archived-code-advanced

[hushed moth]

I get it, but still, I can receive almost real-time webrtc media stream in 1920x 1080 w/o problem, but when I change to 4K the fps goes down to 1

[undone coral]

you should confirm hardware decoding is used for 4k. the sending server has to use the correct SDP for it

there's a level limit for it

did you author the webrtc signaling yourself?

@hushed moth

or are you using a pre-existing relay of some kind?

or hwat is it?

[hushed moth]

lets say half-half
I had a problem with the encoder, in obs, if the encoder is nvidia nvec x264 then google blocks it, it only works with x264 encoder + profiler = baseline

[hushed moth]

looks like the ffmpeg recode solved the problem, but still, I dont think that setting the texture in each receive is a good solution

[undone coral]

okay i really want to help you

but all your answers have been like this

you have to just say it what it is or that's it

i can't help you if you don't tell me straight clear answers with truth

[hushed moth]

thanks for the tip, i will try it in the future

[undone coral]

and i'm the only person who knows about any of this stuff

so it's your call

[hushed moth]

i will also look into the suggested branch

[undone coral]

okay, i think i can't help you anymore i'm sorry

[west jolt]

Hey guys, is it possible to clone a prefab and make changes to it before instantiating it?

[quartz stratus]

Why not create a prefab variant, if your changes are that significant? Otherwise, making changes immediately after instantiating should get the job done. I would argue that if you have to change so much about the prefab that that solution is not viable, it should be a variant.

[jolly token]

You can't make a prefab on runtime.
However you can instantiate a inactive game object and use it as prefab 😌

[west jolt]

It's a small change, but this change is always different, which is decided at runtime. Alright, I guess there's no other way besides instantiating it first in a way that allows me to do the changes afterwards. Cheers :)

[gilded shuttle]

damn a real discord mod uh oh

[gilded shuttle]

Good point my friend thank yoh

[warped mural]

Hello all, I am working on a particularly mathematic game in Unity and I need to write some tests, I wasted far too much time tracking a math mistake through the grapevine.

I want to ask if it's possible to add each MyClassTest next to MyClass so that I have them together at all times, regardless on how I choose to refactor the folder structure.
I saw some asmdef shenanigans are required so I am thinking perhaps it's impossible. But I still want to have it like this as my scripts hug my prefabs where I use them 1:1.

Thank you!

[torn basalt]

Sup room

[jolly token]

You need to, and would want to separate assembly for Runtime and Test. Test classes tends to not exactly match with runtime classes anyways. I recommend you to use Test Framework package, if you're not using it.

[rich estuary]

Heyas. So wondering if anyone has had an issues with using Unity's Device Unique Identifier (UnityEngine.SystemInfo.deviceUniqueIdentifier). This is supposed to give a unique HASH from their device hardware (except for iOS), but I'm starting to think that it's maybe not unique as the documentation does not mention anything for Linux based platforms (https://docs.unity3d.com/ScriptReference/SystemInfo-deviceUniqueIdentifier.html)
I have implemented an auto login system based off their device Id for the game I'm working on (using PlayFab), but now that we have a higher influx of players after going public free-to-play, I'm noticing some players can login to another players account due to possibly having same deviceId, so if anyone has a better understanding on how this device Id works, that'd be appreciated.

[pliant crest]

why not just instead of using a gimmick like this

just have a token

[jolly token]

It is not good idea to use something that user can easily fake like deviceUniqueIdentifier to authentication. Even if it is guaranteed to be unique, that does not mean it's going to be secure.
Anyways, the doc says it will return unsupportedIdentifier if not supported, so you should check the value on Linux.

[pliant crest]

this is pretty common pattern however

inheritance based game design

[jolly token]

Convention-wise I like this better
other.TryGetComponent(out ISomeInterestingInterface interfaceCommonName)
Anyways of course, that's good way of abstract components 😄

[pliant crest]

from a C# standpoint always use the generic method if you can

[jolly token]

It is generic method

[pliant crest]

thats not what generic means thou

😐

you are using the literal definition i'm assuming

yeah this sounds good in theory

eventually you hit inheritance hell

and go there must be a better way

and thats where entities come in

with the whole ECS pattern

yeah i understood your intention

the main problem with inheritance is, its really hard to build an object dynamically

you can't create classes dynamically thou?

unless you pre create all combinations of them

of course

[jolly token]

ECS is great but it's not swiss knife

Always can do AddComponent 😄

[pliant crest]

yeah but thats no longer inheritance

so if you wanted a collision handler on your object now

you have to go through components

i'm just referencing this

on collision grab the ICollision handler which you can't dynamically add to your object

[jolly token]

Inheritance is one of tools to abstract things

[pliant crest]

unless you add the component

[jolly token]

There is no silver bullet 🤔

[pliant crest]

what do you mean who cares about dynamic lol

i'm not talking about literal dynamic keyword

i mean like if you had a rpg player who suddenly got a shield

you can just add IShield or something

but only as a component

shrugs does it matter?

yeah but what i'm saying is you are already using the C in ECS

[jolly token]

I contribute to Svelto.ECS and there we don't allow dynamically adding component
It's not really ECS thing, it's just different pattern
If you need to add or remove some feature dynamically, you need different approach

[pliant crest]

wait just to confirm

you have an entity and it must be created with all components

so you can't add or remove components from that entities life time?

[jolly token]

It needs to build with pre-defined entity descriptor.
Yes you can't add or remove components

[pliant crest]

hmmm

i guess the question is why

sounds like an arbitrary restriction

[jolly token]

It's for performance, To optimize memory structure

[pliant crest]

but shouldn't that be left up to the consumer to decide

whether or not they want that particular entity to be optimized or not? (in that way)

[jolly token]

But by supporting it we'd have to support different type of memory layout 🤔

[pliant crest]

well yes true in your case

[jolly token]

I'm thinking it as characteristic of Svelto.ECS

[pliant crest]

how do you add or remove functionality from an entity - unless you have to destroy it and recreate it?

and if you have to destroy it and recreate it how does caching work

cause alot of ecs is reusing memory space

[jolly token]

Have flag in component, or put it in filters, or change groups that can be queried from

[pliant crest]

so that means every single functionality possible must be attached to the entity

and just have something to disable it (flag)

[jolly token]

If you're curious, see articles linked, it may make more sense https://github.com/sebas77/Svelto.ECS

Yes

[pliant crest]

i guess its just a different way of thinking

not a fan of it however

i could see the benefits for extreme performance

[jolly token]

Yeah you can get very optimized memory layout if everything is statically determined

It may cost your productivity, yes 😌

[pliant crest]

also costs query time

if there are lets say 100,000 units with some sort of "debuff" component even thou they aren't active

you'd have to find the active ones

unless you track that real time

but then that costs you space

[jolly token]

Depends, user can move entities between query groups.
So user can have debuff-active group and only query from there

[pliant crest]

yeah its kinda like indexing

so its either cpu cycles or memory (small amounts)

[jolly token]

Kinda. It still supports indexing feature separately to support some less optimal cases
But yeah always have trade off

[undone coral]

deviceUniqueIdentifier on iOS by default will be unique per installation of the app

if you uninstall and reinstall it on iOS, it will not be the same

unless you obtain the right entitlements

you are better off generating an ID and storing it in the app's documents for it to persist

you should probably use an identity service's pre-existing guest or device accounts. for mobile users this is usually centered around a message to their phone number

[undone coral]

i wouldn't overthink it. you don't have to use unity testing framework as your testing approach

you can also write whatever you want wherever you want, and call it wherever you want

if you want the tests in the script, go for it

[rich estuary]

Yeah for iOS I'd use the user's App Store I'd, as that seems to be the safest way

[rich estuary]

I'm thinking that I should also get the users MAC address, add it to their unique device ID string, and re hash it to use that as the auto login key. But yeah, not sure if that's really that secure still

But it would mean that users on a platform that does not support unique identifier would mean it is now unique

[pliant crest]

this seems overly complicated

for no reason

just given them a login token and save it to their system if they want to remember me (after successful login)

cause what happens if they do this on a public device

kinda stupid

[rich estuary]

Right yeah good point, like at a cafe.
I kinda wouldn't just thought that storing their login token in their options file for example would be bad as well, incase someone else gets their hand on options file and just copies the key

[pliant crest]

if you hit remember me

on a device someone can steal

thats kinda on you

but thats true for any login token right now

like i could ask for or steal your discord token

and just pretend to be you

(just to clarify this is hypothetical and i in no way want it)

[rich estuary]

Yeah fair enough. Good point.

[pliant crest]

the good thing about tokens is you can have expiry dates and people can invalidate them

cause you can just generate a new one every x days

[rich estuary]

So you'd recommend generating a token separate from any device identifiers, and just storing that if they hit remember me

[pliant crest]

yeah

this way you can handle collision server side as well

(in the rare case it does)

[rich estuary]

Yeah makes sense.
Do you have a recommendation for how I should generate that token? (C#) Like basing it off certain hardware serial numbers, or something like the current date where it's a one-time creation that will be completely unique?

[pliant crest]

eh

just generate a guid

[rich estuary]

Yeah had a feeling youd say that haha

[pliant crest]

Guid.NewGuid()

lol

[rich estuary]

Yep

Thought so

[pliant crest]

you want it to be changable

so you don't wan it to be based on anything

[dapper cave]

is DrawInstanceMesh still the fastest way to display many times the same thing or is there a faster way now in 2021?

[livid kraken]

So lets say I want to pass a NativeArray<uint> to a c++ funtions that has unsigned int* destination as the parameter. What I have to do is (IntPtr)NativeArrayUnsafeUtility.GetUnsafeBufferPointerWithoutChecks ( myArray ) ; Is this correct ?

how do I declare the function on the c# side? I dont want to wrap everything in unsafe, passing the IntPtr results in a sigsev. Not using NativeArrays and simply passing a uint[] works fine

Idk what I'm doing wrong with the nativearrays

[maiden turtle]

probably make a couple of adapters for the functions that you need, taking in native arrays

or use unsafe

[stray torrent]

Hi guys, I'm making a unity project that requires external dependencies aka plugins. I would like to know what is a way in unity to deal with external dependencies. Is there a some kind of package manager as npm for unity? So any new developer, who clones my project can install all required dependencies with a single command. Because now I'm forced to install all dependencies manually by clicking on unity package files and importing it.

[maiden turtle]

afaik no, you have to add them to source control

[stray torrent]

and what if my packages are like hundreds of megabytes?

[maiden turtle]

git lfs

[stray torrent]

Got it, thanks

[pale pier]

Unity does have a package manager, it's somewhat new and less feature rich than npm but it's viable.
If you install any packages via it then they will be cached in your "Library" folder and added to a JSON file in the "Packages" folder.
https://docs.unity3d.com/Manual/Packages.html

[stray torrent]

as far as I see not all packages support it, right?

For example Zenject doesn't or does partially

[livid kraken]

Thats the problem if I pass a NativeArray instead of a simple uint[] the data on c++ side gets jumbled idk why. So I think Im forced to pass in the pointers explicitly

[maiden turtle]

show more code

maybe you don't convert the length right or something

you can pass the pointers, just make a wrapper function that encapsulates the pointer business so that you don't have to put unsafe everywhere but just in that function

make the function take a native array

and do the pointer business there

[maiden turtle]

if it's just code, consider forks + git submodules

[stray torrent]

good point

[maiden turtle]

if it's nugets, I'd suggest managing them with a custom script

nugetforunity exists, bit it's meh in my experience

[stray torrent]

so I can just add zenject git repo as submodule to plugins folder

and whenever I clone project again, I'll just go git submodule update

[maiden turtle]

basically yeah

[stray torrent]

and it will fetch it for me

[maiden turtle]

if it's possible, it's usually a good approach

[stray torrent]

but what about building library?

[livid kraken]

Give me a min I will show you the exact code I was away for lunch

[stray torrent]

Are there any pitfalls with that?

[maiden turtle]

what's that?

[somber swift]

DrawMeshInstancedIndirect could be even faster

[stray torrent]

well, if I install library from unity asset store / or by downloading it manually, it comes like a built binary, no?

[maiden turtle]

oh you mean restoring the cache

[stray torrent]

while in gitsubmodule it's just a code

[maiden turtle]

so what? is that bad?

if it's in a separate asmdef, it would get cached and only build once anyway

[stray torrent]

I mean that code would be build on my local machine and there might be problems with environment or no?

[maiden turtle]

i guess it depends on the project, but I'd expect it to just work

[stray torrent]

okay

[livid kraken]

@maiden turtle so this is the function on the c++ side c++ size_t meshopt_simplify(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, unsigned int options, float* result_error); I have it wrapped like this cs [DllImport("meshoptimizer.dll", CharSet = CharSet.Auto, CallingConvention = CallingConvention.Cdecl)] private static extern UInt64 meshopt_simplify(NativeArray<uint> destination, NativeArray<uint> indices, UInt64 indexCount, NativeArray<float3>[] vertex_positions_data, UInt64 vertex_count,UInt64 vertex_positions_stride,UInt64 target_index_count, float target_error, uint options,float[] out_result_error); if I use a simple uint[] instead of the native array stuff it works as expected so I'm guessing the pointer passed to the native array does not point directly to the data

[maiden turtle]

is it extern(C) on the c++ side?

[livid kraken]

yes

[maiden turtle]

it's weird to me that it works at all. I'm not sure on the details of function signature mapping, but to me these two look completely different. an int* is not the same as a NativeArray<int>. let me consult the docs

a native array is pointer + length

afaik

[livid kraken]

well the function itself is made to take the pointer of the first element and the size of the collection and iterate over that

so it works exactly as expected when passing in a uint[] is it's a pointer to the first element anyways( I think) not sure what NativeArray points to

I want to get the NativeArray part down since I'm going to have to spin up jobs calling this dll in the future

and I have to use this new mesh API

[maiden turtle]

it would appear to me it passes the pointer to the array object, not the array data. have you actually tried iterating over data and confirm it works?

[livid kraken]

iterating over what data ?

[maiden turtle]

if you pass a uint[], you say it works. by that you mean the code on the other side gets a valid pointer?

[livid kraken]

yes it also returns valid indicies that build a valid mesh

[maiden turtle]

because arrays are reference types in c#

aka pointers

[livid kraken]

so I can confirm that passing uint[] works fine

also it seems I cant pass a IntPtr struct to a int* in c++

[maiden turtle]

again, idk the details. ehat I'd do is I'd replicate the signature perfectly, make it unsafe, and write out a bunch of adapters for native arrays, then it would work for sure

I can't find the relevant docs on this marshalling, i think it's called, rules right now

the adapter would just call the actual binding, having extracted the pointer from the native array that was passed in

[livid kraken]

I wanted to somehow avoid using a unsafe context but looks like it cant be avoided

[maiden turtle]

just use it in the wrapper function

you can call it normally in a safe context even if the function is unsafe

unsafe doesn't mean you can't use it in safe code

[pale pier]

Yes, most people don't support it

[floral gulch]

get_gameObject and get_transform allocate 40 B, is that a recent regression?

[abstract root]

I have also noticed 40 B GC in OnTriggerEnterExit events when calling collider.TryGetComponent. Its get_gameObject that allocs 40 B

[floral gulch]

hmm, is this related to my question?

[undone coral]

you would need to declare ref NativeArray...

then it iwll work

the correct way to use a nativearray in your extern is

[livid kraken]

I just did a test with ``` [DllImport("meshoptimizer.dll", CharSet = CharSet.Auto, CallingConvention = CallingConvention.Cdecl)]

private static extern void myTest(uint* destination, uint count);





public static void NativeArrayTest(NativeArray<uint> data, uint count)
{
    uint* ptr = (uint*)NativeArrayUnsafeUtility.GetUnsafeBufferPointerWithoutChecks(data);
    myTest(ptr, count);
}```

and this works

it adds 1 to the native array and prints it out from unity

[undone coral]

i was writing the declaration

[livid kraken]

your telling me that if I do ``` private static extern void myTest(ref NativeArray<uint> destination, uint count);

[undone coral]

you would use (IntPtr)data.GetUnsafePtr() in an unsafe context

in unity

[undone coral]

it's nuanced

i wouldn't use the ref keyword at all here

[livid kraken]

what I did in my test is what you suggest right ?

[undone coral]

yes but i would use intptr

[livid kraken]

let me test with intptr

[undone coral]

because that's the actual type that is passed to your extern

[livid kraken]

should be UIntPtr right ?

[undone coral]

no

just intptr

[livid kraken]

cause the c++ side expects unsigned int*

[lusty salmon]

Good day all! I am attempting to automate a build process for hundreds of webgl mini builds. Its a single scene. Set up the decorations, build, new decoration layout, build, repeat. Using the official documentation and each time I get error CS0234: The type or namespace 'Build' does not exist in the namespace 'UnityEditor' (are you missing an assembly reference?)

Any ideas on how to overcome this? Is this a known issue? So far my reseach has not returned much.

Link to the doc:
https://docs.unity3d.com/2020.3/Documentation/ScriptReference/BuildPipeline.BuildPlayer.html

My code:

using UnityEditor;
using UnityEditor.Build.Reporting;

...

[Button]
    public static void TestBuild()
    {
        BuildPlayerOptions buildPlayerOptions = new BuildPlayerOptions();
        buildPlayerOptions.scenes = new[] {"Assets/Scenes/Dev.unity"};
        buildPlayerOptions.locationPathName = "Assets/Builds";
        buildPlayerOptions.target = BuildTarget.WebGL;
        buildPlayerOptions.options = BuildOptions.None;

        BuildReport report = BuildPipeline.BuildPlayer(buildPlayerOptions);
        BuildSummary summary = report.summary;

        if (summary.result == BuildResult.Succeeded)
        {
            Debug.Log("Build succeeded: " + summary.totalSize + " bytes");
        }

        if (summary.result == BuildResult.Failed)
        {
            Debug.Log("Build failed");
        }
    }

Any help would be greatly appreciated!

[undone coral]

intptr is just void*

as a value

since structs are passed by value in c#

[livid kraken]

so then who will do the vast from void* to unsigned int* ?

[undone coral]

it is a static cast, they are the same 👀

x-files music

[undone coral]

you've basically figured it out

[abstract root]

try putting your script inside folder named "Editor"

[undone coral]

the only reason to do this is to make it bulletproof

this is the example in the unity source i found

https://github.com/Unity-Technologies/UnityCsReference/blob/66ae70d07033d2275ef6d8b42614367747439e58/Modules/ParticleSystem/ScriptBindings/ParticleSystem.bindings.cs#L119

it's actually very rare to pass native arrays to extern declared methods

in their bindings

they usually package it all in a struct as with the async readback api

using IntPtr means it's unambiguous how C# should marshal i think

[lusty salmon]

Its not an editor script, it has object references in the scene

[abstract root]

then wrap all editor related code with #if UNITY_EDITOR and #endif

#if UNITY_EDITOR
using UnityEditor;
using UnityEditor.Build.Reporting;
#endif

...

#if UNITY_EDITOR
[Button]
    public static void TestBuild()
    {
        BuildPlayerOptions buildPlayerOptions = new BuildPlayerOptions();
        buildPlayerOptions.scenes = new[] {"Assets/Scenes/Dev.unity"};
        buildPlayerOptions.locationPathName = "Assets/Builds";
        buildPlayerOptions.target = BuildTarget.WebGL;
        buildPlayerOptions.options = BuildOptions.None;

        BuildReport report = BuildPipeline.BuildPlayer(buildPlayerOptions);
        BuildSummary summary = report.summary;

        if (summary.result == BuildResult.Succeeded)
        {
            Debug.Log("Build succeeded: " + summary.totalSize + " bytes");
        }

        if (summary.result == BuildResult.Failed)
        {
            Debug.Log("Build failed");
        }
    }
#endif

[lusty salmon]

Gotcha. Thank you I will try that now

[livid kraken]

Thanks doc if anything this will make the code cleaner

[undone coral]

what's the game?

are you trying to turn N assets inside a viewer into N webgl builds?

[balmy trellis]

hi all, does anyone have any pointers to making a good pre commit hook for c#? I'm trying to stop my devs from committing if there are any stylistic or typo errors in their code

[lusty salmon]

Creating mini scenes for a client. Using an algorithm to create all possible combinations and spit them out as webgl builds little dioramas

@abstract root Thank you but wrapping the editor bits did not work

[abstract root]

any specific errors?

[lusty salmon]

The same, claiming I am not using UnityEditor.Build, which I am

[abstract root]

is the script inside a separate assembly?

[flint sage]

You can't have any references to UnityEditor outside of Editor assemblies, the error says that you are and thus you need to exclude those from builds

[lusty salmon]

No, It is in the scripts folder

[lusty salmon]

Ok, I'm sorry a bit lost on how to resolve that

[flint sage]

Nixa gave you a good example on how to exclude things, you should do that for everything that uses UnityEditor

[lusty salmon]

I did that, no effect

Unless I missed something

[flint sage]

Or move your editor only code to an editor assembly, either a folder named Editor or if you're using AsmDef files then you need to create an asmdef and mark it as editor only

Yes you probably missed something

the error will point to where it found an issue

[lusty salmon]

This is where it says the issue is:

using UnityEditor.Build;
using UnityEditor.Build.Reporting;

Not sure how I could have possibly messed that up

[flint sage]

Looks like it's not wrapped in an #if

[lusty salmon]

The include lines need to be wrapped as well?

[flint sage]

Yes

[lusty salmon]

I wasnt aware of that

Thank you

[flint sage]

It's in Nixa's example 🤷‍♂️

[lusty salmon]

I see, I just missed that is all

[maiden turtle]

yeah, this is what I meant, that can never go wrong

[livid kraken]

yeah thanks for the help. I think I have it figured out now

now to continue restling with the unity api

btw what the doc suggested for IntPtr also works fine at least in the test, hope so with the actual codebase as well

[maiden turtle]

yeah since all pointers are the same size, it only matters from the type safety standpoint which type of pointer gets passed

and if you made adapters, you wouldn't care about that anyway

[livid kraken]

so what is "safer" ? IntPtr or explicitly using the same pointer type as the c++ code ?

[maiden turtle]

Neither one ensures you got the function signature right, I think.

using the same pointer type, well, reflects the type

so imo that's better

but make the wrapper functions anyway

[livid kraken]

in the context of what you are talking about is this ``` public static void NativeArrayTest(NativeArray<uint> data, uint count)
{

   IntPtr ptr = (IntPtr)data.GetUnsafePtr();
   myTest(ptr, count);
}``` the adapter ?

[maiden turtle]

yeah

[livid kraken]

cool we are on the same page then 🙂

that was how I'm making it anyways. I'm not calling the extern directly

[undone coral]

does anyone have any pointers to making a good pre commit hook for c#?
you probably want to use something that can compile into a single binary for all platforms, like golang or zig

then, author it as a plugin for pre-commit

however i will warn you... this is a colossal waste of time

[maiden turtle]

in my experience, it's best to have a custom cli tool for all CI stuff. you could try Nuke maybe, I didn't like it much but it does a lot of stuff for you.

[unborn vessel]

hello, i'm wondering does anybody know of a way i could acquire names of the files that a folder holds? for example if i go to a special directory i can create files File1.txt File2.ogg File3.exe and so the game acquires the names and creates strings so now i have 3 strings each one named that (order doesn't matter) and if there's only one file it only acquires a single string etc.

[maiden turtle]

Directory.EnumerateFiles

@unborn vessel

[unborn vessel]

thanks, but how does it work by the way?

[maiden turtle]

it's a function

[dapper cave]

I'd forgotten about this one. Cheers.

[somber swift]

That requires some shader knowledge but should be super fast

Also because you have to give the object positions etc. as buffers, its really easy and fast to modify them using compute shaders

[spice ocean]

Should I always declare my classes as sealed when posible?

[sly grove]

No

it's just noise

I'd say only bother with sealed if you're making a library for other developers to consume.

[spice ocean]

Well, bother with sealed is good practice and in some cases you can gain a little bit of performance

[torn basalt]

Not sure if I am asking in the wrong channel but I ran across something and can't figure out why. I am raycasting to an inverted cube and its not returning the correct normals. A basic cube returns the correct normals but not an inverted one. The player is inside of a box

[devout hare]

You have taken into account that inverted faces also have inverted normals?

[torn basalt]

I tried using the inverse but it still hasn't returned the correct normal

this is what I am using

        {
            Debug.Log(normal);

            switch (normal)
            {
                case Vector3 v when v.Equals(selectedObject.transform.forward):

                    _surfaceDirection = SurfaceDirection.Front;

                    break;
                case Vector3 v when v.Equals(-selectedObject.transform.forward):

                    _surfaceDirection = SurfaceDirection.Back;

                    break;
                case Vector3 v when v.Equals(-selectedObject.transform.right):

                    _surfaceDirection = SurfaceDirection.Left;

                    break;
                case Vector3 v when v.Equals(selectedObject.transform.right):

                    _surfaceDirection = SurfaceDirection.Right;

                    break;
                case Vector3 v when v.Equals(selectedObject.transform.up):

                    _surfaceDirection = SurfaceDirection.Top;

                    break;
                case Vector3 v when v.Equals(-selectedObject.transform.up):

                    _surfaceDirection = SurfaceDirection.Bottom;

                    break;
            }

            return _surfaceDirection;
        }```

[balmy trellis]

damn haha. Why would it be a waste of time?

[jolly token]

Hmm you could use dotnet format..

[balmy trellis]

Yeah, I'm thinking of enforcing dotnet format, but the goal of the pre commit hook is just to stop devs from committing if there's anything we don't allow style-wise

[warped mural]

I dont have enough information to go with "what I want". I don't know what are the proper options to form an opinion or decision.

[lament salmon]

Hey, should I use DrawMeshInstancedIndirect for a custom detail renderer? I'm using DrawMeshInstanced at the moment, but noticed that a lot of other solutions use the indirect method.

[sly grove]

indirect has the potential for better performance - but is also a bit more complex to set up

[undone coral]

what exactly are you testing? what is the game?

this one - https://store.steampowered.com/app/1198900/The_Underground_King/

?

[warped mural]

Yes

[undone coral]

okay

and what do you want to test

[warped mural]

I have some spreadsheets that dictate the proper loot and progression rate but my actual Unity-based generation is giving me buggy results

[undone coral]

like concretely

because i read your original message

[undone coral]

okay

that makes a lot of sense

[warped mural]

So my need is straightforward, I can either put Generate.cs and GenerateTest.cs in the same folder, or I can mirror my Scripts folder with a Tests folder

[undone coral]

yeah

[warped mural]

I would prefer pairing the classes together Generate.cs and GenerateTest.cs

otherwise I have to sync the Tests and Scripts all the time

[undone coral]

besides the aesthetics of it, why is it essential that the files be side by side

and why do you use the word sync? you could, or could not, write tests for any code, or no code

[warped mural]

The asmfdef is my blocker because I see everybody put all tests in a single folder

[undone coral]

what about it exactly is blocking you?

[warped mural]

If I put all the tests in individual folders sprinkled around, I have to make an asdmfdef for each, right?

[undone coral]

no

[warped mural]

An asmdef takes the folder it's in and the subfolders and turns it into a dll yes?

[undone coral]

yes

are you asking how to organize tests in unity?

[warped mural]

yes

[undone coral]

you're a very experienced developer so i hear you

there's a particular way you want to organize the files

which, it sounds like it's the golang pattern to me

i just have to understand if this is essential

like besides the aesthetics of it

why does a ClassNameTest have to exist for every ClassName

and why does it have to be side by side in the file system?

[warped mural]

I restructure my project shape frequently and recently I split the code in several submodules / git projects to reuse in other projects

Thus I wanted the class and the test class to stay together

[sly grove]

Unity testing definitely kinda forces you into having parallel folder structures, due to the way asmdef works

[undone coral]

okay but

[warped mural]

@sly grove I was afraid of this, but tutorials are kinda all identical and I hoped to find a solution

[undone coral]

what if you just had a file called GeneratorTests that tests all of your generation code?

and it's just 1 file for "all" your classes

[warped mural]

And have a 5000 line code omega class? sounds painful to update it

[undone coral]

you have a working game, usually for a less experienced programmer i would tell them that these aesthetic constraints are getting in the way of things

but if it's important to you to organize something in a very specific way

you can have the files side by side, and each their own named thing or whatever

it just depends do you really need Tests or tests?

are you going to run these automatically in a build server?

do you need them to correctly respect accessor visibility?

[warped mural]

I like single-responsibility classes and I structured my features in the form of:
Feature
-FeaturePrefab
-FeatureView
-FeatureViewTest (this is what I gravitate to add everywhere)

I will do them manually, or make a local script before doing git commit

[undone coral]

do you need a testing framework is really the question

do you care for any of the value that XUnit / NUnit (peer of unity's testing framework) gives you?

[warped mural]

Well, I want to do Asserts and I like the UI panel that shows me what's failed and what's succeeded

[undone coral]

okay

[warped mural]

I don't really care about the tool, just the results (painless as possible)

[undone coral]

does it matter to you if your build ships with the small testing framework dll?

[warped mural]

included inside the export?

[undone coral]

yes

[warped mural]

I guess it's fine

[undone coral]

okay

then you can just reference the testing framework package in your Scripts asmdef (or however you have it organized) and you're good to go

you can declare the tests

side by side

[warped mural]

how small is the testing dll

[undone coral]

well i'm not sure, but small

comparable to having an additional few textures

[warped mural]

Then problem solved! I get to have my cake and eat it too

[undone coral]

lol

[warped mural]

thank you @undone coral

[undone coral]

basically you're making your player assembly contain the tests

which are just methods with the annotation [Test] or whatever

in principle, it will all get stripped out in il2cpp anyway, since nothing references it

[warped mural]

I will put the contents of my Test.cs classes inside #if UNITY_EDITOR anway

[undone coral]

yeah i think it will all be fine

[undone coral]

for future reference a NativeArray has a bunch of fields, the first field happens to be a pointer to its malloc'd buffer. declaring NativeArray in the extern statement incorrectly sizes the function call, because NativeArray is a struct and C# passes structs by value (copies them)

an unsafe cast of nativearray to a void* happens to work because the first field is indeed a pointer to a buffer

so if you interpreted the nativearray as an IntPtr or void* unsafely, Things Would Work

that wouldn't be true if the declaration of NativeArray ever changed

if you did ref NativeArray, the method would receive a pointer to the struct, whose first field is happens to be a buffer point. so the receiving method would actually get a void**

no good.

[polar sedge]

https://cdn.discordapp.com/emojis/530707944195620864.gif?size=48&quality=lossless

[jolly token]

Can I make 'nested' class as MonoBehaviour/ScriptableObject?

How do I match the filename?

[warped mural]

1 mono per .cs

[upbeat path]

you mean derive a class which itself derives from monobehaviour ?

[jolly token]

public class OuterClass
{
    public class TempBehaviour : MonoBehaviour { }
}

Outer class is not MonoBehaviour

[sly grove]

you can but you won't be able to add those components in the editor or really see them properly in the inspector. You can only interact with it via AddComponent etc..

I don't recommend it

basically the editor doesn't know how to handle it if it's not 1:1

[upbeat path]

why would you want to use such a construction?

[jolly token]

It's related to convention of my package
I use generic outer class so it's gonna be extra challenging 😌

[jolly token]

Well, it's actually gonna be SO and user will not directly change content of it
So imma give it a try...

[upbeat path]

I cannot see that working

[sly grove]

you won't be able to create Assets of the SO, only instances in memory

which defeats the purpose, right?

the SO asset needs a file GUID to point at for the script. It won't have such a thing if the class is not 1:1 with a file unfortunately

[jolly token]

Hmmm yeah that will be shame
I can just go with interface 😌

Though It will be still 1:1 match

Outer class is not Mono or SO

[upbeat path]

That is the problem

[sly grove]

It won't be able to detect that sadly 😢

[jolly token]

public class OuterClass
{
    public interface ITempBehaviour { }
}

public class TempBehaviour : MonoBehaviour, OuterClass.ITempBehaviour { }

I might end up doing like this, it's also fine 😌

[upbeat path]

that will work

[jolly token]

But it's still gonna need generic 😛
I'll post here when I sort this out

[upbeat path]

generic on the interface or on outerclass?

[jolly token]

It's gonna be on outer class and SO

I saw this repo and pretty interesting https://github.com/SolidAlloy/GenericUnityObjects
So I probably can create asset with generic args same way

[upbeat path]

be careful with builds, a lot of generics will work in Editor but get removed from a build

[jolly token]

Yeah, as long as they are being used it should be fine, but I'll make sure test them

[upbeat path]

code stripping can be an absolute bitch

[jolly token]

This repo turned out it's codegen solution with Reflection.Emit
That automatically generate empty child class lol

Interesting but I'll have to find out a better way

[dapper cave]

DrawMeshInstancedIndirect could be even

[chilly nymph]

The unity API is mostly just a wrapper for those systems to interact with GO's yes, but there's also other things like sound, UI, inputs.

80% of the time you don't need a Mono class.

[undone coral]

is the goal to have a script that "holds" a data class or something

for the spreadsheet thing?

you can use the approach addressables does

[jolly token]

Yup it is
Do you mean AssetReference thing?

[undone coral]

yes one perspective is that the end user knows exactly what type they will need

they are never going to make the error, oh no i forgot what type this is

they can handle CatheiReference ref; ref.AsRow<RowType>()

otherwise code generation is unavoidable. you would probably want to create an IDL, and then generate the types from that using someone else's tool.

[jolly token]

Yeah actually I pretty much settled with same conclusion
So my SO will only hold JSON serialized string that can be used for any row

With a type metadata for editor script to filter them

[undone coral]

i also do not find it that burdensome, in general, to declare the class MyClass : BaseClass<RowType> {} myself

if i am using your sophisticated thing, i can handle creating that file

i wouldn't be bummed you didn't make that for me

however... how are you generating RowType?

that's still codegen

tricky codegen at that

i've seen JOOQ's code it's nuts

[jolly token]

I don't do codegen 😄

User create C# class themselves

It's like DB ORM

[undone coral]

gotchyu fam

[jolly token]

Then I parse spreadsheet by that type user defined

[undone coral]

gotchyu famalam

i would suggest doing the hdrp config package approach

where you get a copy of a bunch of things you can customize

then you can give the end user the file

[Serializable]
public partial class DefaultRowType : IRowType {

}

that they can edit... they're going ot have 1 row type most of the time

then include a file ```cs
public partial class DefaultDataHolder : DataHolder<DefaultRowType> {}

in your package

[jolly token]

Hmmm how will it help :0

[undone coral]

DataHolder would be a monobehaviour*

public abstract class DataHolder<TRow> : MonoBehaviour where TRow : IRowType { ... }

you can actually just put both partials in your package

and then build your docs around filling out the partial

on their side, in their own space

since most people will have 1 row type they'll at least be able to get started easily

you just can't easily rename / upgrade this stuff, but that's a problem with all serialization libraries

[jolly token]

Ah that kinda thing 🤔 Isn't it basically codegen without 'gen' but manual copy

[undone coral]

or no copy

it's a low ask

protobufs ships all its classes as partial for C#

in case you wanna add stuff

[jolly token]

public class ItemSheet : Sheet<ItemSheet.Row>
{
    public class Row : SheetRow
    {
        // use name of matching column
        public string Name { get; private set; }
        public int Price { get; private set; }
    }
}

This is how it looks like rn

The thing is I'm supporting multiple importer/exporter with same object

public abstract class SheetScriptableObject : ScriptableObject, ISerializationCallbackReceiver
{
    [SerializeField]
    private string _serializedRow;

    [SerializeField]
    private List<UnityEngine.Object> _references = new List<UnityEngine.Object>();

    private ISheetRow _row;
}

SO's gonna look like this, I'm happy with the format itself

[undone coral]

okay so this is pretty similar to addressables

which is also good

[jolly token]

User can use ItemSheet.ScriptableObject type to filtered reference of SheetScriptableObject and get ItemSheet.Row through it

Which I'd provide with PropertyDrawer

So that one is pretty much settled, I have another problem, though

About how to support column referencing specific asset, while supporting server side assembly

Something like AssetPath<GameObject> is not going to work in non-Unity env 🤔

[undone coral]

you can use unity stubs

they exist

they're alright

[jolly token]

What is that?

[undone coral]

hmm

there used to be a github repo

with unityengine.dll and unityeditor.dll

stripped of externs

so it's just the types

[jolly token]

Sounds useful for this purpose

[undone coral]

maybe with the C# source repo it Just Works

okay here we go

https://github.com/Rabadash8820/UnityAssemblies

[jolly token]

Setting looks like requiring some complex work but very interesting, thank you

I'll either go with using mock classes for Unity, or just use generic AssetPath

[undone coral]

it is pretty painful to install the unity editor into a container btw

could be a huge distraction for you

[jolly token]

Exactly. 😢

[undone coral]

although i think most innovation lies there

it's very tough

i pitched trilib guy on it too

[jolly token]

Only if there was [Conditional] for generic parameter lol

[chilly nymph]

Is there a reason?

[jolly token]

I do use codegen in my own projects, roslyn generator is cool, just I don't need codegen for this library

[chilly nymph]

Ah was wondering if there was hate against it.

[jolly token]

Plus making codegen configuration generic is messy XD

I want my libraries Vanilla as possible so it can be easily maintained

[shy spruce]

should the UserSettings directory be gitignored?

[chilly nymph]

When you make your repo on github, add the default ignore file.

[shy spruce]

I don't use GH unfortunately

and the project is already made

If you have your .gitignore handy, could you please tell me if UserSettings is included in it?

no worries if not

[fresh vortex]

Just Google out unity git ignore default template

And see if it's in it

[shy spruce]

ah yeah that's a good point -- id already done that, and am using a template, but i guess the update to 2021 added this directory. mine didnt have that. thanks

looks like UserSettings is indeed included in 2021

[jolly token]

They should be ignored

It's personal settings per user

[shy spruce]

Thanks!

[dapper cave]

ComputeBuffer.GetData, is there something faster and non blocking?

[chilly nymph]

When you initialize an object you have control over what it can reference as well, as oppose to having to resort to using the api to pull a reference.

      SomeRegisterService(this);```

[sage radish]

ComputeBuffer.GetData is the fastest way. AsyncGPUReadback gives you non-blocking, but is a bit slower.

[jolly token]

In theory, you should not create GameObject if it doesn't have "view"
Which would be physics collider or rendering component, they would only interact with physics

But if you count logical case like "Player HP went down so HP bar should decrease" as interaction then there is much more way to interact

[dapper cave]

Do you know how much slower Async is?

[sage radish]

Documentation says:

adds a few frames of latency

[chilly nymph]

Sure, if your system doesn't need to do that. But relying on (for an example) like a health system mentioned above you shouldn't have a GO health (just a health bar) you shouldn't be using any API calls to get a reference to anything.

HealthBarUI.Set(Player.Health) There's no API calls to unity, you have all direct references so I don't know why you don't think it's a good idea to "couple" one Mono to another when you can have managers / controllers.

[dapper cave]

I'll take that, thx

[chilly nymph]

"That game object would have something that represents the health system" You're saying you would put the player health on a healthbar UI?

Why would you have health data on a GO?

       public uint health;```

You would pass your view a value and it would render it, right?

An observer doesn't need to be a GO

[chilly nymph]

"That game object would have something that represents the health system" You are saying that it represents a health system, not the health value.

[undone coral]

how are you using it? usually for low latency stuff you want to keep it on the GPU

[chilly nymph]

I'm trying to explain that 80% of the time you don't need to interact through game objects (and you shouldn't). For the above example your game objects should only receive and render the value of health, there shouldn't be health anywhere in the scene as you don't need it.

For things like a player attacking enemies, yes that's a system that needs colliders, and you would have to use the API.

For a system that has only ONE enemy you wouldn't use colliders unless it was physics based and you would directly call Enemy.Hit(damage) from your Player.Attack(), or even a observer.

You wouldn't want to rely on another layer of interface if you can just do it by directly referencing.

I'm adding to your central point by saying most of the time a system doesn't need to rely on the API to get references, you can arcetect it around direct references when you can.

[dapper cave]

it needs to come back to the cpu to get mesh colliders

[chilly nymph]

what exactly is the cause of your in game interactions that isn't a raycast I literally just showed you Enemy.Hit(damage) from your Player.Attack(),

[undone coral]

that's tough

imo you should do a rough simulation in CPU first

[chilly nymph]

void IsAttackButtonPressed(){ Player.Attack(Enemy, damage)

[undone coral]

or iteratively do the gpu work

what is the platform?

[chilly nymph]

Mate you have inputs.

[jolly token]

Instantiation is also interaction 😄

[undone coral]

on modern AMD and NVIDIA gpus, compute can run parallel to graphics, and in principle you can get your data same frame

[chilly nymph]

      Player.CurrentEnemy(this)```

[jolly token]

The spawner and instance

[dapper cave]

it's not sim, it's marchingcube to mesh a voxel volume

[chilly nymph]

I think we're trying to explain that you don't need to use the API in most cases.

From a manager.

Static.

Singlton.

Where would you like to get it?

[undone coral]

have you looked how mudbun does it

[chilly nymph]

And relying on a raycast is better?

You're free to do what you want. I'm trying to argue that having a direct reference as opposed to relying on a API interface is better.

I disagree 100%.

But you're free to do it your way.

You know that relying on the API is "Middle management" too right?

[dapper cave]

does what?

[undone coral]

marching cube to mesh a voxel volume

[dapper cave]

they keep it on the gpu

[chilly nymph]

I understand it, but your only argument as to why you prefer using the API is that "managers suck" When you're using managers.

[undone coral]

it has colliders too

i don't know if it's realtime colliders

[dapper cave]

ok, lemme look

[undone coral]

or like if you have to call a method to get the mesh

[chilly nymph]

.Hit(), .Collide(), .Find() pick what you amt.

[undone coral]

it has SDF raycasting

but it soundsl ike you want physics objects right?

[dapper cave]

i don t know if it s realtime colliders

[chilly nymph]

But a raycast (or any other collision) is a lose manager.

You're literally using a manager when you use those.

What do you think .Hit() is?

Then use a Singleton.

Alright mate. You keep using the API.

What do you think Physics. is?

You think it's some magic code? 😄

I don't use managers uses manager.

What is the Physics. lib to you?

How does it exist?

I'm listening, you're "Barfing" at managing references but then don't acknowledge it's a standard across Unity.

Alright, anyways you're free to do what you want.

But you refuse to write one because "Barf"

IMO you're just relying on weaker code and adding smell when you rely on physics when you can easily contain a direct reference where you need it, but you can do what you want.

We don't have to agree, and that's fine.

To be fair your only opinion was "Barf" 😄

[pliant crest]

why would you have both Instance and Singleton?

[chilly nymph]

I think you're lost mate...

[zenith ginkgo]

Who... Actually cares?

this whole thread is very childish.
theres no 1 way to do something

Move on

[chilly nymph]

Well you should get it right, SomeManagerManager.Instance().Enemy()

[jolly token]

Manager doesn't have to be singleton 🤔

Dependency Injection 🌈

[pliant crest]

whether you DI or singleton i think his problem is more having to write a manager

sad thing is

this means ECS is out the window

cause S is basically a manager

would be nice if they supported DI out of the box

thats not what DI means

[jolly token]

As many as your contexts 🤔

[zenith ginkgo]

Call me a sinner

[pliant crest]

i think most people have something similar

[jolly token]

For example each prefab can be each small context if they need dependencies
A combat can be context, a scene can be context

[pliant crest]

that doesn't really mean anything thou

[zenith ginkgo]

Singletons are good, if used in moderation Imo.
GameManager is the only singleton here, and it has a reference to the other managers

    public AudioManager GetAudioManager() => audioManager;
    public SaveManager GetSaveManager() => saveManager;``` and yes, i prefer this, instead of a property

[thin mesa]

but that's effectively the same thing as a property

[zenith ginkgo]

we all have our kinks boxfiend

[pliant crest]

why not just make it a field at that point?

seems like pointless ovehead

[jolly token]

😌 It is not single point of entry if it's getting referenced from all over your code

[pliant crest]

if its many:1 its single point of entry no?

how else do you define single point of entry otherwise

[jolly token]

Single point of entry should not be referenced by other code in that context..

[pliant crest]

???

[jolly token]

It should reference dependencies, not getting referenced

[pliant crest]

dependencies of what?

this is a wrapper around the resource

to give a single point of entry

what dependency are we talking about here

[jolly token]

Composition root is where the dependencies getting injected..

Not the other way around like every other script referencing Composition root

[pliant crest]

oh if you are referring to push vs pull sure

[jolly token]

I'm talking about Dependency Injection

[pliant crest]

yeah i know

[brisk pasture]

yeah its just DI without a framework. actully prefer it this way

[pliant crest]

but whos gonna build DI

i was more referring to this is single point of entry

i didn't really care about the CR comment

but i see what you are getting at

[jolly token]

Single point of entry is where program or context starts, composition root can be initialized

[brisk pasture]

it just makes it easy to manage dependencies, they can be constructed at different points with the top being main, and passed down to what depends on them

[pliant crest]

only if we're talking in context of DI sure

[jolly token]

Referencing single point of entry is like calling Main from your script, which is nonsense in most of scenario

[chilly nymph]

Y'all want to fight about Scriptable Objects now?

[pliant crest]

i've still never used an SO before

[zenith ginkgo]

(╯°□°）╯︵ ┻━┻

[jolly token]

Keep it that way, SO sucks

[chilly nymph]

It's actually funny, on the other discord (GDL) they advocate for SO's.

[austere jewel]

They're an awesome part of the toolkit that people should consider when implementing anything. Just like almost everything else you can use them inappropriately or wastefully

[pliant crest]

huh its probably the type of concepts i code, most of it doesn't have (large amounts of) hard coded data

since its mainly roguelike/lite stuff so haven't had a need for a data class

like the test tower defence i'm trying out right now you build your own tower

so there is no definition

[jolly token]

Using SO something other than immutable data is basically visual scripting 😌

[chilly nymph]

Don't you build a tower in every tower defense? It's in the name

[pliant crest]

shrugs if you can't infer what i mean by that

then doesn't really matter

[chilly nymph]

ᴵ'ᵐ ᶠᵘᶜᵏᶦⁿᵍ ʷᶦᵗʰ ʸᵒᵘ

[jolly token]

Where do you store data like waves, items, initial health, etc?

[pliant crest]

its all math formulas

there is however a baseline

for the (stats)

and there is no predefined wave

[jolly token]

So baseline values are just set in inspector I guess

[pliant crest]

its on a manager

but yes set in inspector

(technically its on a System but w/e)

[jolly token]

I see, nothing wrong with that if you don't have to edit or refactor values a lot

[undone coral]

oo, a scriptable object gold star

you don't have to start

don't even think about scriptable objects

[undone coral]

this is very pleasing and reassuring to me

the comforting, snug feels of the scripts that you finish and don't revisit, and then when you need to write new stuff, it gets added on

the warm fuzzy of a game getting made

[undone coral]

this is psychotic and puts me into an emotional tailspin

okay i can think of worse things:

public class managers : ScriptableObjectSingleton<managers> {
 public Func<ScriptableObjectSingleton<audioManager>> getAudioManager;
 void Start() {
  getAudioManager = FindObjectsByTag("audio manager")[0].GetComponent<AudioManagerFinder().audioManager_FindAudioManager_SO().Deserialize();
 }
}

[jolly token]

Too artificial

Unless you make YT video with lot of tutorial follower

[undone coral]

lol

[undone coral]

okay i'll work on this bit more

_SO is pretty authentic though

[austere jewel]

doing it on start is somewhat sane, you've really got to make some basic logic mistake in a property that means it's retrieved via Find every time it's called

[rotund sorrel]

Hello, I have a problem with collider and rigidbody. I force my ball with 2000f and it bounce off even dont hit the bound (pic1) - exactly it cant because the ball smaller than empty between two bounds.
When I force my ball with 1000f , it go through normaly.
I had test my Ball with Continous Speculative and Continoues Dynamic but nothing change

[shy spruce]

Does OnMouseEnter get called for any collider in 2D, or just the first one the raycast hits?

[scarlet copper]

Guys I have weird problem. I have a player object which is always in DontDestoryOnLoad Scene. It is a VR game. It has in Start() SceneManager.sceneLoaded event. But if a new scene is loaded all references in this script gets null. But only in the sceneLoaded assigned event method. I am really confused what is happening. I already checked with Awake and Update but there references are NOT null but in the event method they are null. And the funny part is this happens not always. Even if (gameObject) causes error

[long ivy]

looks like PlayerManager isn't DDOL after all

and any other GameObjects it references which aren't also DDOL will be lost when their scene is unloaded

[scarlet copper]

The references are all childs of the main gameobject. Like I wrote I used Update() to print the references and they are never null there and it happens randomly. I also checked OnDestroy.. also not invoked. In the editor I can see the gameobjects all exists and even in the inspector I see them they are assigned.

Ok I tried something else. It has definitely something to do with SceneManager.sceneLoaded. I changed to LoadSceneAsync and used complete event and there gameObject exists but in the event of sceneLoaded it shows still error.. Weird.

[chilly nymph]

public class DDOL : MonoBehaviour
{
    void Awake() => DontDestroyOnLoad(this);
}

Also if you can, I would suggest removing your dependency on Unity child / parent. And hold references directly to your instances.

A parent spwans a child, so you have your reference right there.

May fix some of your issues.

void SpawnChild() => new Child(this);

[long ivy]

maybe time to show some code. The fact that your OnSceneLoaded takes an extra argument tells me maybe you have a closure over something and are not looking at what you think you're looking at

how are you setting this event up?

[scarlet copper]

This is how it looks like (simplified)
Player (PlayerManager.cs)
-->Camera (CameraFade.cs)

Player has the PlayerManager script.
Camera has the CameraFade script. PlayerManager has a field for CameraFade and I just dragged and dropped dropped it like usually. Its also a prefab. So it cannot lose actually.
But still does not makes sense, why this.gameObject should be null?

But like above I said with AsyncOperation it never shows any errors. Just check the image. Async also uses gameObject and no error while in the SceneLoaded it shows error.

[scarlet copper]

Ok in a sec

@long ivy

void Start()
{
    SceneManager.sceneLoaded += OnSceneLoaded;
}

private void OnSceneLoaded(Scene scene, LoadSceneMode mode) {
    previousScene = currentScene;
    currentScene = scene.name;

    Debug.Log("OnSceneLoaded invoked");
    Debug.Log(gameObject); // this shows error sometimes

    //if (gameObject) // this even cannot work, it shows error
    //    StartCoroutine(DelayedInvoke());
    //else
    //    Debug.LogError("This gameObject does not exists...");

    //IEnumerator DelayedInvoke()
    //{
    //    yield return new WaitForSeconds(1f);
    //    camFade.SlowFadeIn();
    //}
}

// This gets called if I want to change the scene
public void TransitionToScene(string scene) {
    sceneName = scene;
    camFade.SlowFadeOut();
    Invoke("LoadScene", 1.5f);
}

private void LoadScene() {
    previousScene = currentScene;

    //SceneManager.LoadScene(sceneName);
    asyncLoadScene = SceneManager.LoadSceneAsync(sceneName);

    asyncLoadScene.completed += (ao) =>
    {

        Debug.Log("asyncLoadScene.completed invoked");
        Debug.Log(gameObject); // NEVER shows error (10x tested).
        if (gameObject)
            StartCoroutine(DelayedInvoke());
        else
            Debug.LogError("This gameObject does not exists...");

        IEnumerator DelayedInvoke()
        {
            yield return new WaitForSeconds(1f);
            camFade.SlowFadeIn();
        }
    };
}

And of course this one

void Awake() {
    DontDestroyOnLoad(gameObject);
}

The version I am using is 2021.3.4f1

[long ivy]

this doesn't match your previous code. Does the PlayerManager GO have a parent? Does PlayerManager exist in the scene as a prefab instance?

[scarlet copper]

I used in OnSceneLoaded PlayerManager argument just to compare the instance. I thought maybe there are two of them and weird things happens. I just changed back to original code.
No, PlayerManager has no parent. PlayerManager is a prefab.

[long ivy]

does PlayerManager do anything else? Is it a singleton? Is the AwakeScene ever loaded again?

[scarlet copper]

This is not even my project and not my code. I just got hired to complete some milestones but I couldn't even start because of this error 😅.
However, as you can see it works with asyncLoadScene.completed event.

But I also noticed if I go into playmode if I don't get error.. so the error never comes. But if I restart and it shows error.. so it happens always. But still does not make sense for me.

[scarlet copper]

Yes but in that time I do nothing else. In AwakeScene is just idled until I call LoadScene (TransitionToScene). AwakeScene is only loaded on gamestart and from there I can switch to another scene which I can choose via the UI or via ContextMenu in the editor (for faster debugging).

The singleton is a good hint. Let me check that

Omg you are right. It is a singleton and they activated the Enter Play Mode without Domain Reload. Now it makes sense.. Holy...

Ok, I finally figured out what's actually happening. The singleton is NOT the problem. Oh dear.. what a headache.
Since this SceneManager.sceneLoaded is a static, it does not gecleared if I replay because Enter Play Mode was activated in the project settings. So it just contains the handler and reference to the gameObject which of course is null when I turn back to editmode.

Now it also makes sense why it happens sometimes. If I change code and compile the domain gets reloaded so the sceneLoaded handler list is cleared. That's why it worked also all the time until I restart it without any code changes because sceneLoaded had one handler in the list not two with null. What a headache. I hope I was able to explain it clearly.

[livid kraken]

At this point I'm at a total loss. https://pastebin.com/1DyXju4F I know that the data in newIndicies is correct. The triangles get set, the submesh has the correct number of indicies, the bounds are correct etc etc. Yet no mesh is rendered at all. I also know that the vertex data is there and correct since I dont modify the verts at all. Wtf am I doing wrong.

[torn basalt]

wrong group sorry

[fresh salmon]

If you're going to post elsewhere please at least make the effort of deleting your question here!

[torn basalt]

effort made

[undone coral]

where do you interact with the submesh here?

anyway i think it's just something here that has to actually be called

i am not sure what it is

[livid kraken]

Lines 40 to 42

I set the new index count and apply the submesh to the new data

[livid kraken]

If something else has to be called Im not sure what it is too since its nowhere in the docs

[ionic quartz]

Hey guys, I'm wondering if anyone has some insight in an algorithm that would create a pattern similar to this? I am planning to use this for a continuous 2d cave system

[undone coral]

where did this picture come from?

did you draw it?

[ionic quartz]

Ye I drew it

[undone coral]

okay

you can try drawing it again, and write down the decisions you made while you were drawing it. that is the start of your algorithm

there are many approaches to procedural graphics generation, but the real problem is that they are incomprehensible to people who have never done it before

and it wont' be useful to use a preexisting one for you

[ionic quartz]

Ah that makes sense thank you

[undone coral]

i would probably use an L-system to generate lines. self closing loops are challenging to express for uh, laypeople

[gentle topaz]

honestly just looks like some modified version of perlin worms

one pass where it mostly moves right, then another pass where it adds offshoots

[ionic quartz]

Can I guarantee that a perlin worm will overall move right?

[undone coral]

the biggest part of doing this is the problem solving

if you like math try to express it mathematically

if you like updateloopese

you know, make a character that walks in that path

[mortal orchid]

I am using this method to change the direction of the enemy towards the player in a 2D top-down game. But the animation is flickering do you see any problem in this way of implementation?

[mortal orchid]

Nope not showing the progress. I understand that the channel is for advanced code. forget to write the description. I have edited the post

[tender light]

This is at best #archived-code-general kind of question. @mortal orchid

[fresh salmon]

Don't cross-post.
You'll need to provide more info, like whether you're getting the logs or not. Make a development build if needed. Check the Player logs if needed.
Ask in #archived-networking if the error comes from any networking related method.

[foggy bramble]

sorry, It is a development build which is how I know that the error is nullreference object

[fresh salmon]

You'll need to post the full error message, which contains the stack trace

[foggy bramble]

OK

[foggy bramble]

https://pastebin.com/51BmLJJg

[fresh salmon]

NullReferenceException: Object reference not set to an instance of an object
at Launcher.OnJoinedRoom () [0x0000c] in C:\Users\fdmcp\Valorant 2d\Assets\Launcher.cs:22
Error on Launcher.cs, line 22.

[fresh salmon]

playerPrefab was null

[foggy bramble]

but how

[fresh salmon]

That's what you're trying to solve

Maybe the PlayerContainer object isn't there at the time the method runs

Try referencing the script in any other ways than drag-drop.

[foggy bramble]

how would I do that if its a prefab

[fresh salmon]

Like any other object. If you're able to drag-drop, you can do all the other ways. This includes using GetComponent<T>, Find, FindObjectOfType<T>

[foggy bramble]

I'm just so confused why it works in the editor and not in the build

[fresh salmon]

That's why I'm telling you to try other ways of referencing that component

It's a networking issue, and I'm not experienced in networking, so we do the regular C# and Unity ways first

If that leads to nothing, ask in #archived-networking

[foggy bramble]

Ok

[wintry wind]

Hi I have a script that generates a mesh and assigns that mesh to a property. When I try to Apply the overide mesh to the prefab (I want to save it to the prefab) it says that I can't because the mesh is a scene reference. But the mesh is generated by code and assigned as a property to the object. How can I do this? Do I NEED to do an editor script?

[foggy bramble]

I just launched the build agian and it worked idek what was wrong

[dusky tiger]

Hello my Trigger is not working ```csharp

private void OnTriggerEnter2D(Collider2D collision)
{
trsf.position = new Vector3(5, 5, -1);
if (collision.gameObject.tag == "Player")
{
Audio.Instance.PlaySFX(win);
SceneManager.LoadScene("Winning Screen");
}
}```
If I go to that trigger with my Player but it does nothing

[jolly token]

You didn't check isTrigger flag. #💻┃code-beginner

[dusky tiger]

I tried that but it just dissapeared

[grave vortex]

how would I change the skybox material in a custom render feature? I tried RenderSettings.skybox = newMaterial, but that set the skybox for the subsequent pass as well, would just like it in my scriptable render feature.

[gray pulsar]

maybe cache the original skybox material and set it back at the end of your feature's pass?

[grave vortex]

At what point would I reset it?

[gray pulsar]

I haven't done much with custom render features, so I'm just guessing here. But looking at the docs, it seems like doing it at the end of your Execute method would do it. I'm guessing any time after you execute the command buffer

[grave vortex]

Well, I'll just keep hacking away at it. To be clear I am trying to render the skybox with a unique material only in my render feature.

[quartz stratus]

CRTs are a bit of a nightmare tbh, double buffered CustomRenderTextures only got WebGL support like 6 months ago. Best of luck 💪

[grave vortex]

how do I define a custom LightMode tag in shadergraph?

[sly grove]

What disappeared

This is a #💻┃code-beginner question BTW

[undone coral]

this might be helpful

https://github.com/Unity-Technologies/UniversalRenderingExamples/tree/cc4f75e932d9d04bf658886c35e5a93d0a5e0f06/Assets/_CompletedDemos/3DSkybox

[grave vortex]

I have that one - and yes, it is quite helpful.

[bronze geyser]

What I wish to achieve is a "movement" distance outline similar to that of XCOM. Currently I have a bunch of individual planes to represent each "tile". I imagine the way this would be most easily done:

1. Grouping all the tiles within range.
2. Filtering out all non-edge tiles.
3. Then using the edges of the edge tile's meshes to get a series of points with which to procedurally generate a plane that would represent.
4. Applying a transparent material with a border shader.

I'm primarily concerned about the opaqueness of step 3 since I don't have a lot of 3D visual programming skills. I was wondering if anyone was aware of any resources that cover similar procedural mesh generation techniques?

[bronze geyser]

I guess related to this, when constructing a mesh. Aside from minor performance hits, can I just ignore some vertices? One approach that comes to mind is to simply create a bunch of vertices to represent a rectangle that would be the max height/length of the movement area, and then simply render each moveable square with 2 triangles and just ignoring any vertices that are not on a corner of a valid tile.

[maiden turtle]

It has to be stored in an asset in order to be serializable. Prefabs cannot hold mesh assets.

you can create a new asset and add the mesh to it

[plucky laurel]

technically they can

you just need to serialize it with callbacks

[plucky laurel]

i think this can be solved with just a shader

command buffer at most

maybe, im not sure, some of those corners on the screenshot look sophisticated

[bronze geyser]

I'd probably be even further behidn with that approach. At least when it comes to proc genning a mesh, I understand the theory behind it, but maybe not the best practices.

[novel plinth]

This can be done easily with custom renderer feature, https://github.com/Unity-Technologies/UniversalRenderingExamples

see the one with hull method....

oh wait, misread, that's not related to yours I think.. scratch that!

[spring gorge]

there is anyway to setting this by code?

[mortal orchid]

Not able to get any help on that so thought maybe it is an advanced problem. 😅

[unborn lava]

When changing the gameobject name via code "go.name = 'something'" I don't see it changing in the Hierarchy when the game run

[analog spruce]

hey im making a sort of prop hunt game and i was wondering how i could make a script that automatically goes on top of the floor instead of floating in air or in ground? sorry if im unclear

[sly grove]

It works. You're probably just looking at the wrong object or something

Or your code isn't running

[brave viper]

Not at all, they are used a lot

[flat marsh]

not at all, if anything using MB for literally everything is a code smell

[tender light]

Agreed.
Interfaces have issues in Unity and Abstract classes are preferred where possible for MB.
But when writing any kind of code you should stop and think if it needs to be a MB at all.

You can do a LOT with [SerializeReference]

[jolly token]

If you don't try to mix them and use MB only for views (and don't use SO) then you won't have problem

I work on gameplay on days and work on library at night

[tender light]

Well if you want them to be a SO then use abstract classes with <generics> you can accomplish a lot. 🤔 @regal olive Would that not be good enough for your use case?

I think I don't quite get the issue. I am sorry.
It this just not liking that the inspector does not allow you to reference a SO by an interface? @regal olive

You can write your own editor script. Maybe use Odin for that (I think the functionality is there but not 100% sure)
You can also perform checks for interfaces in "OnValidate". I think you can even create your own generic class and your own mono behavior base that checks all of its fields via reflection and runs their "OnValidate" method maybe. But these are workaourds upon workarounds 🤔 I am just brainstorming atm...

Sure 👍 tho its not [ReferenceField] but [SerializeField] 😛

Yeah. As I said. Unity pushes you towards abstract classes. They work much better in the inspector in general

Sure it does. Again. Push your code into an abstract base class and it will work in the inspector.

Unity prefers very flat class structure for most things. Each component should do 1 thing and 1 thing only. Therefore there should be no need to have interfaces on components.
That does not mean you don't use interfaces in Unity. You do. But for non-MB non-SO scripts. You can use interfaces for [SerializeReference] Lists for example (I love this use case. You can easily create a list of conditions and checks that build up into something like a quest/achievement system)

[brisk pasture]

i would actually argue interfaces go very well with components and doing only 1 thing, since it makes swapping of implementations of that 1 thing very easy

though you can still GetComponent by interface type, which is rather helpful for things like raycasts, and taking a action on the hit object

[undone coral]

it's in PlayerSettings

[brisk pasture]

think its because i also do server side stuff and work with Go, but composition and interfaces are my default way to handling most things over more OOP style stuff

really the most annoying part, is not being able to drag in UnityEngine.Object based stuff, on a SerilizedField

otherwise it works decent

though the gopher in wishes i had implicit interface implementation

for non UnityEngine.Object based stuff it works great, or if you call GetComponent yourself

like most of my state machines, are regular C# objects, saved with SerilizeReference

and all by interface type

[undone coral]

Do I NEED to do an editor script?
the simplest thing to do is to declare a context menu function using [ContextMenu] on your monobehaviour, then click the triple dots icon in the upper right corner of the component's title in the inspector to run the function you annotated with it

#if UNITY_EDITOR
using UnityEditor;
#endif
class MyScript : MonoBehaviour {
 public Mesh myMesh;
 [ContextMenu("Generate and Save Mesh")]
 public void GenerateAndSaveMesh() {
#if UNITY_EDITOR
  var mesh = GenerateMesh();
  var asset = AssetDatabase...
  myMesh = asset;
  ...
  MarkDirty()...
#endif
 }
}

[brisk pasture]

yeah can be worked around, issue is for example the faking it with a abstract class, is you can only extend from 1

so kinda interface like, but it pushes you into larger interfaces instead of lots of small tiny 1 or 2 function ones

[tender light]

That is a very nice functionality I agree. BUT
What allright was writing about were local (I assume) children object references. Which... you can totally go without interfaces for that.

[tender light]

No. Why would it. You instead have more components. You can still have tiny classes with a singular purporse.

[wintry wind]

I ended up finding an answer online. It seems the issue is that meshes are not serialized as they are assets. I ended up serializing the vertices and triangles and it worked like a charm

[brisk pasture]

interfaces for this are good for cases of the interface defines what something can do, but multiple components implement it so the how can be different and the caller does not have to care exactly which component handled it

[undone coral]

when you say serializing the vertices and triangles...

you can definitely save a unity mesh as a .asset asset

that is the editor-native mesh representation

[tender light]

To clarify. We are talking about MBs so more than one base class is present for sure.
I am NOT talking about aggregating multiple interfaces into a single base class. And then just saying "f it" and not implementing some of the interfaces methods in the subclasses. (Who would even do that. Creating 0 side effects methods for everything? or throwing exceptions. Thats terrible architecture)

[undone coral]

i think what you found online is wrong

[tender light]

Different class for AWeaponEffect ADebuff AUpgrade etc...

[undone coral]

here's the head of one such file

%YAML 1.1
%TAG !u! tag:unity3d.com,2011:
--- !u!43 &4300000
Mesh:
  m_ObjectHideFlags: 0
  m_CorrespondingSourceObject: {fileID: 0}
  ...

it does indeed have vertex and index buffers in it

when i see interfaces in a unity game i see "junior programmer"

you control 100% of the source code in a typical unity game. if you need functionality X in place Y, you can declare it there.

when you author libraries that will be open source, it also makes less sense to create apis that accept interface arguments

exhibit A: cinemachine

the IVirtualCamera interface is implemented by only two classes, and its purpose is to work around the fact that you cannot new an MonoBehaviour but would like to mock the behavior of one, from the point of view of how it implements blending, in a very specific corner case in their implementation

interfaces make sense when you cannot possibly know all the types ahead of time

in unity, for the most part, you know the types ahead of time

UGUI / EventSystem's use of interfaces is pretty flawed

you look at the Unity API, only EventSystem does interfaces

there's IJob

[dapper cave]

is there an attribute to prevent auto API update?

[undone coral]

which is really "a function signature"

[undone coral]

there is but i don't remember what it is 😦

for example EventSystem makes you do something like "you must implement IBeginDragHandler in order to use IDragHandler"

UIElements, as far as i can tell, is just "the DOM"

which of course makes more sense

what should Unity have done instead? i suppose author a single UIBehaviour class (which exists by the way) with the one property public EventContext eventContext

eventContext.onPointerClicked += ...

eventContext.Cancel()

etc. etc.

EventContext.From(this).onPointerClicked += ...

no interfaces.

i can see how if you're a junior programmer and you like to plan out your thing using interfaces - i don't know why you wouldn't just make an empty class instead, but there is something in the education of this stuff that doesn't illuminate that strategy

i've seen it with my undergraduates

and it's fine. i tell them in their editor to just click "inline interface" and it's all better

[gaunt spoke]

I didn't read everything but are you saying we should never use interfaces in Unity ?

[undone coral]

i think almost all of them can be eliminated, provided you aren't using closed source external libraries whose APIs require them

i am not opposed to "thinking by using interfaces"

which is why people actually use them

but you should try to remove them once you are "done thinking"

if that makes sense

you can also think using "a class"

you don't have to use an interface at all.

there's just something in the education that sticks to intefaces

if you can't possibly anticipate a type ahead of time

then you have to use an interface

that is what they are actually for

for making libraries that have to interact with stuff whose type you cannot possibly know ahead of time

for many APIs, generics obsoleted interfaces

[gaunt spoke]

If I have a player that can take damage and be attacked, I find convenient to create a IDamageable and a IAttackable interface, instead of deriveing from a DamageableAndAttackable base class

[jolly token]

Though I agree as “You don’t need more abstraction than you need”
Interface is a protocol, you won’t have same modularity and abstraction if you use class instead.

[undone coral]

yes, but you only have 1 player

[sly grove]

I think it's even better to just make Damageable a separate component and forget about the interface. Then communicate with other scripts via events or direct references.

[undone coral]

pretty much everything in your game will be damageable and attackable.

or it won't

or, consider that such games, where you have these more complex rules, you have to do something that a junior developer is going to flop around and not do really well anyway

i don't think Jai has interfaces. if you really want to explore more about, what makes sense when programming games, i think jonathan blow has spent more time thinking about this than maybe anyone

[ionic quartz]

Does anyone know if there's a first derivative of Mathf.PerlinNoise()?

[undone coral]

what is your objective?

are you trying to ask for "a texture that shows the magnitude of the slope of the point at x,y of the perlin noise texture?"

[ionic quartz]

I want to use the gradient of the perlin noise at a specific block like these:

[undone coral]

it really depends how their perlin noise is implemented to get a perfect answer

[sly grove]

take two adjacent samples and calculate the difference between them / the distance between them

that gives you the slope between the samples

[ionic quartz]

Not a texture, I'm going to try adding a second layer of perlin noise on top which has a greater range based on the gradient

[sly grove]

you can do this for both/all of the neighboring samples for a given sample and average them to get the slope "at" that sample point

[undone coral]

well i might as well give you the "betetr" thing

what i wrote isn't really true

because you said derivative but you really mean something else

you're showing me a line, and Mathf.PerlinNoise is f(x,y)

[ionic quartz]

Ah ye I've fixed the y value

[undone coral]

which is more or less prateorblue's answer

[ionic quartz]

I see, that makes sense

[undone coral]

it really depends how right of an answer you need. it doesn't sound like you do

weren't you trying to make a cave path?

you're trying to find "turns" in your path?

i would do so iteratively

[ionic quartz]

Yea I'll show you what I'm hoping for

[undone coral]

in my experience it's easiest when you try to do this "only looking forward"

[ionic quartz]

The darker line is perlin noise, the lighter line is a second layer of noise added where the range is greater for higher gradients:

[west latch]

hey is there an stl in unity like deque in c++

?

[ionic quartz]

Adds the back and forth aspect from my diagram yesterday^

[undone coral]

so you can do

f(t) = g(t)

f(t) = g(t) + f(t-1)

f(t) = g(t) + f(t-1) + f(t-2) + ... + f(0)

[undone coral]

you can either do something with the value you have immediately, only look at the previous value, or consider all previous values

hmm it looks like you're really asking how to layer noise

float NoiseWithFrequency(float x, float freq=1f) => Mathf.PerlinNoise(x * freq, 0f);

@ionic quartz does that make sense?

[ionic quartz]

Would that be for the 2nd perlin noise?

[undone coral]

hmm

i'm doing what you did

which is setting y to zero

do you see how "frequency" just means taking the argument to the noise function and multiplying it with frequency?

[ionic quartz]

I see what thats doing

But would frequency be my (probably clamped) gradient?

[undone coral]

you might want to add an offset so that it doesn't look like the line squished

hmm

no

it would just be

2f.

[ionic quartz]

That's just 1 layer of noise right?

[undone coral]

https://www.wolframalpha.com/input?i=cos(x)%2B1%2F2*cos(2x%2B139013)+%2B+1%2F4*cos(4x%2B7871827481)

do you see what's going on here?

[ionic quartz]

I'm not sure I see what your seeing xd

[undone coral]

okay

well it' sokay

let me ask you this - what is the derivative of cos?

[ionic quartz]

Sin?

[undone coral]

okay

would you say perlin noise looks periodic?

so if you want something periodic with a known derivative, that's one of your options

another derivative of cos(x) is cos(x+2pi)

(-1)^x is also periodic

and weirdly neglected

[ionic quartz]

Why would I need something periodic?

[undone coral]

here is some insight - https://farazzshaikh.medium.com/generating-noise-using-fourier-transforms-b6ccf64afb08

perlin noise is fast that's why people like it

[undone coral]

well you don't

https://en.wikipedia.org/wiki/Perlin_noise#/media/File:2D_sample_of_Perlin_noise.png

"rescaled and added onto itself"

that's what i was trying to illustrate for you

these are f(t) = g(t) methods, "no looking into the past" methods

maybe for cave it'll be easier for you to do, "i look into the past" methods

you can transform many look into the past methods into no looking into the past methods

@ionic quartz https://www.google.com/search?q=biased+random+walk&client=safari&rls=en&source=lnms&tbm=isch&sa=X&ved=2ahUKEwiPvMfk0pT6AhXDLEQIHbhoAUIQ_AUoAXoECAEQAw&biw=1012&bih=987&dpr=2

[obsidian glade]

don't think so, but I'm sure someone will have written an implementation in C# somewhere

[undone coral]

^"biased random walk"

@ionic quartz to me this looks a lot like a cave

but here's the thing, we can hand you this stuff

or you have to like, slog through a lot of math and learn the names for these things

that's 90% of the problem

so if i had to make a cave, i would probably do "10,000 biased random walks across a grid. i keep all grid points that were visited N ~ 1,000 times or more, and all closed loops that were visited M ~ 100 times or more"

or maybe the amount of grid point i keep is based on some other factor

it's an art

[ionic quartz]

I'm understanding bits of what you are saying now but not lots

[undone coral]

yeah

i mean you'll google biased random walk

and it kind of look slike a cave

that's what matters 🙂

i think there is stuff like "function zoo"

where you can just look at a graph of tons of stuff

and fish around for something that looks like what you want

[ionic quartz]

Oh that's cool

[undone coral]

i don't know what it'll be called though

[ionic quartz]

Perhaps this? Except there arent lots of visuals, https://dlmf.nist.gov

[undone coral]

maybe you should make one 🙂

[ionic quartz]

I love the motivation for more resources but I'm not sure I have the passion for something like that xd

[undone coral]

yeah i know

it's okay

anyway this is a fun journey

[ionic quartz]

Prehaps by the time my uni dissertation comes around I'll feel differently about functions lol

[torn basalt]

Hey, I ended up with everything working out really well but I have one more question about vert manipulation. Even after updating the bounds and reassigning the shared mesh, the scale of the mesh doesn't update. I tried using the bounds extents size but that was inaccurate. Is there a way to update the scale after manipulating the verts.

?

[undone coral]

not 100% sure

[torn basalt]

I tied some subtle haptics to the dragging of the verts. Its trigged by changes to the scale of the mesh. I guess I could cycle through the verts being dragged and tie it to the position changes. But if someone has a shaky hand it would constantly call it

sorry just type thinking.

[undone coral]

i think you're going to do great just keep cranking

this is part of the procgen journey

[stable moat]

Hey guys, i have a problem with Bossfight, if someone can help us i would invite you in our discord server (we are 4 people) and can share the screen. would be nice ty 😄

[chilly nymph]

Is there a good way to convert a http request for a .wav into an audio source?

[jolly token]

https://docs.unity3d.com/ScriptReference/Networking.DownloadHandlerAudioClip.html

[chilly nymph]

Thank you 😓 I should have just googled it...

[shy spruce]

Is there any easy way to establish whether a collider is entirely inside of another collider? ie no intersection with the outer colliders perimeter? Or something I have to write myself?

[regal lava]

Could just check for positional values when they collide and clamp a range for what you'd would consider full inside of it.

[shy spruce]

Interesting. It gets weird though with square colliders (box 2d) near the corners I think...