Starlancer AIFix v3.13.2 | EnemyEscape v3.0.0

[gleaming robin]

what am I lookin at bruh

[twilit drift]

the emit delegate stuff

[gleaming robin]

is my understanding correct then?

[faint copper]

mrov didn't highlight the ldloc_1

[twilit drift]

🥺

[faint copper]

feels weird to me to have self be the last argument but I'm sure that's because they emit the ldarg.0 immediately before the call

personally I would just emit the ldarg.0 myself to make it the first one lol

[gleaming robin]

they do yeah

[faint copper]

wait who are you talking about? did you look at the MonoMod source?

[gleaming robin]

oh sorry, the example code I shared had
c.Emit(OpCodes.Ldarg_0); // load argument 0 'this' onto stack
right before the EmitDelegate

[faint copper]

wot

@twilit drift you didn't load arg 0 in the code you linked

is it automatic or not??

[twilit drift]

😳

umm

it's working?

I don't know 😭

[gleaming robin]

Just for you to look at yourself in full, this is the example I'm talking about
https://lethal.wiki/dev/fundamentals/patching-code/monomod-examples#my-first-ilhook

[faint copper]

thank you

it's much easier to tell what's going on with context lol

[gleaming robin]

yea lol

if I'm understanding things correctly (probably not) then I just need to load arg.0 and i4.0 on the stack, not emit a ldsfld, and from there let the EmitDelegate handle the method?

👀

idk if it works, but I loaded up with no error this time

[twilit drift]

👀👀

[faint copper]

what's the constant 0 for?

can you post a diff of what you're trying to do to the vanilla method?

from what I saw earlier, you only had a self argument to your delegate

[faint copper]

tbh I'm not sure how, I see nothing in EmitDelegate on GitHub that indicates it tries to load this

[twilit drift]

I am not sure as well

[faint copper]

what's loc 1?

[twilit drift]

I'm planning to change it to a harmonytranspiler instead of ILmanipulator

[faint copper]

(where's the comment about what the load is for )

[twilit drift]

😳

    // // to be completely honest, I have no idea what I'm doing
    // // https://github.com/SylviBlossom/LC-SimpleWeatherDisplay/blob/2d252b92dcd4d8ef259b8072d9339ff5ccdc4d0b/src/Plugin.cs#L127-L153
    // // this is just this, but repurposed

here

[faint copper]

also looking at EmitDelegate it seems like I was right not to want to use something like that

it stores a reference to the delegate in a list to keep it alive for the duration of the program, whereas if you just emit a call yourself you can avoid any indirection

[faint copper]

oh I missed that

[gleaming robin]

man, it might help if I turn warning logs back on in bepinex

[faint copper]

WHAt

turning off logging...

why are y'all making me judge you so much today

[twilit drift]

😳😭

it's a bowl of spaghetti

[gleaming robin]

prior to this I was trying to test something I needed loginfo for, and meaningless vanilla warnings (failed to init audio spatializer) were spamming

so I just wanted peace

[rocky fable]

mrov needs lotsa spaghetti!!!!

[faint copper]

true those ones are very annoying

I won't ever turn off warnings though, I don't trust myself to remember to turn them on again

I don't mind copying logs into npp and scanning them

[rocky fable]

https://tenor.com/view/super-mario-and-luigi-mario-and-luigi-mario-luigi-chat-gif-15119085

[gleaming robin]

understandable lol

okay, well I load in with no IL error, but my code isn't firing

I was really hoping to see logger.LogWarning("Baboon hawk is trying to carry scrap back to the nest!"); flooding in ; w ;

[twilit drift]

🥺

so at least you're targeting correctly? 🥺

[gleaming robin]

I guess? xD

[faint copper]

@twilit drift I understand why it works now, ldloc.1 isn't loading the string, it's loading this but within a generated Enumerator.MoveNext()

silly enumerators

[twilit drift]

oh gosh

[faint copper]

what do your codes look like now?

[twilit drift]

I'm not surprised I didn't understand what I was doing 😆

[faint copper]

also why are you planning on changing to a transpiler?

[twilit drift]

because I don't like the current implementation 🥲

it's yoinked

[faint copper]

hmm I see

if I were you I would stick with ILCursor

[twilit drift]

it's better for the stuff I'm doing?

[faint copper]

well, I haven't directly compared them, but I would imagine ILCursor is a nicer experience than CodeMatcher

[gleaming robin]

I'm gonna screenshot it so I don't get thrown in jail again

[faint copper]

you can emit the same code with ILCursor that you can with transpilers

[twilit drift]

I know, but I've played with CodeMatcher and it was easier to understand then that

[faint copper]

oh? huh

[faint copper]

where is that ldc.i4.0 going?

you're not popping it there

[gleaming robin]

to me it seems like the basics of IL are super unintuitive (alphabet soup nonsense) but I feel like I'm getting somewhere

[faint copper]

I would've expected this not to pass validation

[gleaming robin]

¯_(ツ)_/¯

[faint copper]

you only have one argument to your delegate but you're emitting two instructions that push to the stack

I'm not sure if MonoMod validates the IL, but if it was doing so I would've expected it would not be very happy about that

[gleaming robin]

well removing the ldc didn't throw any errors so... idk

I'm just trying to figure out the process lol

[faint copper]

I guess having an extra value on the stack isn't necessarily fatal unless the VM doesn't like it when ret happens with a non-empty stack

and I don't know if that's valid or not

[gleaming robin]

well it's gone now anyways

[faint copper]

yeah, I'm mainly trying to think through whether it should've detected that after the patch ran

wait

you don't even have those in the order it would need to be in, that should crash

so maybe that's one point for Harmony

I'm pretty sure it would've seen that the call is trying to pop a value of the wrong type off the stack and complained

yeah, looks like a transpiler explodes when I do that

so maybe you should switch over to a transpiler and use CodeMatcher instead

it'll be much easier for you to start out if it actually tells you when your code doesn't make sense

it's not the best at telling you why, but you'll at least see an error on startup instead of it happening when your function is finally called

the CodeMatcher functionality seems quite similar to what you're working with right now as well

[gleaming robin]

what order should it be in?

[faint copper]

so

you had

ldarg.0
ldc.i4.0
call delegate void(BaboonBirdAI)

if you go through that in order, it would:

• push BaboonBirdAI this
• push i32 0
• pop BaboonBirdAI arg0, receive i32 instead

it can detect that the call is trying to pop a BaboonBirdAI instance and getting an i32 instead, and say "mmm I think this isn't quite right "

but it didn't

transpilers would

[gleaming robin]

huh, why didn't MM explode

[faint copper]

so they are better

[gleaming robin]

or at least return an error

[faint copper]

it's because transpilers specifically validate the IL they generate after they finish

[gleaming robin]

ah

[faint copper]

it may also apply to ILManipulators but I'm not sure

[gleaming robin]

well even after removing the ldc it still doesn't work

[faint copper]

yeah

if it hit your code you would've gotten a weird error

did you print a message when it runs your patch so you know that it's actually running it?

[golden basin]

Print the il to see if it's actually changing at the end

[faint copper]

not the delegate, but the code that generates the IL

[golden basin]

:p

[faint copper]

is it possible for it not to change it?

if the ILCursor is invalid and you try to emit, does it eat the error?

[golden basin]

It shouldn't eat any errors afaik

Here lemme bring the expert

@rustic needle

[faint copper]

uh oh

Hamunii isn't gonna be happy to see my hating

[rustic needle]

I'm gaming, what

[golden basin]

Lmao

I dont think ive ever seen u game

They're doing il hooking stuff

[rustic needle]

lol

[faint copper]

would it be possible for an ILCursor to not modify things if it's told to emit? I assume it wouldn't eat errors

[golden basin]

Yeah that

Also what're u playing

[rustic needle]

LC

[gleaming robin]

I put it with GameNetworkManager, so if it wasn't loading then none of my methods should be running on an enemy, but I went ahead and added a log beside it

[faint copper]

is that printed inside your ILContext function? if not, then it's not really a good test

[gleaming robin]

oh I can put a normal log in there? I assumed it had to be all IL stuff only

[faint copper]

you're sending a delegate to MonoMod and then assuming it's running it

it's just any other function, yeah

[gleaming robin]

oh, well then

lemme try

[faint copper]

you could do all kinds of messed up stuff in there

[golden basin]

Ayoo what happened to playing the update together

[gleaming robin]

@faint copper hoping this is what you meant lol

[faint copper]

ah yes

[rustic plume]

are you finally gonna update it

the enemyescape mod

[faint copper]

you should also be able to log the ILCursor instance to get information about the code generated, I believe

hopefully with enough IL context to be able to see if it worked

although now what I wonder is, have you given it time in-game to actually have a little guy pick up an item?

you'd have to scatter a lot of items around for them to find one I think

[gleaming robin]

I'm will say this once. I will update it when the update is ready.

[gleaming robin]

I always make sure it picks up an item

[faint copper]

ah ok hmm

[gleaming robin]

last test I made it real quick and spawned an item for it to pick up lol

[faint copper]

did you check that the behavior state corresponds to the one you're injecting into?

[gleaming robin]

lemme doublecheck

[faint copper]

on next run it would be good to get the final IL you generate by printing the ILCursor and then check with UnityExplorer whether the baboon hawk is in the correct state

[gleaming robin]

Imperium shows me a neat little box above the enemy with the info

How would I print the ILCursor? Just log $"{ILCursor}"?

[faint copper]

print your local variable instance of it, but yes

[gleaming robin]

logger.LogInfo(c);```

[faint copper]

indeed

also... aren't you injecting into behavior state 0 and not behavior state 1 where it tries to navigate to the camp? can you show me the code you're trying to inject into in context?

[golden basin]

When's the lethal escape update

[gleaming robin]

https://tenor.com/view/deus-vult-deus-vult-intensifies-sword-triggered-gif-15356993

[faint copper]

what's wrong with the current update anyways lol

are people just complaining for no reason?

[gleaming robin]

nothing's "wrong" with it necessarily

[faint copper]

I figured

classic

[gleaming robin]

just compared to where I'm currently headed

it's very crude lol

[faint copper]

yeah

it happens

[golden basin]

Are there people complaining?

[faint copper]

not you, but [this person](#1206494982521753620 message) has asked a similar question at least once before iirc

if not more than twice already

[gleaming robin]

the actual code from the hawk's DoAIInterval?

[faint copper]

yes

[golden basin]

Though tbf users are usually split up into two categories, the constant askers, and the constant supporters who are low key askers

[gleaming robin]

it's supposed to be in behaviour state 1 when it's carrying the scrap to the camp

[faint copper]

I haven't gotten that much of that, but I guess most of my mods are for nerds

[gleaming robin]

but my code shouldn't care for the purpose of what we're trying to do rn

[golden basin]

Me neither since I'm not popular :3 but I've seen some mod creators get asked daily lol

[faint copper]

oh wait I forgot you had it checking if baboonCampPosition was loaded, never mind me

but yeah lemme know when you have that printout of the generated IL

[gleaming robin]

IL_0000: ldarg.0```

[faint copper]

uhhh

is that at the start or the end?

[gleaming robin]

yes

; w ;

[faint copper]

wot

[faint copper]

if the answer is that it's at the start then you should move it to the end

we need to see it after you make changes

[gleaming robin]

OH okay I totally misunderstood the assignment

https://tenor.com/view/homer-brain-monkey-gif-11098413

me rn ^

[faint copper]

understoodable

[golden basin]

It's okay, il hooks killed me today too

[faint copper]

I just hope that it's gonna print out more than one instruction

[golden basin]

I couldn't even manage to do the simple il hook I was doing lol

[faint copper]

the MonoMod docs say that it should print "surrounding instructions" so I'm not sure why you only saw one

[golden basin]

Whatd you print specifically

[faint copper]

the cursor

which will use ToString()

[golden basin]

I dont think you print the cursor

Aren't u supposed to print the param il

[gleaming robin]

[Info   :Starlancer EnemyEscape] // ILCursor: System.Void DMD<BaboonBirdAI::DoAIInterval>?-812375320::BaboonBirdAI::DoAIInterval(BaboonBirdAI), 369, None
IL_0000: call System.Void MonoMod.Cil.RuntimeILReferenceBag/FastDelegateInvokers::Invoke<BaboonBirdAI>(T1,MonoMod.Cil.RuntimeILReferenceBag/FastDelegateInvokers/Action`1<T1>)
IL_0485: ldarg.0```

[faint copper]

you mean the context?

perhaps I suppose

[golden basin]

Ye

Pretty sure u are

[gleaming robin]

okay I'll try that

[golden basin]

It should give you a large text dump

Its what I logged to check changes

[faint copper]

wow I guess "surrounding instructions" is one instruction following the current one lol

[gleaming robin]

nope here we go, one sec

[golden basin]

Yep he's got it lol

[faint copper]

it's gonna be a big print for this method which is why I wanted ILCursor to work

[golden basin]

I usually just screenshotted the relevant parts

Otherwise it just looks like a huge mess

But idk what you're looking for specifically

[gleaming robin]

I wanna let Zaggy decide what's relevant

[golden basin]

Aren't you changing something in this method?

[gleaming robin]

yea I just felt like presenting the whole log rather than risk somehow leaving something relevant out

[golden basin]

That's fair

Good luck zaggy

(Its not too hard just annoying to read at first lol)

I'm glad I have a second monitor

[faint copper]

it's fine

[golden basin]

And here I was being told I learned nothing from the il hooking I did today 🥺

[faint copper]

does GotoNext() put you at the index of the first predicate instead of after the last??

[golden basin]

I'm gonna say yes because I think i know what you're talking about but idk what predicate is

[gleaming robin]

the code should be at IL_0485, which right before that is some IL_0000 stuff which I'm assuming is supposed to be me?

[golden basin]

I had to do moveType.after and then - index by 1

[faint copper]

predicate is the lambda that returns a bool to determine whether an instruction is what you're looking for

[golden basin]

Ah icic

[gleaming robin]

wait I think I goofed, lemme check something

[faint copper]

yeah I'm gonna be honest the fact that the MonoMod docs don't say where the cursor ends up for a predicate array with length > 1 is kinda annoying

I'm gonna assume that it puts it at the start and you need to += predicates.Length; which isn't exactly ideal since that means you have to toss a magic number or local array in there

it looks like your Index-- is going here and then deleting this instruction:

[gleaming robin]

Yea I was dumb and meant to get rid of that attempted removal

[faint copper]

well

the removal is fine if you pop twice three times

actually, since it expects it to push a value two pops should do

lol

[gleaming robin]

regardless, removing the removal didn't change any of the surrounding code (except for not removing 0483)

[faint copper]

yeah it wouldn't

you were deleting the branch instruction I showed above

if you want to keep using ILCursor, you would need to c.Index += [number of predicates]

but... I'm gonna be honest, it seems like CodeMatcher's MatchEndForward function would be a lot smoother for this

I actually had thought that ILCursor had similar functionality, but the docs don't mention anything that I can find

if you want to use CodeMatcher, you can just give your function a [HarmonyTranspiler][HarmonyPatch(Type, string)] and then construct the CodeMatcher(IEnumerable<CodeInstruction>, ILGenerator) from your transpiler's parameters of the same type

then harmony.PatchAll(Type) as usual for prefixes and postfixes

https://harmony.pardeike.net/articles/patching-transpiler.html

[gleaming robin]

At this point I might jump ship yea @_@

[twilit drift]

https://tenor.com/view/wolf-of-wall-street-jordan-belfort-leonardo-di-caprio-one-of-us-jonah-hill-gif-5441859

[gleaming robin]

I remember trying to use CodeMatcher and something was going wrong tho

[faint copper]

I'm sure we can figure it out

[gleaming robin]

and I think it was somehow mrov's fault?

[faint copper]

uh oh

thrown under the bus yet again

at this point mrov might as well just become the bus's skid plates

[gleaming robin]

ah yes, this was the issue

[twilit drift]

because i was using different harmony version because i was targeting net48

oopsie

[gleaming robin]

yea that

[faint copper]

so does that mean you already know the solution?

[gleaming robin]

nope

[faint copper]

what does your csproj look like?

[gleaming robin]

I think that's what lead to me going to MM

sec

@faint copper

lemme just paste it

or not, it's too long lol

there we go

I'm a fool that's the old one

aspodfijasopdfij

[faint copper]

try this for your packages and remove the references to BepInEx, Harmony, HarmonyX dlls (and the MonoMod ones if you don't need them)

    <ItemGroup>
        <PackageReference Include="Mono.Cecil" Version="0.11.5.0"/>
        <PackageReference Include="BepInEx.BaseLib" Version="5.4.21.0"/>
        <PackageReference Include="BepInEx.Core" Version="5.4.21.0"/>
        <PackageReference Include="BepInEx.AssemblyPublicizer.MSBuild" Version="0.4.1" PrivateAssets="all"/>
    </ItemGroup>

ope one of the lines changed color when I didn't mean it to lol

[gleaming robin]

no warnings or errors!

and a much cleaner dependency list lol

If I ever hit the lottery @faint copper , you're getting at least ten dollars of it

[faint copper]

I'll buy a half a jalapeno burger with it

[gleaming robin]

🤭 now to figure out what I'm doing here

Okay, so it seems like opcode matching is still pretty simple, but for ldsfld I'm not sure of the syntax. The example I'm looking at on BepInEx's github wiki has this:

new CodeMatch(i => i.opcode == OpCodes.Ldfld && ((FieldInfo)i.operand).Name == "test"))```

To which I assume I can swap Ldfld with Ldsfld, and FieldInfo with BaboonBirdAI

but then I run into not being able to directly reference baboonCampPosition like I was previously

[faint copper]

FieldInfo is a type that represents the field that the instruction is loading

what you could use is i => i.LoadsField(AccessTools.Field(typeof(BaboonBirdAI), nameof(BaboonBirdAI.baboonCampPosition)))

or better yet, put the field in a variable so that you're not requesting the field from AccessTools repeatedly:

FieldInfo baboonCampPositionField = AccessTools.Field(typeof(BaboonBirdAI), nameof(BaboonBirdAI.baboonCampPosition));
matcher.MatchEndForward([i => i.LoadsField(baboonCampPositionField)]);

you can also get the field by doing typeof(BaboonBirdAI).GetField(nameof(BaboonBirdAI.baboonCampPosition)), and there is a similar method for getting methods on a type

[gleaming robin]

Would I store the variable within the code instruction?

[faint copper]

what do you mean?

[gleaming robin]

like should it be declared within the class or the transpiler itself?

[faint copper]

the transpiler is fine, you would only need to put it outside the function if you're going to use it elsewhere

[gleaming robin]

kk

btw you've got MatchEndForward, but mine is MatchForward and I have to declare false/true for start/end

[faint copper]

and actually, it looks like CodeMatch can take an opcode and operand argument, so you can do this instead:

matcher.MatchEndForward([new CodeMatch(OpCodes.Ldsfld, typeof(BaboonBirdAI).GetField(nameof(BaboonBirdAI.baboonCampPosition))
)]);

this will pass the FieldInfo instance into the CodeMatch so that it's cached and then it will just compare after that instance each time

if you were to do a predicate instead like i => i.LoadsField([..]) it would evaluate [..] every time, which if you get the field in there can be costly

hopefully that makes sense

[gleaming robin]

I think so, that's what I was trying to do for a bit but I didn't know the syntax of it :P

Alright, so once I've matched the code I want to transpile, do I make an OpCode.call? Also, it's still there from the tutorial example here, but what is InstructionEnumeration()?

IEnumerable<CodeInstruction> Transpiler(IEnumerable<CodeInstruction> instructions)
{
    return new CodeMatcher(instructions)
        .MatchForward(false,
        new CodeMatch(OpCodes.Ldarg_0),
        new CodeMatch(OpCodes.Ldsfld, typeof(BaboonBirdAI).GetField(nameof(BaboonBirdAI.baboonCampPosition))),
        new CodeMatch(OpCodes.Ldc_I4_0))
                 .InstructionEnumeration();
}```

[faint copper]

you said false means start and true means end, didn't you?

if so, you'll want to pass it true

and InstructionEnumeration() just returns the IEnumerable<CodeInstruction> of the modified method body

to make your call, you would want to call Insert(CodeInstruction[]) and pass it an array of CodeInstructions that do what you want

that would go after you find the injection point with MatchForward()

so i.e.

.Insert(new CodeInstruction(OpCodes.Ldarg_0),
  CodeInstruction.Call(typeof(YourType), "YourMethodNameOnYourType", new Type[] { typeof(ParameterType1) }))

[gleaming robin]

the bool is useEnd so yea true should be end

[rustic needle]

I don't know the full context, but wouldn't GotoNext's MoveType.After argument do the job?

[rustic needle]

the ILCursor is not on an instruction, it is between instructions, similar to a text cursor. So, the first instruction if the previous instruction, and the next is the next instruction

[distant bay]

v50

[faint copper]

Ah, I missed that looking at the documentation, it probably would

[rustic needle]

Yeah, I should really work on improving the (unofficial) documentation on lethal.wiki. And perhaps talk with the MonoMod devs about it too, as it could be nice to have better official documentation

[faint copper]

Oh, I was looking at the MonoMod docs actually

the table of contents doesn't let you scroll through it so none of the GotoNext methods appear there lol

[rustic needle]

yeah, that sucks :(

[faint copper]

as far as the ILCursor, the documentation also doesn't really make that clear, but I would've thought of its current index as referring to the instruction in the list at its index rather than between them

I suppose it makes sense, but I think it would be nice if it printed more than two instructions, or at least if the documentation made it clear that's what it meant

[rustic needle]

well, that's why you should generally just print the ILContext il instead

[faint copper]

sure, but then you get 10 times as many instructions as you actually need

(in functions that are large obviously)

I'm not sure what the use case for printing two instructions at the position of the ILCursor is

seems like it should be configurable by an argument or just do a wider range

[rustic needle]

true. How does HarmonyX deal with printing the position of the CodeMatcher or whatever?

[faint copper]

that I don't know tbh

I don't use either lol

[rustic needle]

But well, I guess it is for seeing where the cursor is, and that's it

[faint copper]

that seems so situational to the point of being useless, but maybe I'm wrong

I guess it prints the offset..

[rustic needle]

It's useful if you don't know how the cursor works I guess

[faint copper]

right

I am wondering while you're here though, am I right that MonoMod IL manipulators (or whatever you would call them) don't validate the generated instructions? it seems like earlier Audio Knight was generating some invalid IL and it wasn't printing any errors despite that

[gleaming robin]

So this .Insert() will tell it to call SetDesToPos after matching the call that I want to modify? How would I feed it my little loop like I was trying to use with EmitDelegate?

.Insert(new CodeInstruction(OpCodes.Ldarg_0,
CodeInstruction.Call(typeof(EnemyAI), nameof(EnemyAI.SetDestinationToPosition))))
         .InstructionEnumeration();```

[faint copper]

I would've expected the code to cause undefined behavior or exceptions when it ran due to popping values of the wrong type off the stack

but transpilers seem to validate that before it gets to running it

[rustic needle]

They should immediately throw something like invalid IL code if the IL code is invalid, like if you don't pop when you should and stuff

[faint copper]

oh, that's odd

hard to say what was going on for him then

[rustic needle]

ye

[faint copper]

do you know if they both use some standard way of validating or their own differing implementations of that?

[rustic needle]

I guess feeding the arguments to methods isn't validated then?

No idea about the validation stuff

[faint copper]

I tried it on one of my transpilers and it caught it, but maybe it depends

[golden basin]

i got errors for doign that tho

[gleaming robin]

oh actually, Harmony borrows EmitDelegate from MonoMod 👀

[rustic needle]

HarmonyX borrows quite a lot of stuff from MonoMod with IL stuff

well, at least the ILContext/ILCursor

[faint copper]

it would, yeah, although there you don't have a Vector3 on the top of the stack for SetDestinationToPosition to use

[twilit drift]

monomod supremacy mentioned

[faint copper]

I would tend to recommend declaring a static method and calling it instead of using that

up to you though, it does make the code a little prettier to use that indirection

[rustic needle]

Why though?

[faint copper]

because from what I saw of the code, it has to statically store a reference to the delegates that are used so they don't get GCed, then make a separate function call just to get it later when it needs to be called

[rustic needle]

huh

[faint copper]

it's probably not super inefficient since it's in an indexable list but I'm a bit of a purist wrt my code injection

if I can avoid overhead I do

[rustic needle]

saving the precious nanoseconds or something 🙏

[faint copper]

yeah

premature optimization is the root of all evil but I can't help it

it probably adds up but if other people are doing it, I'm a drop in the ocean

[rustic needle]

idk, people doing inefficient stuff is what makes performance worse, like getting some gameobject in an Update method or something lol

[faint copper]

yeah

that's probably a whole lot more significant than a table lookup

[gleaming robin]

I'll test if the pretty version works first, since I have it already written out 🤭

[faint copper]

fair enough

[golden basin]

whoops...

[faint copper]

I won't fault you for that too much

[golden basin]

hamunii has 0 faith smh

i was kidding

though i still do something a bit weird in update

[rustic needle]

couldn't that be an animation?

at least the sizing up part

[golden basin]

you want me to animate the forest giant old bird driftwood etc?

[faint copper]

I also haven't looked at how MonoMod's mmhook works, but I'm a little concerned that adding wrappers for a bunch of functions adds a lot of overhead

since those are more common it would be more of an issue than any IL manipulation

[rustic needle]

MMHOOKs use a thing called the HookEndpointManager afaik for managing the hooks

[gleaming robin]

Okay, so now I need to register the transpiler, yea?

[faint copper]

in Harmony you just tell it to patch from your annotated class

[gleaming robin]

Oh yea

[faint copper]

Harmony.PatchAll(typeof(PatchClass))

[rustic needle]

But you probably shouldn't be concerned about that, since just applying and undoing hooks is expensive in general

so if there is any perfomance cost with using MMHOOK, it's nothing compared to that ^

[faint copper]

but applying/undoing would only be done once, right?

my concern is about the cost at the call site, since a fair amount of mods I think use MMHOOK to hook into Update() methods

[gleaming robin]

Harmony didn't scream at me 👀

[faint copper]

it looks like HookEndpointManager just holds a couple of dictionaries, not sure if I'm missing some references to those dictionaries in the rest of the MonoMod codebase though

[carmine shale]

Enemy escape?

That must be new

I wanna see an old bird go ham on a coil head

[rustic needle]

Yeah, you should only do it once. And you should probably have a boolean to determine whether or not your should hook should run your stuff if you need to disable it sometimes or enable it other times

[rustic needle]

There is some performance lost with detouring, but not much

[gleaming robin]

OKAY so its not overriding it, but my thing is being called!!!!!

[gleaming robin]

quite new :3

[rustic needle]

So I guess maybe ILHooks are more performant? dunno

[gleaming robin]

Alright, now that I've got the code actually firing, I'm not sure why the destination isn't getting properly overridden

also I'm renaming self to hawk to that I don't get put in jail again if I share the code

[golden basin]

the naughty word included self?

[warm nexus]

coilheads are immune

[faint copper]

I assume it only detours methods that mods add delegates to the hooks of? But even with that in mind, I feel like the number of Update() hooks could still be problematic given how attractive the easy approach is for a lot of mods, I'd be interested to know just how much overhead it is

[faint copper]

I would think so, although using a delegate in an update function probably would still introduce some

[gleaming robin]

just in case, I'll dm it to you lol

[golden basin]

lol bet

[faint copper]

I'm having trouble figuring out where the hooks are actually injected, even looking at a decomp of MMHOOK

[rustic needle]

not done there, see monomod.runtimedetour

this is what mmhook uses https://github.com/MonoMod/MonoMod/blob/master/MonoMod.RuntimeDetour/HookGen/HookEndpointManager.cs

[faint copper]

right, I was in that file and I didn't see any noteworthy references to the dictionaries it stores

oh wait, does the magic happen in the Hook constructor? lemme check

[rustic needle]

there is definitely stuff in the constructors

[gleaming robin]

@faint copper I've tried a few iterations of my code, but it seems like I need to completely replace the original call for this to work

[faint copper]

you can add branches if you need to skip vanilla code

but...I really think this would be easier if you just injected into SetDestinationToPosition...

[gleaming robin]

https://tenor.com/view/fine-jeez-metal-gear-solid-snake-gif-5014677

[faint copper]

if you do it in SetDestinationToPosition, you can just modify the destination as you see fit since it's an argument instead of a static field

and it'll take care of more AI than just the baboon hawks

[gleaming robin]

true

Should I just put in a catch-all of "if (pathIsInvalid) {return;}"?

[faint copper]

oh, that's not what I meant

you should change the destination there, not skip the function

[gleaming robin]

I'm unsure what to change the destination to tho

[faint copper]

well, if an AI wants to path to a location that is outside of its current domain, it needs to find an exit, yes?

you were trying to do something similar in the baboon hawk's DoAIInterval I believe

[gleaming robin]

ah, so maybe something like this?

bool nodeInOtherArea;
if (isOutside) 
}
  //loop through inside nodes and if any of them are within, let's say 20 units of Vector3.Distance(), set nodeInOtherArea = true and break the loop
}
if (nodeInOtherArea)
{
  //loop through outside teleports, pick the closest one, set position to it instead of the original position passed to SDTP
}
else
{
  //keep original position
}```

[faint copper]

for checking which area you're in, you'd be better off checking the nodes of both areas to see which one has a node closest to the enemy, but otherwise yeah

there's no real guarantee that a node will be within 20 meters, but you can be more sure that the closest one will be in the area that the enemy is in

if you cache an array of Vector3 for each area then you can check that very quickly

[gleaming robin]

oh fair, yea checking the closest node's tag would be way better

[faint copper]

I would keep a separate array of Vector3 for each area so that you can just loop through them really fast

[gleaming robin]

I could certainly make an array in StarlancerAIFix when I run my FindNodes method and cache it

[faint copper]

contiguous memory helps a lot for stuff like this

exactly

you already are caching the node objects, so you could probably get away with caching their locations as well

at least assuming we aren't going to see mods moving AI nodes around (which would be very strange imo)

[gleaming robin]

that would be incredibly weird lol

would it be better to make a dictionary for this?

[faint copper]

a dictionary? how would you use a dictionary?

[gleaming robin]

so that I could cache a node's position with its tag?

I'm not 100% sure of the logistics lol

[faint copper]

Vector3[] outsideNodePositions;
Vector3[] insideNodePositions;

with this, you don't need to care about tags

tags are inefficient

[gleaming robin]

yea u right

[faint copper]

getting anything from the Unity runtime is slow

.transform is slow, .position is slow, .rotation is slow, .gameObject is slow, etc

if you're doing those in a loop 200 times, it adds up

if you have an array of Vector3, all that gets put in your CPU cache and then it just blazes through the calculations

if you use a Dictionary, you're splitting those up into separate buckets which can result in blank spaces that have to be checked, or even indirection, which isn't great for CPU cache

(speaking about iterating a Dictionary that is, but getting values from every key in the Dictionary is also similarly inefficient)

if you absolutely needed to keep track of the tags (like if there were an unknown number of AI node tags), you would be better off doing either

(Vector3, Tag)[] nodePositionsAndTags;

or

Vector3[] nodePositions;
Tag[] nodeTags;

not sure which would be more efficient though

[gleaming robin]

Understood 🫡 I'll keep it simple then lol

for (int i = 0; i < insideAINodes.Length; i++)
{
    insideNodePositions[i] = insideAINodes[i].transform.position;
}```

[faint copper]

[gleaming robin]

Alright, the original code is:

public bool SetDestinationToPosition(Vector3 position, bool checkForPath = false)
{
    if (checkForPath)
    {
        position = RoundManager.Instance.GetNavMeshPosition(position, RoundManager.Instance.navHit, 1.75f);
        path1 = new NavMeshPath();
        if (!agent.CalculatePath(position, path1))
        {
            return false;
        }
        if (Vector3.Distance(path1.corners[path1.corners.Length - 1], RoundManager.Instance.GetNavMeshPosition(position, RoundManager.Instance.navHit, 2.7f)) > 1.55f)
        {
            return false;
        }
    }
    moveTowardsDestination = true;
    movingTowardsTargetPlayer = false;
    destination = RoundManager.Instance.GetNavMeshPosition(position, RoundManager.Instance.navHit, -1f);
    return true;
}```
Would I want to insert my code before, and if it determines the position is in another area it returns true earlier than the original code and thus takes precedence?

(I'd also need to toggle those two bools toward the end)

Also, am I able/would I need to ref the original position and checkForPath?

[gleaming robin]

Just for clarity, I wouldn't need to check which area the enemy is in, just what area the path leads to

[faint copper]

oh yeah, true

whoops

[gleaming robin]

that's why I was initally talking about "within a distance"

[faint copper]

I would just modify the position parameter and then let it run its course

[faint copper]

you still can't really know that the destination will be near an AI node, though it is likely that it often will be I suppose, so you could optimize it by checking if it's in a certain fairly small distance and exiting the loops early if so

[gleaming robin]

Alternatively, I could check the node nearest the position and see which nodePositions array it exists in

I think there's a method for that at least

[faint copper]

that sounds like what I suggested but without the early exit

[gleaming robin]

actually it has a break/return

[faint copper]

nonononono

that one will calculate a path

you don't need to check paths for this, that will be way slower than iterating over every node's position in a Vector3 array

[gleaming robin]

kk, I'm just trying to think of a more accurate way to determine it than a magic number distance

[faint copper]

the magic number distance would just be a shortcut, there's nothing particularly inaccurate about it unless the interior literally clips into the moon

[gleaming robin]

true

which LLL will be preventing anyways iirc

[faint copper]

hopefully, eventually

but even if it doesn't, if those AI nodes are closer than 10 meters from each other then we probably have bigger problems than an interior being on the same height as the moon

[gleaming robin]

true lol
also I do need to check for a valid path to the teleport, the poor thumper was stuttering hard when I tried to force him up a ladder he couldn't climb

[faint copper]

it certainly might make sense to do that, yeah

on most maps that would be 2-4 paths, I suppose, so it shouldn't be a huge deal

not ideal if there's another solution, though

[gleaming robin]

true, but it needs to have a valid path if I wanna set it as a destination

[faint copper]

not really

vanilla code doesn't check that most of the time

[gleaming robin]

yea but vanilla code doesn't really account for enemies being both inside and outside, except for Masked and snakes

[faint copper]

if you wanted to, you could check NavMeshAgent.pathStatus to react to a path being invalid

[faint copper]

true, but if it works in the majority of cases then there may be other solutions for the niche problems

[gleaming robin]

So instead of

NavMesh.CalculatePath(enemy.transform.position, teleport.transform.position, enemy.agent.areaMask, pathToTeleport);
if (pathToTeleport.status != NavMeshPathStatus.PathComplete && Vector3.Distance(enemy.transform.position, teleport.transform.position) < closestTeleportDistance)
{
    position = teleport.transform.position;
}```
I would set it and then check if the path was invalid without calculating it myself?
`if (enemy.agent.path.status != NavMeshPathStatus.PathComplete) { }`

[faint copper]

it's asynchronous, so not immediately

pathPending has to be false

this would be more a thing to check on the next AI interval

[turbid wave]

You have some great stuff cooking over here. Can't wait for it to release

[gleaming robin]

Ah okay. I have to go to dinner and movie, so I'll be back later to hammer away lol

[gleaming robin]

I'm 99% sure it'll be worth the wait!

[turbid wave]

When this code is triggered by an attempt to path to another area.
You'll need to record the original destination.
Slip a call to the rest of the mod to schedule the teleport after reaching the exit.
And restore the original destination pathing after the teleport happens.

With those steps, the areas would become fully seamless

[faint copper]

I don't think storing the original destination should be necessary, since the behavior state repeatedly sets the destination

if the AI is forced to update immediately after a teleport, which I believe he already made them do, then they should immediately start moving to wherever they wanted to go outside

or inside in the reverse scenario

[turbid wave]

Is that specific to the behavior of current enemies, or something that is done by the base enemyAI class?

[faint copper]

it's just a common pattern

we could definitely verify that it's consistent, but I don't think that would be an issue

[turbid wave]

Seems inefficient to spam pathing attempts, when the destination is stationary. The first pathing could be the only pathing

[faint copper]

I kinda doubt it is

I'm sure that the Unity runtime tracks whether the destination or the navmesh are modified

it would be kinda silly if it didn't

even if not, it's a vanilla thing, it's not like we can easily change that

[turbid wave]

Hmm, do you mean that calling it again with the same destination would reuse the pathing from the last call?

I have assumed that enemyAI would call SetDestination on the event. And then wait for the agent to reach the destination.

If the enemyAI class stores the desired destination and calls SetDestination on every update. Then yes we don't need to store it.

But if that is done by a behavior of each enemy, and not the class, then future/ modded enemies could break that assumption

[gleaming robin]

SetDestinationToPosition is a method of the EnemyAI class

Or type, whatever the terminology is lol

[turbid wave]

Oh I guess this is missing from the codeblock. It's calling that on the event, then position gets saved to enemyAI.destination

Then yeah we don't need to save it.

[ruby bobcat]

||Are outside enemies supposed to travel inside with EnemyEscape? xD||

[autumn panther]

Ye

[gleaming robin]

You can disable it per enemy in the config

Is there something causing a problem?

[ruby bobcat]

||Ah, nah I just saw a dog inside on Lunx's stream and was confused xD||

[faint copper]

yeah, I would expect it to avoid that work

the only time I know of that it clearly doesn't is if an invalid target location is set, because when that happens, they reset their target to some default location

[gleaming robin]

Sorry, I think I'm missing something, the function recognizes if the position being fed to it is the same as the previous position?

[faint copper]

not the vanilla function, the Unity runtime

if you were to tell an agent to go to the same place twice, I would expect it to not cancel any existing pathing

[gleaming robin]

ah

neat

that would line up with the blob's debug nav-line ceasing its flickering after I tell it to go somewhere I guess

@faint copper @copper yew @twilit drift idk what time zones everyone is in, so none of y'all might be up/available, but if I want to insert my code at the beginning of the method, would I just:

return new CodeMatcher(instructions)
    .MatchForward([useEnd = false],
    new CodeMatch(OpCodes.Ldarg_2)

Or is there some other logic I have to grab for this?

[copper yew]

if you're just running code before the original, couldn't you just do a prefix patch

[gleaming robin]

I don't know anymore ; w ;

[tulip vault]

depends

return false prefixes still run other patches applied to that function

which sometimes can be problematic

[rocky fable]

I will say I don't find the Dog going inside to be balanced

XD

[gleaming robin]

Like fr I don't know what the right answer is. I've been focused on learning transpiling bc originally I was gonna insert the code into the BaboonBirdAI.DoAIInterval(), but Zaggy suggested just preventing issues in EnemyAI.SetDestinationToPosition() in the first place, which I agree with, but we never discussed switching from transpiling

[gleaming robin]

I don't think I'd really want to return false, I would be just be adding some of my own logic (if that's how prefixing can work, which idk)

[tulip vault]

it is

[gleaming robin]

I was about to say that's your problem, but then I remembered I made a "balanced" preset, so I guess it is kinda my problem

[rocky fable]

Haha

Earlier I went into a Fire Exit and there was a dog that went inside that was instantly aggro'd and it attacked me

XD

I was like "Bruh all I did was walk inside why was that dog so angry?"

☠️

[gleaming robin]

Fair point, I'll change the default to 0 bc it is a one-hit-kill after all

[rocky fable]

Haha yeah

[gleaming robin]

If I prefix it, I need to end it with return true, yea?

[tulip vault]

if your never returning false the function can just be void

[rocky fable]

Are you planning to add the functionality for some enemies like Hoarding Bugs and Brackens to be able to spawn outside btw?

[gleaming robin]

Wasn't planning on it really, but I guess it could be within the scope. I'd probably default that to off if I implemented it

but that's a future consideration

If I can get this working, I can start tidying up for release lol

Prefixing might actually be better in this case since I could ref and manipulate the initial variable calls too, which CodeInstruction wouldn't let me do

[gleaming robin]

Running into a new error this time. The only code that's new is the caching of the positions, and the IL error points to:

// for (int i = 0; i < outsideAINodes.Length; i++)```
Is FGOWT too slow for it to correctly get a length in time?

[gleaming robin]

so the game doesn't calculate the length until after the if statement is done? I thought that it'd be usable after assigning the new value to the array

[tulip vault]

or means if any of the three are true it runs

which means it checks all three

and if that first one is null..

[gleaming robin]

I do and don't understand. The code runs top-down right? So it shouldn't be null when it gets to the loop

[tulip vault]

but if it is null

your checking it's length and first index inside of it

in the actual if statement

i think

[tulip vault]

im doubting myself a little now

[gleaming robin]

let me do a quick log test

[tulip vault]

is outsidenodePositions null?

[gleaming robin]

ohhh maybe that's it

but brb

[gleaming robin]

yep that was it

added outsideNodePositions = new Vector3[outsideAINodes.Length]; to the code and it was able to log all the positions

Code still isn't where I want it, but no more errors at least 😓

[faint copper]

a prefix would be fine here, you can just modify the position as a ref parameter to your prefix

if you use a transpiler, I think that CodeMatcher would just start at the beginning of the function, so you can immediately begin inserting instructions without calling MatchForward

[gleaming robin]

copy that

Like many programmers, I feel like my current code should be doing what I want, but somehow the initial position is still briefly getting thrown in as an actual position assignment

[Warning:Starlancer EnemyEscape] Original Position: (-35.09, -2.43, -11.97) //The position passed by the call to SDTP (baboonCampPosition)
[Warning:Starlancer EnemyEscape] Target position is outside, checking for reachable teleport.
[Warning:Starlancer EnemyEscape] Teleport found, changing destination to (-26.26, -224.23, 46.90).
[Warning:Starlancer EnemyEscape] Final Position: (-26.26, -224.23, 46.90) //The position of the teleport found by my code```

[faint copper]

wait, where? the logs you show here seem like they look about right for what I think it should be doing

[gleaming robin]

The hawk is still trying briefly to go towards the camp while its inside, which makes it path away from the teleport for half a sec, path towards for half a sec, ad infinitum

so it's like the destination is still being set to the camp somehow

[faint copper]

hmm

[gleaming robin]

also that's my point 😭

[faint copper]

yeah your phrasing didn't really make it clear what was going on

[gleaming robin]

fair lol

[faint copper]

now might be a good time to install your mod into the asset rip project to see the AI debug gizmos

but if you look the decomp, it uses EnemyAI.destination in the EnemyAI.DoAIInterval() function in the base class, which means that it sets the destination before your code takes effect every interval

[gleaming robin]

oh dammit

[faint copper]

which means that if anything sets the destination to somewhere else in the meantime that'll take effect instead

it doesn't appear that anything else assigns to destination which would interfere with it, so it should work

you should see if the destination is being set to somewhere else

[gleaming robin]

alrighty
any specific gizmo that you're thinking of in the debug thing? bc I'm using Imperium which gives me a full navmeshpath drawn out and tells me the behaviour state

[faint copper]

as in it shows the agent's path? or does it listen for some EnemyAI function and generate a separate path to display?

if you were in the editor it would execute this

public virtual void OnDrawGizmos()
{
    if (base.IsOwner && debugEnemyAI)
    {
        Gizmos.DrawSphere(destination, 0.5f);
        Gizmos.DrawLine(base.transform.position, destination);
    }
}

I don't know what Imperium uses for that so I couldn't tell you whether that is a useful debugging tool

[tulip vault]

@faint copperbtw NavMeshAIPath or whatevs its called provides .corners[] which is an array of positions used for the path

which you can use for a mroe accurate debug path

[faint copper]

yeah, certainly

that could be used to debug what the actual agent is doing, but that above code is vanilla so it would have to be injected in there

[gleaming robin]

it also lets me visualize all the nodes and stuff

[gleaming robin]

Yea I'm gonna try setting the agent.destination in my thing as well. I wasn't expecting the final destination to change in this, since I didn't set it, but I figured I'd log it anyways lol

[Warning:Starlancer EnemyEscape] Original position argument: (-54.48, -3.85, -6.51)
[Warning:Starlancer EnemyEscape] Original destination: (-54.48, -3.85, -6.51)
[Warning:Starlancer EnemyEscape] Target position is outside, checking for reachable teleport.
[Warning:Starlancer EnemyEscape] Teleport found, changing destination to (2.11, -211.73, 65.89).
[Warning:Starlancer EnemyEscape] Final Position argument: (2.11, -211.73, 65.89)
[Warning:Starlancer EnemyEscape] Final destination: (-54.48, -3.85, -6.51)```

[faint copper]

I didn't mean agent.destination, I meant EnemyAI.destination

it'd be good to figure out why the AI isn't going where you expect, since it should be going to your overridden location after 200ms

the reason I say you should look at it in the editor is then you can see specifically whether EnemyAI.destination is changing

unless you want to look at the Imperium source code to figure out what path it's visualizing exactly

what I would want to verify is that EnemyAI.destination remains what you set it to until the next call to EnemyAI.DoAIInterval

[gleaming robin]

that's actually what I did instead, but still no luck, so I guess I'll try the editor

[Warning:Starlancer EnemyEscape] Original position argument: (22.27, -2.96, 37.38)
[Warning:Starlancer EnemyEscape] Original destination argument: (-113.19, 3.04, -17.65)
[Warning:Starlancer EnemyEscape] Final Position argument: (22.27, -2.96, 37.38)
[Warning:Starlancer EnemyEscape] Final destination argument: (-113.19, 3.04, -17.65)```

[faint copper]

wait what? then what were you doing before?

[gleaming robin]

how would I detect that specifically?

[gleaming robin]

I mean instead of setting agent.destination lol

[faint copper]

you're being very confusing with the order in which you tried things and what exactly they did

pls post code for what you had before and/or now

[gleaming robin]

yea sorry, I'm throwing things at the wall and seeing what sticks lol

[faint copper]

it's very difficult to guess what's going on without knowing what precisely you're doing

there are so many ways it could do what you said without knowing that it's written the way I expect

[gleaming robin]

okay well the context of that last block is that the hawk was on its way to the main entrance teleport when it decided to switch to state 1 and then the position argument and enemy.destination ended up mismatching, so the code that was affecting the pathing was this
edit: This is in my escape component Update() triggered on a 1s interval

[gleaming robin]

This block was created by this:

logger.LogWarning("SetDestinationToPositionPrefix is trying to do something!");
logger.LogWarning($"Original position argument: {position}");
logger.LogWarning($"Original destination argument: {__instance.destination}");
bool destinationInOtherArea = false;
bool teleportFound = false;
NavMeshPath pathToTeleport = new NavMeshPath();

if (!__instance.isOutside)
{
    foreach (Vector3 node in AIFix.insideNodePositions)
    {
        if (Vector3.Distance(node, position) < 10)
        {
            destinationInOtherArea = false;
            break;
        }
        else { destinationInOtherArea = true; }
    }
    if (destinationInOtherArea)
    {
        logger.LogWarning("Target position is outside, checking for reachable teleport.");
        float closestTeleportDistance = float.PositiveInfinity;
        foreach (EntranceTeleport teleport in insideTeleports)
        {
            NavMesh.CalculatePath(__instance.transform.position, teleport.transform.position, __instance.agent.areaMask, pathToTeleport);
            if (pathToTeleport.status != NavMeshPathStatus.PathComplete && Vector3.Distance(__instance.transform.position, teleport.transform.position) < closestTeleportDistance)
            {
                checkForPath = false;
                teleportFound = true;
                position = teleport.transform.position;
                __instance.destination = RoundManager.Instance.GetNavMeshPosition(position, RoundManager.Instance.navHit, -1f);
            }
        }
        if (teleportFound) { logger.LogWarning($"Teleport found, changing destination to {position}."); }
    }
}
logger.LogWarning($"Final Position argument: {position}");
logger.LogWarning($"Final destination argument: {__instance.destination}");```

[faint copper]

where are the methods' declarations?

[gleaming robin]

Sorry, edited that first thing to say it was my escape component's Update() on a 1s interval
Second thing is the SDTP prefix

[faint copper]

can you paste the prefix declaration?

[gleaming robin]

[HarmonyPatch(typeof(EnemyAI), nameof(EnemyAI.SetDestinationToPosition))]
[HarmonyPrefix]

private static void SetDestinationToPositionPrefix(EnemyAI __instance, Vector3 position, bool checkForPath)```

[faint copper]

you didn't make your position parameter ref, it won't modify the value in the caller

[gleaming robin]

I tried that before, but I'll try it again

alright reduce me to atoms

it's sort of working

I swear I tried the ref thing before, ya gotta believe me D:

[faint copper]

it might've had some other issue keeping it from working then

it definitely has to be ref, if it isn't ref it just modifies your local argument and then tosses it

[gleaming robin]

alright well the screwiness I currently have I know the reason for, so I'll report back in a few

there's still some optimizations to be done as well, but for now my goal is functioning

[faint copper]

I wanna say your check for whether the node is inside could sometimes fail since the destination could end up being further than 10 meters from a node, why not check which set of nodes contains the closest node as well as exiting early if one is within that distance?

not immediately urgent obviously but it would probably be good to keep that in mind

[gleaming robin]

yee, good idea

[gleaming robin]

ZAGGY

ZAGGYYYY

VIDEO INCOMING IN A MIN

@faint copper @twilit drift @copper yew @tulip vault

[tulip vault]

oh god

adamance is gonna be hell

[gleaming robin]

I could cry rn

[autumn panther]

peak

[rancid brook]

you're a bad person

/s

[gleaming robin]

https://tenor.com/view/how-mad-can-i-be-playing-guitar-flying-money-money-musician-gif-17939087

[gleaming robin]

huge thanks to all 4 of you, I couldn't have gotten this far without y'all

[tulip vault]

https://tenor.com/view/the-bear-tv-show-jeremy-allen-white-bros-hugging-matty-matheson-gif-27412549

[gleaming robin]

I'm gonna go to bed now, and in the morning I'll do a brief test of the other enemies and make sure I didn't break any of them horrendously. From there I'll optimize/tidy up the code and then update \[T]/

and yes I have committed this to my local repo

[turbid wave]

Amazing dish chef, cooking well done ✅

[faint copper]

oooo very nice! glad it's working

[faint copper]

good work! next stop: actual commit messages

[turbid wave]

Although zaggy,, I thought we wanted to keep the original destination in EnemyAI.destination so that the enemy remembered where it was going even when it's not spamming EnemyAI.setDestinationToPosition()

And instead change the call that is in EnemyAI.DoAllInterval so that calling agent.SetDestination() would lead to the door

[copper yew]

LOL NICEEE

[turbid wave]

Or is there yet another location that saves our original destination?

[faint copper]

the logic that set the destination will re-run and set the same destination in almost all if not all cases, like I mentioned before, so that shouldn't be necessary

the AI interval function runs on an interval, it doesn't stop after it sets a destination

[turbid wave]

Oh so the call to SetDestinationToPosition() comes from doAllInterval()?

[faint copper]

that's where it's almost always called, yes

or in functions called from there

[turbid wave]

alright.
SetDestinationToPosition() seemed like the kind of function that an enemy mod would call once to order a pathing. And then doAllInterval would use the stored destination from EnemyAI to direct the agent

Which if there is a modded enemy that does it that way. It would get lost after returning to the intended area

[faint copper]

yeah, it's certainly possible for mods to do things differently

[turbid wave]

SetDestinationToPosition doesn't seem to be intended to be spammed every interval.

[faint copper]

what do you mean?

I saw no indication that it's not

[turbid wave]

I don't mean that it's too costly.
But it just sets the EnemyAI.destination
You'd expect that to be done to handle an event in the behavior. Like picking up scrap -> go to nest.
Then the EnemyAI can handle going there without reminders on every interval. Cuz the destination was stored in EnemyAI.destination

[faint copper]

there's no reason not to set it every frame, let alone every interval

it doesn't matter what it could be used for, it matters what it is used for

I don't know why Zeekers made it work this way, but it does

maybe there were issues with desync

[turbid wave]

I know there is no reason to avoid doing that.

[faint copper]

obviously we could store the previous destination to make an attempt to make it foolproof, but making a component for something to store that when we can't even test it seems a little pointless

[turbid wave]

But a modded enemy could do it properly. Even if Zeekers didn't

[faint copper]

if there's no bug, we can't verify the fix

it's not "properly" if the agent already does no work when setting the destination to the same location

perhaps it would work fine to do it that way, but I don't see any big problem with how it works now

[turbid wave]

It's properly, because repeating the call is not required

[faint copper]

for one thing, that would mean you would have to write a bunch of code to verify that the asynchronously generated path was valid, and if it isn't, either keep spamming it or try a different one

I've observed agents that have no valid destination continue moving, which means they have a chance of getting unstuck

[turbid wave]

I'm just predicting a bug when a custom enemy doesn't spam that method. That's all

[faint copper]

but I don't know that there's any way to recover if you detect an issue with pathing, because there is probably a root cause that needs to be addressed before writing a bunch of code in the hopes that it will allow AI to leave a stuck state that they probably entered for unforseen reasons

what I'm saying is, why should we predict a bug and fix it when we can't even verify that it will work?

the baboon hawk immediately started pathing to its nest after leaving the interior, so there is no way to know if that happened because of a "bug fix" or because it already worked

if the default settings of the mod don't make any modded enemies path between zones then it's not necessarily "supported" to enable that

so until someone reports such an issue, it shouldn't be a concern

also @gleaming robin not sure if I missed it in the code before, but are you checking whether the enemies have an escape chance before letting them change zones?

[turbid wave]

This is a potential bug. And that's all it is. What are we unable to verify?

[turbid wave]

Cause you appear to be saying that we are unable to determine what calls what.
This is a moot point really, if we wanted to get this bug demonstrated, all it would take is an enemy that doesn't repeat that call.

[faint copper]

if this bug is a hypothetical, we can't test it

[faint copper]

what do you mean "determine what calls what"? I've already looked at the references to the function and I know that the majority are called within DoAIInterval

if you want to make a modded enemy that breaks that assumption so that we can test a way that preserves the destination, then that would be cool

all I'm saying is that reworking to fix some hypothetical bug is both a waste of time and a way to introduce more issues

[turbid wave]

Is the decompiled code for LC available somewhere?

[faint copper]

it's available in a decompiler, of which there are several

[copper yew]

you just need ILSpy or DNSpy

[turbid wave]

So that I don't need to setup the modding environment to look at that stuff

[faint copper]

most people don't use an asset rip for that

[spiral ermine]

wow

yall have been suffering over here havent u

[twilit drift]

LET'S GOOOOOOOOOO 🔥 🔥 🔥 🔥 🔥 🔥 🔥 🔥 🔥

you did it!!!!!!!

[spiral ermine]

HOLY SHIT

THATS HOT AS FUCK LETS GO!!!!!!!!

[ruby bobcat]

o.o
They just hide the loot xD

[twilit drift]

make yippees and baboons fight for their loot nests 🤭🤭

[turbid wave]

hmm, probably something for FairAI to look into.
I wonder if in vanilla, a lootbug would get mad at a baboon for taking items from nest.

[golden basin]

They'd try to grab eachothers loot if encountered, and probably bug out

But won't attack eachother

I don't think zeekers intended for outside going inside and other way around

[ruby bobcat]

Idk if im just crazy but i swear i witnessed them attacking eachother when testing affliction. But then again i might be totally wrong about that xD

[golden basin]

I'm only guessing here but I'd be very surprised if zeekers implemented that lol

[ruby bobcat]

Agreed

[golden basin]

It might just be that collision = damage

[ruby bobcat]

Ill try to force it by testing just to confirm

[ruby bobcat]

Ye probably

[gleaming robin]

Ahh I was not, but that's an easy fix.
if (chanceToEscape > 0) {}

[turbid wave]

hmm, shouldn't chance to escape be applied only when randomly deciding to go to the other area?
When they have a reason to path to the other area, failing this check would just break the ai.

and if chanceToEscape is 0 indeed then they shouldn't be in the wrong area to begin with...
Ah but they could've been spawned into the wrong area by the moon author

Maybe if they get spawned into the wrong area, then their nest is also in the wrong area?

so they don't try to change area anymore

[gleaming robin]

That's a fair point

I may have misunderstood the question, but my own statement still stands that I need to implement a check in case an enemy happens to wander close enough to the entrance/exit

[golden basin]

Btw since you're including it in this thread are you making it part of ai fix?

[gleaming robin]

It's dependent on AIFix

A rudimentary 1.0 is already out

[gleaming robin]

Sorry, still waking up, I forgot I already have a conditional to destroy the escape component after creation if chanceToEscape == 0

[reef salmon]

do old birds attack outside inside enemies?

[gleaming robin]

I don't think so. That's something I wanna do for the next big AIFix update tho

[reef salmon]

oh yeah they chill

spot the thumpers, camoflauged

[gleaming robin]

Alright so there's still a thing or two that needs ironing out, but I verified that the Hawk behavior is at least mostly consistent, AND a hoarding bug that leaves and finds scrap outside will absolutely take it back into the facility

[reef salmon]

with enemies properly pathfinding in and out, will you make brackens able to do this when nobody's inside?

[gleaming robin]

I'd have to hook into it directly, so idk if it'll be this next update, but yea I was planning on it.

[reef salmon]

very epic

[rustic plume]

its already good

enough

please update it

[reef salmon]

for some reason the individual enemy chance configs are gone

[gleaming robin]

gone, or missing descriptions? Cuz I've had that happen to me

[reef salmon]

gone

[gleaming robin]

huh, that would imply that they all failed to register

did you have any errors?

[rustic plume]

u left the game when it was creating the config

thats why

or its a thunderstore issue

[reef salmon]

only one pack of errors, which is i think incompatibilities between terminalplus and LLL

[rustic plume]

refresh the config page

[reef salmon]

i've launched the game like 10 times while it's like this and done full rounds

[rustic plume]

delete the config

and open the game again

[reef salmon]

i did

didn't work

[rustic plume]

weird, for me it worked

[rustic plume]

release the update already @gleaming robin

😭

[twilit drift]

🤔

[faint copper]

pal you need to stop

[faint copper]

did that happen after deleting the config? BepInEx configs are not supposed to clear out unused config options

also, a log would be useful, I believe the hook to register them was in GameNetworkManager.Start() or something similarly required for the game to function

[reef salmon]

randomly the descriptions for the configs were cleared, which usually means the setting doesn't exist anymore, so i tried deleting the config, which regenerated it with only that

[faint copper]

gotcha, that makes sense

would've been good to know first of all if the game was actually functioning, and second of all whether the enemies' configs are registered when they spawned

[copper yew]

wait outdoor thumpers is actually kinda crazy on maps like this LMAOO

[faint copper]

because I believe Audio Knight put in a way for the enemy configs to be bound when they spawned as well (I hope)

[gleaming robin]

Correct, I have two entry points into the bindings

[reef salmon]

hmm, will enemyescape allow masked to exit through fire exits aswell instead of just main?

[gleaming robin]

They can't?

[reef salmon]

they can only go through main

also what in the world

when did the configs start existing

and why does only the flowerman one exist now

also i believe i've found an incompatibility between this and snatchinbracken, if the bracken drags you inside, the game still thinks the player is outside so you get outside lighting and the clock and stuff

[ripe flax]

does this mod effect enemy spawn rate?

[gleaming robin]

Ahh that would make sense

[gleaming robin]

Not at all

[faint copper]

that's likely a combination of an issue with that mod and CullFactory unfortunately

[spiral ermine]

https://tenor.com/view/parasyte-lock-in-locked-in-serious-meme-gif-13189278796605253304

wait

what

[faint copper]

they'll need to override SetEnemyOutside and update the flag of the player when dragging them

[reef salmon]

i don't have cullfactory

[faint copper]

what do you mean by outside lighting then

[reef salmon]

the sunlight clips through the walls

[faint copper]

hmm lemme check something

wow, I didn't know the sun turned off when you entered the interior

but I suppose that makes sense considering that it has a non-shadowed light for indirect lighting contribution

[faint copper]

this still stands though, the flag needs to be set in SetEnemyOutside to account for this

[gleaming robin]

Sorry, is that something I need to do, or SB?

[faint copper]

Snatchin Bracken does

you can't know that the bracken is teleporting a player, unless you make a soft dependency on their mod

[gleaming robin]

I will say convincing a bracken to leave the facility is harder than it would seem. I might have to transpile it lol

later tho

[faint copper]

trying to make it randomly do so you mean?

or I guess you mentioned making it leave if no players are inside

definitely seems like something you'd have to do a little IL injection for, yeah

[gleaming robin]

Yea it's glued to its favorite spot while no player is inside. I tried changing the favorite spot to a teleport, but that didn't seem to do anything. I'll work at it more later. For now...

⚙️ ⬆️

[faint copper]

hmm, are you sure it was trying to go to its favorite spot and not just finding the farthest node from the entrance? I forget which it does in that situation but it does like to do that

[gleaming robin]

oh maybe that's it

regardless lol

[gleaming robin]

Starlancer AIFix v3.6.0 | EnemyEscape v2.0.0

• StarlancerAIFix v3.6.0

  • When finding the AI nodes in the level, AIFix now also caches their locations. This was done primarily for EnemyEscape, but it offers a non-zero performance boost for AIFix as well.

• StarlancerEnemyEscape v2.0.0

  • Almost a complete overhaul from v1.0.0
  • Instead of only allowing enemies a chance to use the EntranceTeleports if they happen to wander close enough, a system has been implemented to make an enemy path directly to an EntranceTeleport and warp to its match on the other half of the level. The range at which they may search for a teleport and the cooldown for an attempted search are configurable per enemy.
    • "Chance To Escape" now represents the chance for them to initiate pathing to the teleport.
  • Implemented a HarmonyPrefix to EnemyAI.SetDestinationToPosition() in order to prevent enemies from getting stuck if the position they want to path to is on the other half of the level.

Source code is now available :3c
https://github.com/AudioKnight/Starlancer

AIFix's source code is so tiny in comparison to EnemyEscape 😅

Don't Touch Me is compatible now btw. I don't know if it can actually move, but at least it doesn't break EnemyEscape now 🤭
@spice kindle @tired ore @edgy topaz

[copper yew]

yippieeeeeee

[gleaming robin]

you get to see how my fevered mind codes

[copper yew]

[edgy topaz]

Yippie

[tired ore]

Thanks I hope it doesn't break anything else

[dusty jewel]

Ooh wait, does that mean monsters can actually chase you in/out of the facility now?

[golden basin]

Last I checked I think he did it based on their behaviour, which is when they're not chasing you, but I could be wrong it's been a while + I can imagine them chasing, you leaving, losing aggro, then they leave, see u again, chase again

[gleaming robin]

Xu is mostly correct. If an enemy is mauling you to death, it won't leave. But if you go outside and its ai continues to target you (or your position) it'll come after you

[spiral ermine]

Yall have done it

Truly the opps can catch you lackin anywhere

[gleaming robin]

The size of this changelog truly underscores the 4 days of work I put into v2

[short lagoon]

Wait ur doing modded monster compat?

[gleaming robin]

Mod enemies are automatically registered, but they default to 0% chance to escape

[short lagoon]

Nice!!

Dude Imma start using this

The fiend outside would be so mortifying

[gleaming robin]

every enemy is registered by my mod, just some like the lassoman and red pill unimplemented vanilla enemies are blacklisted

even though in my code I was silly and called it a whitelist lol

[short lagoon]

So we can put monsters on a blacklist and whitelist too

I fucking love this

[gleaming robin]

well on the user end it's just setting the escape chance to 0 or not 0 🤭

[short lagoon]

Real

[gleaming robin]

the interior/exterior range and cooldown time are also configurable per enemy

[short lagoon]

Ooooo

[gleaming robin]

read the readme 🤭

[short lagoon]

I will

[reef salmon]

my config is still broken with 2.0.0 🥹 give me my config back

[reef salmon]

also, how does chance to escape work? like, when does it roll the chance?

[golden basin]

When I saw it being discuss, it was a dice roll once every cooldown of 10 seconds when it qualifies

[reef salmon]

every 10 seconds is actually pretty frequently

[gleaming robin]

pathing range and cooldown is configurable per enemy

try deleting your config

I've tested multiple times and it should actually re-check the config each load

idk why yours is busted

[reef salmon]

i have 🥺

[gleaming robin]

go into your profile, go into the bepinex folder, into the config folder, and make sure it's absolutely gone

actually

what it looks like is something is preventing it from registering

are you getting an error after you click into the main menu?

or do you have a fast startup mod on?

[reef salmon]

nope

[reef salmon]

testing right now

[16:22:52.0191936] [Info   :Starlancer EnemyEscape] Starlancer is allowing enemies to roam.```
these are the only two logs that mention enemyescape

[gleaming robin]

nothing errors after clicking Online?

[gleaming robin]

send me your profile code

[rustic plume]

i see why

a enemy need to escape only once to the full config appear

[gleaming robin]

nope, it registers all enemies after clicking Online

[rustic plume]

then idk

[reef salmon]

018f0bcc-efdb-b682-008e-00ba9f8fe128
note: i manually imported the beta v50 version of lethallevelloader, which isn't there in the code

[gleaming robin]

I've tested it with LLL v50 and it worked

well

lemme test it with a deleted config and make sure it regens

I just made sure it didn't crash 🤭

[reef salmon]

https://tenor.com/view/wtff-cat-confused-orange-gif-25858653

[gleaming robin]

well it's not LLL at least

gonna test your profile

[rustic plume]

@gleaming robin

yo

the hoarding bug isn't giving any damage to non-hosts

(only the host is with the mod)

[reef salmon]

hm, that's an issue i know lethalescape had (hoarding bugs only deal damage to host)

[gleaming robin]

why does only the host have it? I never said anything about it being server-side only

[rustic plume]

because lethal escape is host only

and this is really similiar

[gleaming robin]

that's a bold assumption to make

[rustic plume]

well, you can add support to serverside

[gleaming robin]

no

just have everyone download it

and if the issue persists I'll look at it

[rustic plume]

k, i'm gonna stay with lethal escape then

[gleaming robin]

okay bye

[reef salmon]

it modifies enemy ai (i think), so that's impossible

[rustic plume]

no

lethal escape is serverside

and it takes the ai ownership

[reef salmon]

lethal escape just teleports enemies around

it doesn't modify ai

[rustic plume]

k

[rustic plume]

you think?

[gleaming robin]

ehhh yes and no, it does set them outside, which is why AIFix never really touched it, but I can't say for sure exactly if/how it modifies things

[rustic plume]

the mod is semi-serverside

[short lagoon]

I thought it was host only

[rustic plume]

they see the hoarding bug outside

it just doesn't give damage

[gleaming robin]

regardless, unless a mod author says explicitly that a mod works when only the host has it, I wouldn't assume that it is. It's a 30kb file, there is no harm in everyone having it

[rustic plume]

nah bro, then i'd use lethal escape

[gleaming robin]

goodbye then

You've been very impolite here

[rustic plume]

cya

[gleaming robin]

and I'm rather tired of you making demands and telling people you know what my mod does as factual information

[short lagoon]

Send him to the backrooms

[gleaming robin]

I try to be very patient, but my patience has limits

[short lagoon]

Understandable

Im naturally not a patient person but I’ve been trying really hard

I’ve been doing pretty good

Simply asking questions about mods so I can know it’s progress

It helps a lot

You’re doin great Audio Knight

Don’t worry bout them haters

[spiral ermine]

Dickhead on my block?

https://tenor.com/view/ltg-now-low-tier-god-gif-3446062039083503850

[short lagoon]

woah? I see you reacting, piss off now

[spiral ermine]

What we don't tolerate around here is unneeded rudeness to devs like Knight

[short lagoon]

Yeah

[stiff pulsar]

Here if you need me

[tired ore]

bailiff, whack his pp

[gleaming robin]

@reef salmon
try disabling Rosie's moons

_>

[reef salmon]

WHY DID THAT WORK

[gleaming robin]

@ruby bobcat WALL TURR- I mean, your moons break my mod

[reef salmon]

i can only guess it's because of the fancy subspecies or whatever (i think rosie's moons has those), but i think those are just texture changes to map prefabs

[gleaming robin]

I think it must be her custom enemy types which are modifications of vanilla enemies?

ye

at least we found the issue

[reef salmon]

why would a texture change to a prefab break it though

[gleaming robin]

it's something in the scriptable object I think, but we'll wait for her before further speculation

[reef salmon]

ohh i see how it works, every 10 seconds while they're in a RANGE of the entrance, not always

that makes alot more sense

oh wait no cooldown time depends on enemy

it can vary!

[gleaming robin]

Yep! Fully configurable

@turbid wave I forgot to tag you, but you'll be happy to hear that it's all config'd

[reef salmon]

default 0 chance for mouthdogs to go inside, cowardice

[gleaming robin]

only because they're a one-hit-kill lol

but that's the beauty of a config, you can jack it up to 100 and give it 9999 units of pathfinding range

[reef salmon]

absolutely doing that for bracken

actually if a bracken is chasing someone but in range of teleport detection, will it go for the player or the teleport?

[gleaming robin]

bracken doesn't do random pathfinding yet

player

you should read the readme for the known issues btw

(everyone should read the readme)

[stiff pulsar]

It's in the name

[gleaming robin]

?

[stiff pulsar]

Read me!

[gleaming robin]

oh

yee

I'll have to transpile the bracken code to get it to allow EnemyEscape to tell it what to do

it's real stubborn

[reef salmon]

what! does this mean

[gleaming robin]

exactly what it says on the tin

it can still potentially follow you out

but like I said, it's stubborn

[reef salmon]

oh if it rolls the chance even with 9999 range, it'll just not go there?

is that what it means?

[gleaming robin]

currently, my code just ignores Bracken for the random pathfinding, before the diceroll, to not waste any computational effort on it
bc even when I tell it to do something, it doesn't

[reef salmon]

then transpile it! i want the bracken to be a fiend

[gleaming robin]

Soon!

[reef salmon]

now!!!!

[gleaming robin]

I just spent 4 days on this, lemme rest 😭

[reef salmon]

[gleaming robin]

alternatively

https://tenor.com/view/jack-sparrow-gif-22521430

[reef salmon]

also from what i've heard, hoarding bugs and baboon hawks, if they go inside/outside, they bring stuff to their respective nests?

[gleaming robin]

yep

[reef salmon]

so it goes

[gleaming robin]

.

vid there of a hawk doing such

[reef salmon]

i like how he sleeps at the end of the clip, job well done

will you perhaps consider modifying the snare flea code to make them actually work outside?

[gleaming robin]

Wdym? In what way do they not work?

[faint copper]

wait, isn't the description for the preset wrong here? @gleaming robin isn't it supposed to only take priority over values set to -1?

also if the author of LethalEscape said only the host needed it that's putting a lot of faith in the networking code of LC lmao

all enemies that I've looked at in the code deal damage on the client side based on the target they have chosen there, so if the target is mismatched between host and client they become useless

[gleaming robin]

I guess the wording is a little unclear, but my meaning here is that values manually set take priority

[faint copper]

ohh

that makes sense, I guess that's partly my sleepy brain's fault

[gleaming robin]

Lol no worries. The description of each enemy config escape chance says it again in what might be slightly clearer terms

[turbid wave]

Yeah, I've been reading everything on this thread, so I've seen that as soon as you implemented it. Now I just need to make my own preset and wait for other mods to release for v50.

Oh one thing that didn't end up being configurable. I wanted to give enemies a higher chance of return than that of escape. And currently there is a single value for both.

You deserve a rest tho. Job well done with this mod. You can just stash that suggestion for later.

[gleaming robin]

That should be easy to implement, I'll look into adding it with the next update :3

[golden basin]

can you make it so that the AI is just naturally better? thanks

[opal bane]

[19:12:13.6619438] [Error : Unity Log] NullReferenceException: Object reference not set to an instance of an object
Stack trace:
EnemyEscape.StarlancerEscapeComponent.BindRegisteredEnemies () (at <a5a07d1df715484e849d3d540a9d2ca0>:IL_0016)
(wrapper dynamic-method) GameNetworkManager.DMDGameNetworkManager::Start(GameNetworkManager)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.TrampolineGameNetworkManager::Start?-1330587096(GameNetworkManager)
LethalLib.Modules.NetworkPrefabs.GameNetworkManager_Start (On.GameNetworkManager+orig_Start orig, GameNetworkManager self) (at <c68aa40cbbae4bd69cd2fcd50c8f1ae1>:IL_0000)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.HookGameNetworkManager::Start?1654550456(GameNetworkManager)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.TrampolineGameNetworkManager::Start?-367031872(GameNetworkManager)
LethalBestiary.Modules.NetworkPrefabs.GameNetworkManager_Start (On.GameNetworkManager+orig_Start orig, GameNetworkManager self) (at <0d184de3cadb4a4a9b8ad443576d9910>:IL_0000)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.HookGameNetworkManager::Start?942834480(GameNetworkManager)

[19:12:13.6619438] [Error : Unity Log] NullReferenceException: Object reference not set to an instance of an object
Stack trace:
EnemyEscape.StarlancerEscapeComponent.BindRegisteredEnemies () (at <a5a07d1df715484e849d3d540a9d2ca0>:IL_0016)
(wrapper dynamic-method) GameNetworkManager.DMDGameNetworkManager::Start(GameNetworkManager)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.TrampolineGameNetworkManager::Start?-1330587096(GameNetworkManager)
LethalLib.Modules.NetworkPrefabs.GameNetworkManager_Start (On.GameNetworkManager+orig_Start orig, GameNetworkManager self) (at <c68aa40cbbae4bd69cd2fcd50c8f1ae1>:IL_0000)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.HookGameNetworkManager::Start?1654550456(GameNetworkManager)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.TrampolineGameNetworkManager::Start?-367031872(GameNetworkManager)
LethalBestiary.Modules.NetworkPrefabs.GameNetworkManager_Start (On.GameNetworkManager+orig_Start orig, GameNetworkManager self) (at <0d184de3cadb4a4a9b8ad443576d9910>:IL_0000)
(wrapper dynamic-method) MonoMod.Utils.DynamicMethodDefinition.HookGameNetworkManager::Start?942834480(GameNetworkManager)

[ruby bobcat]

?! How wtf

Like what even broke?

[ruby bobcat]

Oh wait

V50?

[gleaming robin]

Are you running Rosie's moons?

[gleaming robin]

ye

[opal bane]

yes