repro masked being weird on Galetry with some other things I added disabled

#

reprod

keen ruin Mar 5, 2025, 6:41 AM

#

sounds goog

ashen harbor Mar 5, 2025, 6:45 AM

#

keen ruin sounds goog

Okay I confirmed it's somehow PathfindingLagFix

#

with it enabled masked that spawn at the Galetry have no cosmetics, but with the Beta disabled or the Stable version present they're fine

#

I used the Twitch API to spawn it lol

#

It was bald when I had the Beta enabled

#

XD

#

Idk why it's specifically the Galetry though

#

I'll reenable it and try sending a log cus there were no Stack Traces

#

I'd assume it's causing some odd conflict with Mirage somewhere though but why I noticed it mainly on Galetry is odd

#

But yeah with Beta enabled the cosmetics vanish

#

Lol

#

@keen ruin

📎 LogOutput.log

#

01956515-f83f-afc2-3e70-e2a7cb7a86a6

cold peak Mar 5, 2025, 6:57 AM

#

[06:51:19.1878346] [Error  : Unity Log] Error when initializing enemy variables for MaskedPlayerEnemy(Clone) : System.NullReferenceException: Object reference not set to an instance of an object
  at PathfindingLagFix.Patches.PatchFindMainEntrance.FindMainEntrancePositionPrefix (System.Boolean getTeleportPosition, System.Boolean getOutsideEntrance, UnityEngine.Vector3& __result) [0x00010] in ./Patches/PatchFindMainEntrance.cs:87 
  at RoundManager.FindMainEntrancePosition (System.Boolean getTeleportPosition, System.Boolean getOutsideEntrance) [0x0000d] in <bdf6a080e98a49fd84b92b24894f768c>:IL_000D 
  at MaskedPlayerEnemy.Start () [0x000be] in <bdf6a080e98a49fd84b92b24894f768c>:IL_00BE

ashen harbor Mar 5, 2025, 6:58 AM

#

cold peak ``` [06:51:19.1878346] [Error : Unity Log] Error when initializing enemy variab...

Ah yeah I never caught any stack traces

keen ruin Mar 5, 2025, 6:58 AM

#

ty diff I was just lookin at that

ashen harbor Mar 5, 2025, 6:58 AM

#

That makes sense cus Galetry has no main entrance

keen ruin Mar 5, 2025, 6:58 AM

#

I did a silly

#

yeah

ashen harbor Mar 5, 2025, 6:59 AM

#

So that error goes off which probably then prevents the cosmetics and stuff from initializing lol

green gust Mar 5, 2025, 6:59 AM

#

keen ruin I did a silly

https://tenor.com/view/silly-cat-silly-car-car-stare-10-thousand-yard-stare-10-thousand-yard-gif-14200271775968563996

Tenor

#

banned

keen ruin Mar 5, 2025, 7:07 AM

#

just pushed 2.2.1 Beta to fix dat

storm lava Mar 11, 2025, 12:42 AM

#

Should I be using the beta or the normal pathfindinglagfix mod?

crimson jolt Mar 11, 2025, 12:53 AM

#

the beta is not guaranteed to be stable

#

it gets updates earlier and can be used if you dont mind potential need for troubleshooting

#

the normal mod is less prone to issues because it gets the releases after they've been in beta for a short while

keen ruin Mar 11, 2025, 1:02 PM

#

oh speaking of the beta, @ashen harbor have you been using beta 2.2.1? wondering if it has been fine on galetry since then

#

if so, I'll push to stable, I've just been busy with PathfindingLib API and haven't been able to specifically test it really thoroughly

ashen harbor Mar 11, 2025, 1:42 PM

#

keen ruin oh speaking of the beta, <@165179734194520073> have you been using beta 2.2.1? w...

Oh yeah I haven't had any issues and have been using the beta

keen ruin Mar 11, 2025, 1:45 PM

#

yay nice!

#

I'll push it later today then if I remember

storm lava Mar 13, 2025, 1:59 AM

#

01958d38-7331-b245-2a97-54b29fabdf06
Enemies on certain moons seemed pathfind weirdly. Truth be told, we only went on generic's moons this session but I still figured I'd let whomever know. I had a thumper, a hoarding bug, a snare flea, and even a rollinggiant just pathfind toward something but they didn't move as fast and like stuttered when doing so

📎 LogOutput.log

keen ruin Mar 13, 2025, 6:44 AM

#

a video would be immensely helpful if you have one

storm lava Mar 14, 2025, 5:14 AM

#

If you scroll up in helpandtroubleshooting, you'll find the exact collection of mods (disabled and enabled) plus the correct log for that session

#

Also, I noticed this whenever I saw a snare flea bugging out in the logs

#

let me get it rq

#

[Debug :PathfindingLagFix] Player paths from Centipede(Clone) (CentipedeAI, -495494) are 55594.64ms old, which is more than twice 140ms, using synchronous paths.

#

It went toward me, then went away, then went toward me

#

It was as if it couldn't decide on whether or not to target me

#

This has happened to a thumper, a lord of the manor

#

The hoarder bugs idk. They seem to just stutter walk

#

And I'm not 100% but it seems this happened only only generic's moons. Mostly collition but also icebound. I'm not 100% sure if it's just tied to those moons but that is where I've been routing to, to see

ashen harbor Mar 14, 2025, 5:19 AM

#

Considering I've never ran into it it's prolly a moon or interior issue

crimson jolt Mar 14, 2025, 5:19 AM

#

usually those error messages print because something else is causing excessive frame lag

#

it is usually a symptom and not a cause

#

but that clip is definitely strange

#

doesn't feel like framerate is low enough that it should be causing that much jank

storm lava Mar 14, 2025, 5:20 AM

#

the framerate wasn't low at all, it most likely was just 180 or around there. The lag in the video was cuz I used snippit tool. Never used it before so idk if I did it right

#

I've spent all day disabling mods trying to figure out the cause

#

I hope something is found, that's like all the info I can provide

crimson jolt Mar 14, 2025, 5:23 AM

#

my guess is it's interior related

storm lava Mar 14, 2025, 5:23 AM

#

This has happened on multiple interiors

crimson jolt Mar 14, 2025, 5:23 AM

#

i see

storm lava Mar 14, 2025, 5:23 AM

#

Also I meant colateral. the scrap moon

ashen harbor Mar 14, 2025, 5:24 AM

#

It's prolly an issue with the moon then

#

I've seen Spiders tweak out specifically on Infernis before lol

#

But they're fine everywhere else

storm lava Mar 14, 2025, 5:24 AM

#

I asked generics but noone seemed to find an issue with the monsters on the two moons I listed

#

even said that it isn't generics that is doing it though I didn't provide a log so they was assuming ig

ashen harbor Mar 14, 2025, 5:25 AM

#

Could just be they didn't notice it

storm lava Mar 14, 2025, 5:26 AM

#

So the moon itself can mess with the pathfinding of monsters? Cuz again, a lot of monsters would work then occasionaly it would be as if a monster kept deciding over and over to chase me. I remember a thumper going the opposite way of me. Even a coilhead was going in the wrong direction every time I looked away from it

#

Plus, I didn't have this problem with that moon by itself. I went there with just imperium and checked. And yes this has happened without imperium so I don't believe imperium would be the cause

#

Well anyways, nobody is required to look at everything I gave but I ask someone please do. I have been stuck on this last night and all day today. I just can't find the cause and it's very demotivating. Thank to whoever has looked at it so far.

ashen harbor Mar 14, 2025, 5:32 AM

#

I mean I've been using PathfindingLagFix for a while and never ran into this, however I've had some occasions where all clients freeze from masked enemies and I've been unable to trace the culprit of that, currently we're pretty sure it's a bug with Mirage based on an error I found

#

They all freeze and then I get lag as host that makes even my stream drop frames cus something freaks out in terms of networking

storm lava Mar 14, 2025, 5:34 AM

#

I haven't ran into any masked issue so far but idk if it's the norm but I'm using pathfindinglagfix beta. seems more updated. I don't know if that isn't recommended

ashen harbor Mar 14, 2025, 5:34 AM

#

That's what I'm using and I've still ran into it

storm lava Mar 14, 2025, 5:37 AM

#

Hmm. Well in short it just feels like monsters can't target me properly or something sometimes. I had a bunch of snare fleas barely move from their spot when they fall because they stutter when they move. Also had some go away from me, then back at me, then away like their couldn't make their mind. I am just know remembering something that sounds very familiar with a forest giant on assurance. It happened like a week ago. It just stuttered walked toward me.

#

That could be a different reason entirely but that 100% did happen now that i'm recalling it

storm lava Mar 14, 2025, 7:36 PM

#

Anybody know what it could be the problem? Just had a spider and a hoarder bug run away from me

keen ruin Mar 15, 2025, 3:23 PM

#

really not sure, I also haven't observed that I don't think

#

I haven't had time to look at this for the last few days, but hopefully tomorrow or Monday I can

#

if you happen to have this happen early on a day, a seed and profile code might help so I can try to reproduce it

#

especially if it happens to you in singleplayer

keen ruin Mar 15, 2025, 3:27 PM

#

ashen harbor I mean I've been using PathfindingLagFix for a while and never ran into this, ho...

did Mirage have an update that you suspect caused this regression? what's the error? I would have to have a look at my patches again to make sure that couldn't be related to them, seems unlikely but possible

storm lava Mar 15, 2025, 7:36 PM

#

keen ruin if you happen to have this happen early on a day, a seed and profile code might ...

All these weird issues was on generic moons specifically. Most of this happened on collateral. Though the 1 time I noticed a giant sttuter walk it was on assurance near the entrance. I was standing at the door, it saw me, went toward me but that giant pipe thing was above it. Then it just slowly stutter walked toward me

keen ruin Mar 15, 2025, 7:47 PM

#

does this happen every time on generic moons you think?

#

I can try to repro with just that if so, but if it's intermittent, or especially if it only happens in multiplayer, that makes it very likely in my mind that it won't happen for me too easily

#

(without what I mentioned above at least)

storm lava Mar 15, 2025, 7:59 PM

#

Yes, I just came to the conclusion it was just generics moons. it was happening without anything else enabled in my modpack. I mentioned it to generic and they did notice a few of their monsters acting wonky so they said they'll message wesley about it

keen ruin Mar 15, 2025, 8:07 PM

#

ah I see, hmm

#

and that's even without PathfindingLagFix then?

#

I'm not sure what would cause the AI to stutter step while roaming unless the AI nodes themselves are missing or outside the playable area

#

and I believe you mentioned some AI that would be only roaming

storm lava Mar 15, 2025, 8:15 PM

#

It was weird. I noticed snare fleas like stuttering toward me when they fall but barely move, hoarding bugs stutter walking, enemies going aggresive but going away from me instead of toward me. And enemies that seemed like they where in fast sessions going from toward me to away from me

keen ruin Mar 17, 2025, 8:38 AM

#

hmm, you didn't mention if this was happening without PathfindingLagFix too? I would like to avoid trying to investigate this if it's not a bug caused by it, since I want to focus on the elevator pathing API

#

I'm gonna be working on that today if this doesn't turn out to be my issue

storm lava Mar 17, 2025, 6:26 PM

#

well the giant stutter thing happened with it but after an entire day I just disabled all mods and went to generic moons and it was still happening. I didn't know what I was which is why I was trying to find answers on it

storm lava Mar 18, 2025, 2:06 AM

#

0195a6ff-0d47-58d7-653f-69303c1bf0ff
Was on titan in the ice/snow version of the Rajigar Mines interior on Titan. A bracken was sliding instead of moving and when looking it at it just didn't move for a bit. Eventually it did though

📎 LogOutput.log

#

#

keen ruin Mar 18, 2025, 7:10 PM

#

bracken sliding around is an Imperium bug, you have to jump to fix them after they spawn

#

at least if you are solo

storm lava Mar 19, 2025, 12:30 AM

#

Okay I'm glad on that. I got worried it was an interior issue.

thin solstice Mar 21, 2025, 6:36 PM

#

I have been noticing a thing in my modpack lately where Butlers are opening doors much faster than in vanilla, does what AI this mod is set to use impact this?

crimson jolt Mar 21, 2025, 6:40 PM

#

well

#

butlers normally only take 2 seconds to open doors

#

the reason it usually takes longer is just because their colliders dont overlap correctly sometimes

#

mineshaft doors are notoriously difficult for butlers to open for example

thin solstice Mar 21, 2025, 6:47 PM

#

Respectfully, is that a yes or no?

#

I am playing my modpack and a friend with probably over 100 vanilla Butler kills cannot shut up about Butlers constantly going through doors like they are nothing

#

I would like to confirm if a mod is causing this

crimson jolt Mar 21, 2025, 7:30 PM

#

i am doubtful it is this mod

#

what im trying to say is that butlers taking a long time to open doors is unintended behavior and basically happens by coincidence

#

so expecting it to be consistent in the first place is probably misguided

crimson jolt Mar 21, 2025, 7:35 PM

#

crimson jolt i am doubtful it is this mod

this mod changes how enemies calculate paths, but (and zaggy can correct me if im wrong) should have no effect on their spacing

#

if a butler wants to walk through a door they are going to overlap it with their collider the same way they would in vanilla

crimson jolt Mar 21, 2025, 7:35 PM

#

thin solstice I am playing my modpack and a friend with probably over 100 vanilla Butler kills...

out of curiosity, does your friend (or did your friend ever) run "doorfix"?

#

that mod shrinks door colliders which makes it almost impossible for any enemy to open doors

thin solstice Mar 21, 2025, 7:36 PM

#

I know of that mod, I will have to ask

keen ruin Mar 21, 2025, 7:53 PM

#

PathfindingLagFix doesn't affect door opening at all

#

there is zero possibility of that being affected

thin solstice Mar 21, 2025, 8:16 PM

#

Well thanks, I just wanted some confirmation

keen ruin Mar 22, 2025, 10:14 AM

#

yeah no problem, just wanted to be absolutely clear

jolly wind Mar 23, 2025, 11:14 AM

#

storm lava Anybody know what it could be the problem? Just had a spider and a hoarder bug r...

I think I’ve been having the exact same problem. Enemies either don’t pathfind at all (like coilheads just running in place) or they all pathfind to some random corner of the facility and kinda ignore players. It was definitely happening on generic moons even though it had vanilla interiors. That said, things like this were happening before, but installing pathfindinglagfix seemed to make it worse. So probably not this mod’s issue.

keen ruin Mar 23, 2025, 5:19 PM

#

it does seem odd if it gets worse with PathfindingLagFix

#

but as far as issues on a particular moon, I wonder if it might be something screwed up with the navmesh surface on those moons, that's the only thing I'm thinking of that would affect the interior in such a way

#

other than if the moon has modifier volumes, but that seems much less likely to me

jolly wind Mar 24, 2025, 11:18 PM

#

I figured it out and it wasn't this mod's fault, or even generic moons. It was an incompatibility between brutalcompanyminusextrareborn and some of demonmae's moons

keen ruin Mar 25, 2025, 8:02 AM

#

oh that's curious

#

glad you found the problem!

keen ruin Mar 26, 2025, 10:26 PM

#

I forgor to push this out onto stable, it's been in beta for way too long lol

Version 2.2.0

Patched masked to use the cached elevator controller instead of finding it fresh every AI interval.
Patched the method to find the main entrance to use a cache and avoid a FindObjectsOfType call when masked are pathing to it.
Patched hygroderes to reduce the physics casts per frame from 8 to 1.
Patched the maneater AI to avoid synchronous pathfinding when the adult is sneaking and when it checks if it is stuck.
Patched coilheads' player targeting pathfinding to run asynchronously.

keen ruin Mar 28, 2025, 12:56 AM

#

Pathfinding Lag Fix | Pathfinding Lib

fast cypress Mar 28, 2025, 4:07 AM

#

I think the recent hygrodere change makes the slimes really uh tall

ashen harbor Mar 28, 2025, 4:18 AM

#

I have not seen this happen lmao

#

Wtf

fast cypress Mar 28, 2025, 4:22 AM

#

maybe its a conflict with something else but it and lobby control are the only two mods that have updated

#

and now we're getting this

#

i hate tall slime its cursed 😭

ashen harbor Mar 28, 2025, 4:31 AM

#

Idk I never had this with the Beta so maybe it's some oddity that got pushed to the stable update

keen ruin Mar 28, 2025, 7:41 AM

#

hmm I did see this one time but I'm not sure why

#

not sure if it was with my beta actually, it was a good while ago

#

but if you catch a seed or especially if you see one spawn and can record it that would be incredibly helpful

keen ruin Mar 30, 2025, 2:06 PM

#

ah actually, I might have an idea why it does that

#

I have a feeling it would look normal with ButteryFixes

#

it is probably using a bad initial value in the first fixed update and not fixing itself due to the screwy way it updates the positions

crimson jolt Mar 30, 2025, 4:55 PM

#

keen ruin it is probably using a bad initial value in the first fixed update and not fixin...

do you think it'd be worth consolidating that change back into pathfindinglagfix?

#

it's your patch to begin with so like... not like im gonna fight to keep it or anything

#

or do you think this issue would've replicated still in vanilla

keen ruin Mar 31, 2025, 10:20 PM

#

I... could maybe consider doing so, but I think I would rather fix the incorrect initial state so that it works in the vanilla case, since the patch is supposed to not have much of an effect on gameplay, at least when the game is performing normally

#

that's if it is what I'm thinking it is, I'll have to see when I get to this

#

it'll take a little fiddling with the patches I think

#

it's implemented as a transpiler so it's a bit more of a pain to do than it would otherwise be

crimson jolt Mar 31, 2025, 10:23 PM

#

well, let me know if there's something on my side that needs adjustment for compatibility

#

supposed to not have much of an effect on gameplay, at least when the game is performing normally
i agree with this (and the general mission statement of vanilla gameplay parity) but i think this is sort of a special case

#

in a way

#

since it makes the slime work consistently at all framerates, instead of just at low framerates (an "issue" that might be exacerbated by PLF improving the framerate)

#

hard to say there's a clear answer

#

though

keen ruin Mar 31, 2025, 10:46 PM

#

yeah

#

I just talked to tomato who helps administer (or something) the speedrunning community and it sounds like it would be reasonable to bring into PathfindingLagFix under the OnlyFixes preset actually

#

but I still need to fix the unfixed behavior for the sake of the Vanilla preset

keen ruin Apr 2, 2025, 4:48 AM

#

actually, I forgot that the damage area of the blob is more affected by the bone positions than I thought (though still not totally because it's a bit spaghetti how that works), so maybe I'll have to discuss that further

#

that said, the fix for the tall blobs works

#

Version 2.2.1

Fixed blobs often being stretched to the origin for a while after they spawn.

knotty leaf Apr 10, 2025, 2:47 PM

#

keen ruin ### Version 2.2.1 - Fixed blobs often being stretched to the origin for a while ...

Hello!
I really hate to blame anything in particular, but I wanted to let you know that I'm having a general problem with all the monsters.
None of them are able to open doors; instead, when they stay in the doorway for too long, they end up going right through.
The first one I noticed was the masked monsters in StorageComplex, thinking it was a dungeon bug, but I realized this bug is present in any interior and with any monster: crawlers, coilheads, brackens, etc.

As far as I know, this is the only mod that could make a change that could affect that, so I'd rather let you know and get some help fixing this.

crimson jolt Apr 10, 2025, 3:12 PM

#

knotty leaf Hello! I really hate to blame anything in particular, but I wanted to let you kn...

are you using doorfix?

#

i think it's kind of fallen out of the public eye

#

but it was a known issue for a while that it prevents monsters from opening doors

#

because it changes the size of the door trigger (which enemies have to overlap to open the doors)

#

it's why testaccount fixed the grab action instead of changing the doors

#

in itemgrabfix

knotty leaf Apr 10, 2025, 3:15 PM

#

crimson jolt are you using doorfix?

if it's this mod, no

crimson jolt Apr 10, 2025, 3:15 PM

#

yeah that's the one

#

i wonder if something else you have might be making similar changes

#

but that's my ace in the hole gone

#

at least

crimson jolt Apr 10, 2025, 3:16 PM

#

thin solstice I have been noticing a thing in my modpack lately where Butlers are opening door...

someone reported having a similar (but opposite) issue a while back

crimson jolt Apr 10, 2025, 3:16 PM

#

keen ruin PathfindingLagFix doesn't affect door opening at all

but zaggy's pretty confident it shouldn't be related

#

that is probably also the case here

keen ruin Apr 10, 2025, 3:38 PM

#

yeah the only way it should affect that is if the enemies have stopped pathing to closed doors entirely, and it would be difficult for the pathfinding patches to even cause that

green gust Apr 10, 2025, 4:17 PM

#

knotty leaf Hello! I really hate to blame anything in particular, but I wanted to let you kn...

if you're using ReXuvination can you try without?

#

it should just still be fine though

knotty leaf Apr 10, 2025, 4:17 PM

#

green gust if you're using ReXuvination can you try without?

yes, I have it

#

I'll try without it

#

this is what happens (this is the crawler)

#

it wiggles on the door

#

now testing without Xuvination

green gust Apr 10, 2025, 4:28 PM

#

i think it is rexuvination, that's interesting

knotty leaf Apr 10, 2025, 4:30 PM

#

green gust i think it is rexuvination, that's interesting

yup, it is

#

it opens

green gust Apr 10, 2025, 4:31 PM

#

ill get a fix in a sec, weird that my enemies work but vanilla doesnt lol

knotty leaf Apr 10, 2025, 4:31 PM

#

thanks!

green gust Apr 10, 2025, 4:46 PM

#

knotty leaf thanks!

will update ina a sec

knotty leaf Apr 10, 2025, 4:46 PM

#

green gust will update ina a sec

gotcha!

keen ruin Apr 10, 2025, 9:30 PM

#

green gust ill get a fix in a sec, weird that my enemies work but vanilla doesnt lol

that is curious indeed

#

do you have your colliders set up differently?

green gust Apr 10, 2025, 9:41 PM

#

keen ruin do you have your colliders set up differently?

i didnt think i did, maybe i had interactableobject not excluded but i thought i did

keen ruin Apr 16, 2025, 9:26 PM

#

Version 0.1.0

Made NavMeshQueryUtils.FindStraightPath call through to the Unity runtime instead of reimplementing the algorithm. The results should now match the output from NavMeshPath.corners perfectly.
Deprecated the original overload of NavMeshQueryUtils.FindStraightPath in favor of one that requires less allocations to match the native Unity function's signature.

green gust Apr 16, 2025, 9:29 PM

#

Rikka_clap

#

now release 1.0.0

keen ruin Apr 16, 2025, 9:31 PM

#

maybe if we get enough testing in with the upcoming API teehee

green gust Apr 16, 2025, 9:41 PM

#

call me up for it anytime bbg

keen ruin Apr 16, 2025, 9:44 PM

#

oh I definitely will

#

it's gonna be an adventure

#

this is what I have on my list still though, I feel like it might take a bit longer

vanilla elevator API impl component
allowed link flags
masked smart pathfinding mod
- option to include all nodes and skip going in/out based on player positions? they don't roam too well with this yet??
- option to allow fire exits to be used
- is there a good way to prevent masked from using offmesh links that are intended for non smart AI?
teleports within area
testing...

#

the big one is the masked mod

#

I gotta turn my hacky implementation into proper transpilers

keen ruin Apr 23, 2025, 6:37 PM

#

I thought I should note this here, but I'm looking into an issue on the latest version of PathfindingLib where the change to use Unity's FindStraightPath method is resulting in paths "succeeding" when they shouldn't be, which can cause enemies patched by PathfindingLagFix to stutter step

#

currently investigating what is causing it, but I'm not seeing any differences in how I'm calculating it versus how NavMeshPath.CalculateCorners does

#

hopefully I can find what the issue is before too long, but for now I would recommend downgrading it

tardy pivot Apr 23, 2025, 6:39 PM

#

does it always succeed or just in some cases?

keen ruin Apr 23, 2025, 6:54 PM

#

not sure, but I have a test case where the results differ from NavMeshPath.CalculateCorners and I'm trying to figure out why

#

in the job it's succeeding and returning a partial path followed by a corner that is exactly at the destination

#

the partial path doesn't match the NavMeshPath version

#

trying to figure out if there's somewhere in the native code that causes the difference

ashen harbor Apr 23, 2025, 6:57 PM

#

keen ruin hopefully I can find what the issue is before too long, but for now I would reco...

https://tenor.com/view/anime-loli-gif-20946617

Tenor

keen ruin Apr 23, 2025, 6:58 PM

#

the difference between NavMeshPath and the job version is in the last corner in the partial path only, every other corner is exactly the same afaict

#

very odd indeed

#

and the input polygons match identically

tardy pivot Apr 23, 2025, 7:02 PM

#

wait it returns it as partial?

#

i belive native always returns the destination as the last but if the status is marked as partial it ignores it

keen ruin Apr 23, 2025, 7:04 PM

#

the result of EndFindPath is partial yeah

#

but FindStraightPath is unaware of that and I couldn't find a place where it reduces the size of the managed array by 1 if it is partial in the NavMeshPath

tardy pivot Apr 23, 2025, 7:05 PM

#

tardy pivot i belive native always returns the destination as the last but if the status is ...

I'll have to double check on ghidra this tho

keen ruin Apr 23, 2025, 7:05 PM

#

I was wondering the same thing but I see no place where that logic lives

#

are you able to look now or is this a later thing?

tardy pivot Apr 23, 2025, 7:06 PM

#

it's probably a "in 5min" thing if my RDP works

keen ruin Apr 23, 2025, 7:06 PM

#

ohh I see

#

NavMeshPath.m_Status (my name for the field @ 0x4) should be comparable to the result of NavMeshQuery.EndFindPath I believe, but it doesn't get referenced in CalculateCornersInternal afaict

#

looks like the actual result of NavMeshQuery methods isn't kept so I think that's all that's relevant here

tardy pivot Apr 23, 2025, 7:21 PM

#

what was the result of FindStraightPath?

#

it's a PathQueryStatus no?

keen ruin Apr 23, 2025, 7:22 PM

#

it did appear to be yeah

#

success being 0x40000000 makes it seem very unlikely not to be

#

so that's what I'm currently interpreting it as, and in the case where I'm getting false positives it is giving only that result

tardy pivot Apr 23, 2025, 7:26 PM

#

local_78 = local_78 & 0xffffffff00000000; that's one way to zero out an int out of a long field 👀

keen ruin Apr 23, 2025, 7:29 PM

#

huh

#

not the same as a cast though if it's signed thonk

#

or wait I'm getting it backwards

#

what is that used for??

tardy pivot Apr 23, 2025, 7:31 PM

#

to store the PathQueryStatus

#

or more precisely to dynamically retreive it from a "memoryManager"

#

and that local is the destination and the output of CalculathePathCorners

keen ruin Apr 23, 2025, 7:36 PM

#

yeah

#

that's getting the non-detail flags from the status

#

I'm seeing it returning the number of straight path corners from NavMeshManager::CalculatePathCorners though

#

it's very screwy to follow here since it's doing some weird stuff in the decomp for the calling convention but it is passing a local to FindStraightPath in the second to last parameter as an uint64* and then returning that as an int

#

on my end all I see it doing with the result of FindStraightPath is checking if it failed and returning a corner count of 0 if so

tardy pivot Apr 23, 2025, 7:41 PM

#

uh?

keen ruin Apr 23, 2025, 7:41 PM

#

(which is irrelevant here because I'm getting a partial path back from NavMeshPath.corners, it's not hitting that)

#

oh actually it's passing it as int* I forgor

#

#

with int local_res8[2]

tardy pivot Apr 23, 2025, 7:44 PM

#

thoose are the pathRefs right?

keen ruin Apr 23, 2025, 7:44 PM

#

I'm guessing that maybe what makes it do all this crap with puVar11 is the call site addresses it's assigning to -8

#

what are pathRefs?

#

what I highlighted?

tardy pivot Apr 23, 2025, 7:45 PM

#

in NativeArray<PolygonId> straightPathRefs,

#

the second to last param

keen ruin Apr 23, 2025, 7:45 PM

#

second to last param is the out int straightPathCount

tardy pivot Apr 23, 2025, 7:45 PM

#

on ho it's the path count

keen ruin Apr 23, 2025, 7:45 PM

#

we drop the last parameter in our wrapper

tardy pivot Apr 23, 2025, 7:45 PM

#

uh

tardy pivot Apr 23, 2025, 7:48 PM

#

keen ruin

oh it is an array for you. in release it's a single value

keen ruin Apr 23, 2025, 7:48 PM

#

local_res8 you mean?

tardy pivot Apr 23, 2025, 7:48 PM

#

yeah

keen ruin Apr 23, 2025, 7:49 PM

#

it's just confused about the stack a little in the decomp

tardy pivot Apr 23, 2025, 7:49 PM

#

or you just retyped it

keen ruin Apr 23, 2025, 7:49 PM

#

it's a single value here too

#

retyped and renamed:

tardy pivot Apr 23, 2025, 7:52 PM

#

🤔

keen ruin Apr 23, 2025, 7:53 PM

#

are you seeing something different?

tardy pivot Apr 23, 2025, 7:54 PM

#

nono that checks out

#

i'm trying to find where it's adding/removing that extra corner

keen ruin Apr 23, 2025, 7:54 PM

#

how sure are you that you saw that?

tardy pivot Apr 23, 2025, 7:55 PM

#

not too much anymore

keen ruin Apr 23, 2025, 7:55 PM

#

I haven't seen any evidence of it, and the fact that corner index 3 (with 5 total in our version) is differing from NavMeshPath makes me think something else is going on

#

not sure what though........

#

for your reference, this is with NavMeshQuery:

  - float3(-19.05093f, -219.5336f, 86.91694f) @ 562954802036800
  - float3(-25.335f, -219.5336f, 84.71749f) @ 562954802036794
  - float3(-26.91203f, -219.5336f, 84.02847f) @ 562954802036837
  - float3(-28.71834f, -216.0919f, 69.90083f) @ 281479821131788
  - float3(-28.21279f, -182.1419f, 73.91283f) @ 281479821131788

and with CalculatePath/NavMeshPath:

[Info   :PathfindingLib] 0: (-19.05, -219.53, 86.92)
[Info   :PathfindingLib] 1: (-25.34, -219.53, 84.72)
[Info   :PathfindingLib] 2: (-26.91, -219.53, 84.03)
[Info   :PathfindingLib] 3: (-28.72, -212.83, 63.75)

#

Y and Z differ significantly for the fourth corner for some reason

tardy pivot Apr 23, 2025, 8:02 PM

#

is the destination in the nodes too?

#

oh true

keen ruin Apr 23, 2025, 8:02 PM

#

not sure what you mean

tardy pivot Apr 23, 2025, 8:02 PM

#

nothing

keen ruin Apr 23, 2025, 8:02 PM

#

with NavMeshQuery, the destination is the 5th corner

tardy pivot Apr 23, 2025, 8:03 PM

#

i'm trying to see if somehow CalculateCorners populates differently the parameters of FindStraightPath

keen ruin Apr 23, 2025, 8:03 PM

#

the polygons part is the same at least

#

afaict

tardy pivot Apr 23, 2025, 8:04 PM

#

what do you have mapped as "param2" of the CalculatePathCorners?

keen ruin Apr 23, 2025, 8:05 PM

#

it appears to be the capacity of param_1

tardy pivot Apr 23, 2025, 8:05 PM

#

it's adding 2 to the value of a field in NavMeshPath

#

keen ruin Apr 23, 2025, 8:06 PM

#

I believe they allocate a corner array that is the number of polygons + 2 since it's the maximum amount of corners that it can result in

#

I'll probably mirror that in my job code later

#

I'll call them corners and cornerCapacity

tardy pivot Apr 23, 2025, 8:07 PM

#

could it be that it does not have the space for computing?

keen ruin Apr 23, 2025, 8:07 PM

#

in ours?

tardy pivot Apr 23, 2025, 8:07 PM

#

ye

keen ruin Apr 23, 2025, 8:08 PM

#

no we waaaay overallocate

tardy pivot Apr 23, 2025, 8:08 PM

#

uh?

keen ruin Apr 23, 2025, 8:08 PM

#

we give it 128 corners

#

it results in 4

tardy pivot Apr 23, 2025, 8:09 PM

#

uhm. the param is not the corner count but the node count

keen ruin Apr 23, 2025, 8:11 PM

#

tardy pivot

where are you seeing this?

#

I dkon't have a param_2 in NavMeshPath::CalculateCornersInternal

tardy pivot Apr 23, 2025, 8:11 PM

#

in CalculatePathConrners based on what it's passing to FindStraightPath

#

CalculateCornersInternal -> CalculatePathConrners -> FindStraightPath

keen ruin Apr 23, 2025, 8:13 PM

#

I see the + 2 in CalculateCornersInternal rather than CalculatePathCorners

#

Hmm

tardy pivot Apr 23, 2025, 8:13 PM

#

CalculatePathConrners passes it's param_2 as the last param of FindStraightPath. and it's param_1 as the 4th

#

and by the analysis we did before param_2 is 2 elements larger than the original size

keen ruin Apr 23, 2025, 8:14 PM

#

param_1 should be the 5th but that sounds correct

#

param_2 is maxStraightPath

tardy pivot Apr 23, 2025, 8:15 PM

#

yeah 5th

keen ruin Apr 23, 2025, 8:15 PM

#

it's the capacity of all the pointers

#

the output arrays

tardy pivot Apr 23, 2025, 8:15 PM

#

we called it like that but it also seems to be the size of the in array too

keen ruin Apr 23, 2025, 8:15 PM

#

I mean

#

it's based on the size of the input array

#

but that's because the input array limits the size of the output array

#

it allocates arrays to the size of the maximum number of output corners

#

or I assume it does

#

which is polygons + 2

#

we give it 128 instead

tardy pivot Apr 23, 2025, 8:16 PM

#

then why also resize the input array?

keen ruin Apr 23, 2025, 8:17 PM

#

where?

tardy pivot Apr 23, 2025, 8:17 PM

#

in the "internal" one

#

#

i also wonder how an uint + 2 can result in 0 but that's another story XD

keen ruin Apr 23, 2025, 8:19 PM

#

yeah idk that might be a macro lol

tardy pivot Apr 23, 2025, 8:22 PM

#

also another intresting thing in the non-internal one:
the path size ( aka. param 4 of FindStraightPath ) is taken from a field of NavmeshPath instead of the size of the array 🤔

keen ruin Apr 23, 2025, 8:23 PM

#

yeah, NavMeshPath has a field for the number of polygons

#Pathfinding Lag Fix | Pathfinding Lib | Smart Enemy Pathfinding

Version 2.2.0 (Beta)

Version 2.2.0

Version 2.2.1

Version 0.1.0