archived-code-advanced

[tame tartan]

the drawings are parented to the canvas, but i dont really know what what does "setting transform z" to be pointing out of the page mean? like making transform.forward match with the page forward?

[undone coral]

yes, out of the page

[tame tartan]

one of the problems is that when i set the drawings to local space, they appear basically somewhere else entirely

[undone coral]

i don't know whether your page is facing 0,0,1 or 0,1,0 or..

[undone coral]

i think you gotta take a big breather

and reason about this in your head first

[regal olive]

Ok I'll ask again then

[tame tartan]

the page only has one rotation, and its -39 degrees on the x axis

[undone coral]

you can try googling "unity draw github" and also see some results

[regal olive]

Does anyone have any documentation on what batchmode does exactly, and specifically how it should affect rendering except for not calling Camera.Render() ?

[undone coral]

i think you can also contact unity support and maybe they can give you a more detailed explanation

[regal olive]

Oh right thank you

I'll do that

[tame tartan]

what i was typing was basically saying that i almost have it working, the positions and rotations match the paper the first time https://i.imgur.com/mxNpZiD.png, https://i.imgur.com/fLQLP9o.png
for some reason, the second time you rotate your canvas, this happens https://i.imgur.com/EOZOon4.png

[undone coral]

okay

when you say rotating the canvas

i see the paper graphic

[tame tartan]

okay i see where the misunderstanding could come in

[undone coral]

i mean, surely the line renderer is parented to the paper?

[tame tartan]

so when i turn my camera, the canvas turns with me

[undone coral]

i asked you if it's supposed to be naturalistic

[tame tartan]

the line renderer is parented to the paper

[undone coral]

is it?

[tame tartan]

the paper is parented to an empty which tracks the camera rotation

[undone coral]

is the goal to have everything behave naturalistically?

[tame tartan]

https://i.imgur.com/3Y2gJPS.png

yes, the first two screenshots are more or less how i want it to act, with the exception of the ends of the lines connecting to the center of the canvas for some reason

[undone coral]

did you try googling the thing

i suggested

[tame tartan]

just did

i found a github but not sure if its exactly what im looking for

[undone coral]

it looks like there are a lot of bugs in your code

[tame tartan]

most likely lol

[undone coral]

i think the hard part is translating the behavior into code

[tame tartan]

there are a few bugs ive fixed between my posting of the code and now, evidenced by how the canvas now rotates the first time at least

but yeah

a lot of it is really messed up

[undone coral]

so

i mean this is for a game jam

you can use the github repos that try to solve this

there are many ways to draw

[tame tartan]

i was kind of hoping that someone would provide me a solution to rotate an array of points around another point in space so i could just scrap what i had and use their solution

[undone coral]

the solution you are asking for is reinventing local space

which if you have all these bugs

why try to reinvent the idea of local versus world space

you're not going to do that correctly, as you can see

[tame tartan]

i suppose so

[undone coral]

so...

you have to use local space

you are actually really close

your raycasts are wrong

you should just try to fiddle with things in play mode

[tame tartan]

hmm

can i run an idea by you

[undone coral]

like change the alignment to transform z

i mean i said all this stuff already

[undone coral]

sure

[tame tartan]

so right now i get the position of the line by raycasting onto the page, then basically lifting the line off the page by moving it in the opposite direction of the raycast direction (so the line doesnt clip into the paper)

but this hit.point is a global position

when i set the line renderer to local space, it makes the lines appear like, 50 miles away from the canvas

do you think that the fact that hit.point is a global position, combined with the local space setting is causing this?

and if so how would i take the hit.point and convert it to local space

[undone coral]

why are you offsetting the point by half a meter

[tame tartan]

let me show you instead of explaining

[undone coral]

well yes of course

because it's not transformed to local space

i've been saying what to do

anyway i think you'll figure it out 🙂

[tame tartan]

https://i.imgur.com/FUld55p.png lines clip into paper if i dont offset it

would i use transform.inversetransformpoint to turn the hit.point into local space?

[undone coral]

yes

okay i gotta go

[tame tartan]

ok thank you dude

seriously thank you

[misty glade]

What's your goal? To ensure you don't forget a particle component on a GO with a script that depends on it? If so, I tend to just add my Assert.IsNotNull() directly in the script itself (since they're compiled out for production builds) - I find out right away if I forget a component or whatever

@regal olive

public class MissionIcon : BetterMonoBehaviour
{
        [SerializeField] private TextMeshProUGUI StatusText;
        private void LocalAssert()
        {
            Assert.IsNotNull(StatusText);
        }
        private void Awake()
        {
            LocalAssert();
        }
}

that's my standard pattern

[jolly token]

If you wanna spend whole day making tests for cosmetic stuff sure 🤔

[terse bobcat]

does anyone know how to add camera shake if a player is running?

[snow rapids]

Kinda feels like one of those tests you write to verify requirements instead of behavior. Having a particle system with spheres is probably not sufficient to prove it's what you need.

[split folio]

By rotating it if the player is running

[undone coral]

i also use the asserts occasionally for library code

as my super discount documentation

we don't have infinite time on this earth to write all the instructions

then again anything i share as source people can always modify if GetComponent based workflows are inadequate

so it's worth just doing everything as easily nice friendly base case as possible and owrry about power users later

[misty glade]

I thought I would try to export my docs from my shared library DLL .. oof.

1500+ errors

(since i have warnings as errors in my codebase)

solution:

👀

[haughty sun]

is it possible to have a custom package has a dependency on a another custom package? because i'm getting errors.

[jolly token]

No unless you use scoped registry like OpenUPM

[hardy sentinel]

fix plz

[crystal ridge]

How can I write my own built in function that has a specific execution time? Specifically, I want to make my own globally accessible (identified by monobehavior) method called LateStart(); that fires off after Start()

Is that something pretty simple to do?

[jolly token]

There is section called Script Execution Order in Project settings

You can make specific script execute faster/later than others

You could expose event from that script for other scripts to subscribe

[crystal ridge]

Awesome, thanks!

[haughty sun]

@jolly token is open-upm widely used?

[austere jewel]

Yes

[dapper nest]

I'm trying to download files through .NET's class Webclient and have a callback when the file is downloaded. I cannot for the life of me figure out why I cannot subscribe to the DownloadFileComplete event though.

https://learn.microsoft.com/en-us/dotnet/api/system.net.webclient?view=net-6.0
anyone that has a better idea than me?

[shadow seal]

Your method needs parameters which match the delegate

https://learn.microsoft.com/en-us/dotnet/api/system.net.webclient.downloadfilecompleted?view=net-6.0

[balmy cargo]

does anyone know why my database initialization doesnt work ?

it's a brand new project with an empty scene that has an object containing the script above

[flint sage]

Probably beacuse it's async and it's not done connecting

[random dust]

Async operation

You're checking for fb_db immediatley after calling Check_Dependencies

Which is synchronous but FireBaseApp.CheckAndFixDependenciesAsync is not

If this method is only done once, consider making a component that initializes it for you when the application starts

Then add a method to it that accepts an Action that can be called when you actually initialized the database

Because you never know when it is

So you store those Actions in a list in the monobehaviour and have an Update method inside of it invoke them all as soon as the database exists

Just remember all those calls will be asynchronous as well, because they are in databases

So the tas that represents this operation will have to be stored aswell

Welcome to Asynchronous programming in Unity

Unity avoided asynchronous programming and now it will bite you in the ass

Creating a general system that can store all of this for you is probably the best answer. You will store all delegates in it and whatever should happen with the result. It's basically the old callback hell that you used to have before async-await programming, but now you manage it yourself

But you can also try it all with ContinueWith(). Just remember that the moment you call _task_.Result, the result might also throw an exception

[mint sleet]

Hello! I have a backend system to make it work in Unity. But do not want to cal APIs directly. I'm thinking to code a library sitting in the between Unity and Backend.

In order not to suffer from versioning in the future, I need to make a very wise abstraction, for sure.

Do you have any suggestion before I start? Thank you

So from user login to all CRUD operations will be handled by this Library with Unity objects.

in the meantime, however, the API v2.0 will come and I do not want to break my previous applications using the version Library 1.0

[undone coral]

.NET's class Webclient
use UnityWebRequest

[wraith snow]

@mint sleet it sounds like you need to separate any calls into your dB into a separate datalayer class or project. Then you can code a data adapter between your client calls and the data layer, which would abstract the calls and help wit versioning

@balmy cargo if your async method needs to finish you can call GetAwaiter().GetResult()

I use that often in business applications where we have async and sync colliding

[undone coral]

are you saying you authored a REST API method POST /login that takes a JSON object of the form {"username": "some username", "password": "somepassword"} ?

[undone coral]

do not do that

[wraith snow]

It's literally the MSDN recommendation

[undone coral]

it is bad advice

[wraith snow]

Why

[undone coral]

it isn't the right way to interact with unity or the firebase unity api

the original user is gone anyway, but i think @balmy cargo will quickly start to comprehend what is going on

without anything too complicated

@balmy cargo doesn't know what the Async means and provided he googles that, he'll be fine

[wraith snow]

You can't await in synchronous context. He needs to either get awaiter or use async code. You haven't explained why you shot down my answer at all

[limpid prairie]

Hey guys, I have a question about detecting "bunched up" gameobjects.

Let's say I have a map of a few mobs in it, and I want to cast a AOE spell onto the area with the most bunched off mobs

How do I detect via where to place the spell?

[wraith snow]

Pretty sure this is traveling salesman

Good luck

[undone coral]

i think as you become more experienced in using unity it will be easier to give good advice

[wraith snow]

I'm very experienced in untiy. You still habet provided an answer

[limpid prairie]

I have map 1), and need to know where to cast spell such that I cast on map 2)

[jolly token]

So you want most optimized place to use your spell

[limpid prairie]

Yes, exactly, but detect it through code, not player controlled

[wraith snow]

You could do a KD tree and spit into certain size zones then count in zones

[balmy cargo]

alright i'll look it up

thanks everyone

[wraith snow]

Or do n^2 comparisons and measure all distances

[limpid prairie]

😳 A KD Tree, true

I haven't touched that stuff since Uni, but I feel like that's its purpose

[wraith snow]

Having an ai always have the right answer could be very difficult

Maybe try randomly selecting 10 units and then see which has the most neighbor's?

You could find a local maximum but not be the most efficient

[limpid prairie]

For most simplest and basic way, I could just divide the map into 4x4 cubes, and run overlap sphere on it, and see which one resulted in most mobs

obviously accuracy wise this isnt better than a KD tree but this is very cheap comp wise

[wraith snow]

Cheap and fast is good you can iterate on it later

[limpid prairie]

yeah true, I'll start with this for now

Thanks!! @wraith snow

[undone coral]

this is a pretty hard problem

[wraith snow]

Aye it's traveling salesman just worded differently

[undone coral]

consider that most of the time, the center of the optimal circle will not lie on a unit itself

[limpid prairie]

Well, once you have a rough idea of the bunched up quadrant, then take all the mobs in it

then average out their pos

[undone coral]

additionally, k-d trees don't really give you the right shape

[limpid prairie]

it just involves many layers of steps

[undone coral]

there is a datastructure for circular ranges

but i think that's pretty premature

consider what the problem would look like in 1D

[wraith snow]

Yeah honestly from game play perspective, randomly selecting a unit and then doing a spherecast for other units or something might be the best bet

But then you would always center on a unit

And you'd have to accept "good enough"

[limpid prairie]

I think what you can do with KD tree is, find out the most populant quadrant,

then discard the KD tree, take the all the mobs in that quadrant, get the center mob, then now, scan all mobs near this mob within X distance, and get average of that to get ur circle's center

[wraith snow]

The goal of the attack is to hit a unit? Wouldn't you rather always be centered on one to increase those odds?

[limpid prairie]

so even if the most "bunched up" mobs fall under 2 squares, and not just 1, then you can still cover both

[undone coral]

hmm... think about what the problem would look like in 1D first

don't get anchored on this kd tree thing

[limpid prairie]

it's a boss jumping up and smashing down at the most grouped up units, so if he misses any targets in the deadcenter, it makes sense

[wraith snow]

I could see it working either way

Up to you

[limpid prairie]

I definitely got some ideas here to expand on it, but Ill do the basic one first

[undone coral]

hmm

really seriously try to do this in 1D

you are getting derailed by someone

[wraith snow]

1d? That's just a line. Do you mean 2d?

[undone coral]

i think i figured it out by following my own advice

[jolly token]

You could adapt dynamic programming

[undone coral]

i wish you would try...

[wraith snow]

You could also try building some kind of heat map and jumping to the place with the highest "temperature"

[undone coral]

it is a dynamic programming problem i don't think there is an improvement from doing it dynamically

[undone coral]

this is close

you don't really care about the region with the most points you don't care about a rectangular area that has a lot of points

[wraith snow]

At least I'm giving ideas not riddles

[undone coral]

a k-d tree doesn't help you

i think the other key insight here is that, why assume there is only one point that is the best location to cast

there will be a region of points

this is clear in 1D

in 1d you can draw lines of length Radius of Spell on each unit. the interval (the region between two points) with the most overlapping lines - any casting position in that interval will have the maximum number of units targeted

@limpid prairie in 2d you would draw circles on every unit, of radius spell effect. any point in the region in your image with the most overlapped circles would have the most units hit

@jolly token in 1d, the dynamic programming is easier to see - i think the reason it works is because as you query more points

the overlapping regions with the most number of points can only be the same length or narrower

i'm not really sure how to translate that into 2d though

it might not be true in 2d if you are querying point by point, there is something else you have to do

yeah i think you can do it point by point in 2d. you will want to preprocess the distance between every point - in this sense the data structure that supports circular range queries will help you. the range you are interested in is 2 * spell casting radius, because that will identify the (narrowing) region of best cast positions

i'm just not sure how you deal with the weird shapes you get when you overlap circles :/

anyway pretty good problem

okay i think it's

This kind of problem is NP-complete problem.

[jolly token]

Wrong one http://www4.comp.polyu.edu.hk/~csbxiao/paper/2003 and before/PDCS2003.pdf

[undone coral]

yeah

i think that makes sense

yeah

@limpid prairie you should use the greedy algorithm from that paper with number of disks = 1

if you have a lot of units you will have to use RKCP2 as specified

and it will be an approximation

but it wasn't tested with 1 disk

thanks for finding this @jolly token 🙂

greedy is fine - it's O(n^2)

[jolly token]

Yeah it is pretty interesting question

[undone coral]

i was having trouble expressing it

now we know

"robust disk covering"

[limpid prairie]

Oh damn wtf

But isn't that computation-heavy though?

Ofc that gives the best results

collision checking in Unity isn't the lightest, but I guess this spell isn't going to be cast often so it's totally feasible

[languid sinew]

Hello, I need help saving a list variable as a json file

and then decoding it back into that variable

[shadow seal]

Search up JsonUtility Unity

JsonUtility can't serialise a List<T> so you'll need to put the List into some wrapper class like

public class SaveBlah
{
  public List<T> MyListOfThingsToSave;
}

[placid jacinth]

As other channels are busy I'll pop this in here, is there a way to convert or pass in a System.Type variable in place of writing the actual class name?

[misty glade]

I did something that works but.. might be godawful, lemme dig up the code

    /// <summary>
    /// If you can read and understand this method without semantic satiation on "type", 
    /// you win a prize: https://stackoverflow.com/questions/43080505/c-sharp-7-0-switch-on-system-type
    /// </summary>
    private Container GetContainerFromType(Type type) => type switch
    {
        Type _ when type == typeof(PlayerModel) => _cosmosPlayerContainer,
        Type _ when type == typeof(SpaceGameSettings) => _cosmosSpaceGameSettingsContainer,
        Type _ when type == typeof(UserMetricsModel) => _cosmosUserMetricsContainer,
        _ => throw new NotImplementedException(),
    };

comment included because it still makes me laugh

i read that article one day and after could not make my brain comprehend the word "type" any longer

[placid jacinth]

lol

[misty glade]

(obviously you don't have to switch/return something, but since you're in advanced I'm sure you can pick out the useful bits without help)

[placid jacinth]

Thanks, I'll try and digest that :)

[undone coral]

i think just do it the right way?

[jolly token]

There is usually a non-generic version accepts Type parameter for those kind of method .. 🤔

[placid jacinth]

pardon?

[jolly token]

There is no AssetUtilities.LoadAsset(type) kinda thing?

[sly grove]

public Object LoadAsset(string name, Type type);

this exists

you need a name/path though

[jolly token]

Yeah for Unity APIs they have it, Idk what AssetUtilities is

[placid jacinth]

AssetUtilities is mine

i can change the function on the other end

it basically returns a list of all assets of a type

[jolly token]

Okay how does that one look like, you can have overload that accepts type parameter

[placid jacinth]

1 sec :)

[sly grove]

you'd have to make the type an actual normal parameter

instead of a type paramater

List<UnityEngine.Object> LoadAssets(Type t) {
  ...
}```

Of course you will lose compile-time type checking here, which is expected since the type is now determined at runtime.

[jolly token]

Yeah then use AssetDatabase.LoadAssetAtPath(string assetPath, Type type); version

Converting to your preferred type later

[placid jacinth]

No errors, so I will test this rn, but is this how you guys meant? (thanks for the help by the way :D)

[jolly token]

Yes

[placid jacinth]

oki checkin now :)

@jolly token @sly grove Big thanks to you both, it works all good, I was driving myself insane over that as usual lol

[maiden turtle]

you can still make it generic if you cast to T after loading

[odd shale]

Hi im trying to read data from my mysql database, and it gives me an TlsException: Handshake failed - error code: UNITYTLS_INTERNAL_ERROR, verify result: 4294936704
code:

[fresh salmon]

You might need to explicitly tell MySql to not use a certificate when connecting to the server. This is usually done in the connection string. You'll have to look at the docs to know how, though.

Fair word of warning though, I googled your error message, and told it to search for this error explicitly (no fuzzy searching), and I got 4 results, in which 3 of them were copies of the first one

[odd shale]

so u r saying my only hope is just to find a programmer that knows how to fix this problem

[fresh salmon]

Yeah

Looks like a Unity issue though, see if you can reproduce the error but outside of Unity, in a console application for example

[odd shale]

You want me to create this issue again outside of unity?

[fresh salmon]

Yep, so you determine whether the issue comes from Unity, or MySql

If you don't get the error with the exact same setup, then Unity is at fault

[odd shale]

OK, also idk if I'm going overkill cuz I'll I need is couple of buttons and gui with mysql, do u recommend something less overkill?

Ok*

[fresh salmon]

Like, you're making a SQL query runner or something?

[odd shale]

It's something for school with a machine we built for producing energy, and every student has a score, so the data goes to mysql and the app just takes the data and displays a leaderboard

[fresh salmon]

So no game at all, just GUI?

[odd shale]

Ye that's why I'm saying I think I'm going overkill but idk any other engine

[fresh salmon]

Yeah Unity isn't the thing you're looking for here, if you wish to keep C# you can make GUI apps in C#

[odd shale]

And now that I think about that, a website would be better cuz ios problems and stuff

Is it easier in Java script to work with mysql?

[fresh salmon]

Depends on the framework that you use, but I'm sure most of them (like Node.js, React, etc.) have MySQL connectors

[odd shale]

Alright, I appreciate the help 🙏

[upbeat path]

DM me for MySql problems

[icy aspen]

hi, i have a question. I only want my touch control buttons to show when the game is running on mobile. The problem is that my game is built on webgl.

[coarse charm]

        public class myClassA:MonoBehaviour
        {
            public myClassB;

            async private void Start()
            {
                await myClassB.DoSomething();
                //continue with other stuffs
            }
        }

        public class myClassB : MonoBehaviour
        {
            async public Task DoSomething()
            {
                await Task.Run( () => {
                    //stuff
                    
            });
        }
``` Im having a NullReferenceException and a TaskCanceledException

I did assigned myClassB through the inspector so myClassA has a reference, any ideas on how to solve this?

[fresh salmon]

First things first, post the full errors, stack traces included

[coarse charm]

okey ```
NullReferenceException: Object reference not set to an instance of an object
Namespace.ClassName+<FunctionName>d__49.MoveNext () (at ScriptLocation:LineOfCode)
--- End of stack trace from previous location where exception was thrown ---
System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw () (at <695d1cc93cca45069c528c15c9fdd749>:0)
System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0)
System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0)
System.Runtime.CompilerServices.TaskAwaiter.ValidateEnd (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0)
System.Runtime.CompilerServices.TaskAwaiter.GetResult () (at <695d1cc93cca45069c528c15c9fdd749>:0)
NameSpace.ScriptThatCalledTheFunction+<Start>d__11.MoveNext () (at ScriptLocation:LineOfCode)
--- End of stack trace from previous location where exception was thrown ---
System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw () (at <695d1cc93cca45069c528c15c9fdd749>:0)
System.Runtime.CompilerServices.AsyncMethodBuilderCore+<>c.<ThrowAsync>b__6_0 (System.Object state) (at <695d1cc93cca45069c528c15c9fdd749>:0)
UnityEngine.UnitySynchronizationContext+WorkRequest.Invoke () (at <3be1a7ff939c43f181c0a10b5a0189ac>:0)

[coarse charm]

the other error TaskCanceledException: A task was canceled. System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.ValidateEnd (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.GetResult () (at <695d1cc93cca45069c528c15c9fdd749>:0) Namespace.ClassName+<FunctionName>d__2.MoveNext () (at ScriptLocation:LineOfCode) --- End of stack trace from previous location where exception was thrown --- System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw () (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.ValidateEnd (System.Threading.Tasks.Task task) (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.TaskAwaiter.GetResult () (at <695d1cc93cca45069c528c15c9fdd749>:0) NameSpace.ScriptThatCalledTheFunction+<Awake>d__91.MoveNext () (at ScriptLocation:LineOfCode) --- End of stack trace from previous location where exception was thrown --- System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw () (at <695d1cc93cca45069c528c15c9fdd749>:0) System.Runtime.CompilerServices.AsyncMethodBuilderCore+<>c.<ThrowAsync>b__6_0 (System.Object state) (at <695d1cc93cca45069c528c15c9fdd749>:0) UnityEngine.UnitySynchronizationContext+WorkRequest.Invoke () (at <3be1a7ff939c43f181c0a10b5a0189ac>:0) UnityEngine.UnitySynchronizationContext:ExecuteTasks()

[undone coral]

what are you trying to do?

it's telling you what the error is

[fresh salmon]

Namespace.ClassName+<FunctionName>d__49.MoveNext () (at ScriptLocation:LineOfCode)
That's not valid, please don't redact information from messages

[undone coral]

you obscured too much of your code

//continue with other stuffs
ah yes, stuffs

[fresh salmon]

Stack traces contain essential info, especially where in the code the exception was thrown

We need that info in order to debug

What I can say though is that both errors are related, the NRE forces the task to cancel at some point

[coarse charm]

yeah, its kind of hard when you have a NDA, anyways Im trying to await the response of a fetched data from a database, and according to whats in the data continue with the logic, the script that handles the fetch is separated from "myClassA" thats why I make a reference through the inspector.

[undone coral]

async public Task DoSomething()
            {
                await Task.Run( () => {
                    //stuff
                    
            });

what is this actually doing?

[coarse charm]

lol

[undone coral]

you should be using UniTask

are you trying to access a database with a blocking IO API?

and you are trying to "taskify" it?

[fresh salmon]

Most database providers provide async calls

Check the documentation of yours

Without the essential info, can't help further

[jagged harness]

Hello. Maybe someone can help me read value of some UUID? I am using Bluetooth LE for iOS, tvOS and Android. My script

public void ReadBattery()
{
BluetoothLEHardwareInterface.ReadCharacteristic(_deviceAddress, ServiceUUID, ButtonUUID, (data, bytes) =>
{
Debug.Log("Read Succeeded " + bytes);
});
}

[sly grove]

is this Java?

[jagged harness]

in respond i am getting "System.Byte[]".

[sly grove]

https://learn.microsoft.com/en-us/dotnet/api/system.guid.-ctor?view=net-6.0#system-guid-ctor(system-byte())

C# Guid is the same as UUID

assuming your byte array is actually a 16 byte guid, you can use this constructor

[jagged harness]

i tried this one - var str = System.Text.Encoding.Default.GetString(result);

But looks like i am not getting value, but just uuid...

[sly grove]

ok so it's encoded as a string

so then just pass that string to this:
https://learn.microsoft.com/en-us/dotnet/api/system.guid.-ctor?view=net-6.0#system-guid-ctor(system-string)

[jagged harness]

Its too hard for me, sorry 😄

[sly grove]

wdym

what's hard?

[jagged harness]

And this is looks like work with UUID not of this value?

[coarse charm]

im looking at it, Im having a hard time to what to use, All I want is to wait for the response and then continue with the main thread

[sly grove]

I don't understand your english here, sorry.

[brisk pasture]

guessing the implementation of uuid is not the same as the guid

[sly grove]

UUID and GUID are the same

[undone coral]

that is very easy

[undone coral]

okay i'll share you the snippet

in a sec

but you have to tell me what htis is actually for

[coarse charm]

fetch data from a db...

[undone coral]

i mean what is the application / game

i'm sure you can say enough in broad strokes

this is just part of being an engineer

[coarse charm]

idk what to answer, like, what genre is it?

[undone coral]

for example, "i am building a virtual test drive experience in Unity for an auto brand"

does that make sense?

or "i am making a first person shooter strongly influenced by CS:GO" <-- valorant

[coarse charm]

why would you want to know that?

[undone coral]

it's just table stakes for getting help for free

[coarse charm]

then it isnt free at all

[undone coral]

i'm saying part of what you can learn today

is how to describe what you are doing

and understand that it doesn't violate any NDAs

i.e. "Not Doing Anythings"

it's too bad because

the solution to your problem is 3 lines

[coarse charm]

x for doubt it? lol

[undone coral]

alright well so it goes

consider that there are certain things that i categorically do not want to help, as i am entitled to!

everyone has a line

i want to help people make games

you're talking about accessing databases

from inside unity

and i'm not the only person who told you that the amount of obscurity you are operating in is so annoying

two other people who have tried to help you ran into basic issues understanding what it is you are trying to do

[coarse charm]

        {
            public myClassB;

            async private void Start()
            {
                await myClassB.DoSomething();
                //continue with other stuffs
            }
        }

        public class myClassB : MonoBehaviour
        {
            async public UniTask DoSomething()
            {
                await UniTask.Run( () => {
                    //stuff
                    
            });
        }```  Im getting a new error, not all path return cant recall what

[undone coral]

🙄 who knows really

@jolly token c'mon "NDA" = "Not Doing Anything" is pretty good

[jolly token]

Honestly, didn't notice it was pun 😂

[undone coral]

lol

[brisk pasture]

thing is that code makes little sense, how are you expecting the new task to run

[undone coral]

i think the user is trying to access a database with a blocking API

but that's just my guess

[brisk pasture]

Tasks and async and await, does not make things concurrent its just tools for working with concurrent stuff and repersenting futures

[undone coral]

i'm not trying to be a blowhard, it just would really make my life easier to know some basic facts about what is going on

[brisk pasture]

like if its truly a blocking database call, and you need it not to be. well you would need to setup your on thread to run the work on providing its thread safe

[coarse charm]

here https://forum.unity.com/threads/how-to-detect-if-a-mobile-is-running-the-webgl-scene.440344/

[brisk pasture]

then you can use the Tasks to represent work happening on that thread

[jolly token]

I was just going to say "doc is being a Google that collects analystics data for their service"

[undone coral]

lol

[brisk pasture]

now it is not too hard to setup your own TaskFactory and TaskScheduler that runs work on its own Thread providing all calls in it are threadsafe

[undone coral]

well i wanted to share

UniTask.Void(async () => {
 await UniTask.SwitchToThreadPool();
 var data = BlockingIO();
 await UniTask.SwitchToMainThread();
 textLabel.text = $"Retrieved {data.id}";
});

as an example of how elegant the API is

[brisk pasture]

never used that lib, just was talking about standard library ways to do it

though looks pretty smooth for one offs

though think i would toss the BlockingIO call in a try/finally, that way 1 exception does not prevent it from switching back to the MainThread

[undone coral]

it won't get into a stuck state AFAIK

it will be peaceful

[undone coral]

in the context of a specific RPG? what are you authoring?

a conventional baulder's gate style RPG?

when you say "level changes" what do you mean?

hmm okay

are you asking how to express waiting for a long time, in code for example, so that it looks like the quest text?

this is a good thing to want

so if your quest sounds like

Find the Dragon and slay it.
Return its tongue to the Mayor.

you want code that looks like the english words? because quests are already algorithmic / procedural?

you're saying "level changes"... i think that's sort of a whatever concern

do you just replace the scene based
replace a Scene? or a scene?

what do you mean

focus on what i'm asking for a sec

do you mean replace the scene as in loading a new unity scene? or do you mean metaphorically replace the scene?

it's just important to use the right words you know, when you ask these questions

and people who ask questions on the unity code channels... they have different experience levels and say the wrong words sometimes, is all

it sounds like you mean an actual unity scene

Currently I'm trying to use visualscripting for making a graph
this is probably a bad idea too

so is using conditions, and all this stuff in the inspector

i don't think you should use Scenes, but that's a different story

i'm trying to help you see things in a clear way, which is that there's this long running Task - c# doesn't have such a thing really

where you are awaiting things to happen

because that's what the quest text looks like

?

hmm it's too bad this was an interesting question. i think you can use a long running Task

i think the core issue is that you don't see things like the quest text as procedural/algorithmic

you didn't know that's what it was

[jolly token]

Is this dialogue tree type thing

[undone coral]

dialogue trees kind of suck

[jolly token]

What part are you having problem with?
Making the system? Managing the data? Populating event?

[undone coral]

like people who write well don't use dialogue trees

they prefer Ink

something that looks like screenplays

and then, if you know how ink really works internally, it's a great insight into how to make quests - really, Long Running Tasks - generally

[jolly token]

Gen Z is coming

[undone coral]

lol

[jolly token]

They don't know what ink is

[undone coral]

ink's engine is, a Task that replays up to where you left off

what people want is Serializable Task

[jolly token]

I want that

[jolly token]

Ah ink is script language 🤔 https://www.inklestudios.com/ink/ interesting

[brisk pasture]

ah have done similar things, where i have game actions and stuff tied into the loc files, so stuff can trigger in certain dialog

[undone coral]

so for example

Find the Dragon and slay it.
Return its tongue to the Mayor.

async UniTask DragonQuest() {
 var dragon = FindNpc("The Dragon");
 await dragon.DiesAsync();
 var player = GetPlayer();
 await player.inventory.OnItemOwnedOrAddedAsAsync(FindInventoryItem("Dragon's Tongue"));
 var mayor = FindNpc("The Mayor");
 await mayor.inventory.OnItemOwnedOrAddedAsAsync(FindInventoryItem("Dragon's Tongue"));
 player = GetPlayer();
 player.gold += 100;
 player.xp += 1000;
}

[jolly token]

Generally I put them in the spreadsheet with conditions that can happen and actions to do 😅

[brisk pasture]

it very heavily depends on the game and its setup

[undone coral]

if you look at the internal scripting system for RPGs, they tend to look like what i wrote there

with the idea being that you can remember where you were along the way of that Task

the state of that Task gets serialized

[brisk pasture]

last time i had to do it, steps of a quest where setup using a Command pattern

[undone coral]

it is hard to do that generally, but not troublesome to do it within the limitations of a game

[brisk pasture]

so it was easy to save state changes from each one taking place

and revert any if needed

[undone coral]

@jolly token i believe the simplest way to do it is to store a "save point" integer that increments for every side-effectful method call, like player.gold += 100

[brisk pasture]

the actually tracking part is just a statemachine per quest

[undone coral]

i think all these systems eventually reinvent serializable tasks

in some way

you can do it informally or formally

[jolly token]

Then I guess the await part would be pseudo?

[undone coral]

not sure... i think you would just wrap it in a decorator

[placid jacinth]

I'm interested now too, I don't mean to disrupt the answering, but with this setup surely as soon as the device is shut down or perhaps the program crashes, you lose any progress made upon that quest, because it's awaiting a async call, not checking conditions when one condition is updated?

[undone coral]

that is passed a container for state

kind of like how react works actually

[brisk pasture]

really feel trying to shove await syntax into this problem just makes it harder to follow and debug

[undone coral]

nah i mean, c'mon look at my snippet, it looks like the english language description of the quest

and absolutely none of this other shit does

anyway i am just trying to do, the formal version of what really big RPGs do

they have a scripting language that is the same energy

[brisk pasture]

you are showing a best case, i am referring to it all, how does it work with our savefiles, how does it reapply its state to the world on load, does it allow undo etc

[jolly token]

Hmmm Yeah this could be somewhat generic

Just restoring state part is not easy and can be messy
I guess it's like DB migration in a term

[brisk pasture]

why i like the idea of things effecting state that needs to be restored using command pattern

[round wharf]

How would i load a c# dll at runtime (il2ccp game)

[jolly token]

AOT and dynamic DLL loading is conflicting by definition.
You'll need to use script language like lua or python or something

[undone coral]

yeah a few approaches

class Quests : MonoBehaviour {
 void Start() {
  // slay the dragon
  GameController.instance.data.AddQuest(async context => {
   // method sets context.Quest variable and other things
   await UniTask.WhenAll(context.player.level.Where(lvl => lvl >= 8).First(),
              context.player.visitedRealms.Any(realm => realm == "Dragon's Realm"));
   // mix animations, rules, dialogue. all in one place
   // why not!
   await context.StartQuest();
   
   var dragon = context.FindNpc("The Dragon");
   if (!dragon) {
     context.EndQuest();
     await UIController.instance.ShowToast("The dragon has already died")
     return;
   }

   await dragon.Killed().FirstOrDefault(evt => evt.Killer == context.player));

   await context.Effect(async () => {
     // do some animations ghost animation
     await dragon.transform.DOMove(new Vector3(...));
     // give the player gold
     context.player.gold += 1000;
     // do an analytics event
     GoogleAnalyticsV4.Event({,,,});
   });
  });
 }
}

[undone coral]

class Quests : MonoBehaviour {
 void Start() {
     // slay the dragon
     // reactive
     var dragon = context.FindNpc("The Dragon");
     var dragonDied = dragon.Hp.Select(hp > 0).Pairwise().Select(pair => !pair.CurrentValue && pair.PreviousValue);
     var dragonKilledByPlayer = context.player.Kills.Where(k => k == dragon);
     var playerLevel = context.player.level.Where(lvl => lvl >= 8);
     Observable.CombineLatest(dragon, dragonDied, dragonKilledByPlayer, playerLevel)
       .Select((a,b,c,d) => a && b && c && d)
       .Where(shouldFire => shouldFire)
       // extension method that doesn't take if you've already taken this once
       // during htis playthrough
       .TakeOnceInPlaythrough("slay the dragon quest")
       .Subscribe(async _ => {
         await dragon.transform.DOMove(new Vector3(...), cancellationToken:token);
         context.player.gold += 1000;
       })
       .AddTo(this);
  }
 }
}

@jolly token lots of ways to skin this cat

in the first example you'd need to use methods on the context or a custom async executor to increment the "step counter" of the serializer for the task

the positive is that it is really easy to read and debug

in the second example you don't need to do that, it can be stateless, but it is way more complicated

just like react it would rerun the quest every time anything in it changes

and that is also hard to implement

the second one might as well just be a subscription to a reactive thing

like rxjs wasn't in React's vocabulary

i might as well have just done an observable

it isn't doing anyhting for me to do it the react way

it would make things harder

[jolly token]

Let's make it modular and decoupled and serializable 😌

Rx is neat for UI but I feel like it's becoming Perl

"Write once, read never"

[undone coral]

lol

yeah i fixed the example

it wouldn't be that hard for multiple step things in observables

it would be "ContinueWithPlaythroughSavePoint("you found the dragon")"

if that makes sense

or really

Observable.TakeMemoized(..., "a name for this checkpoint")

what do you think

i think the first one is more powerful

but really hard to implement

[jolly token]

Thing is when a designer implement this

They have to understand where to save and what to consider

[undone coral]

designer would want context. then autocomplete gives them all the options

yeah

[jolly token]

And I don't trust designers

[undone coral]

that would only be possible in the first case

the designer isn't going to come up with an animation

someone else will do that

[jolly token]

That too 😄

[undone coral]

i personally like to be able to author Timeline content in code

so that's why i think it's important to have #1, it makes it easier to add in animations, dialogue, UI effects, etc.

and just messing with game objects

these are the two approaches that can make these things very robust / low bugs

but only the first approach really looks like the text of the quest

[jolly token]

And that's a good thing yeah

[undone coral]

no i never write tests for unity

someone i'm sure has tried to do serializable tasks

in java i used a serializable fibers library but never serialized anything

wasn't necessary

i think akka can also serialize actors

very Erlang

it is nice to be able to just declare a variable, side effects be damned

if you wrap side effects in context.Effect(() => {}) you can do it

[jolly token]

Hmm only if we can just jump in to middle of Task

[undone coral]

it's like writing haskell

[undone coral]

yeah in the java library it was smart

it kept a program counter

and leapt through the block points

[jolly token]

Maybe possible in world of pure function

[undone coral]

so if you deserialize a fiber, it (1) restored all the scoped variables and (2) skipped all the code up to the save point

[jolly token]

Isn't versioning gonna be very challenging for that?

Like how do you even modify code with it 🤔

[undone coral]

it's only possible to restore a previous version if the variables (local, closure) stay the same

but that's true in the RPG script code too

i wonder how those RPGs migrate for patches

probably jsut declares a "migrate" method

[jolly token]

I wonder if that is possible in C#

[undone coral]

someone i'm sure has done it

[undone coral]

the cheapo way is to rerun the function, and just don't run the stuff in context.Effect

that's the Lispy way to do it

[jolly token]

tHaT's NoT pReTtY

[undone coral]

very cheap

yeah i know

but in my example it would work

and you wouldn't need a degree in CSharpology to pull it off

you would just hijack await

which you can do

...
Task Effect(string name, Action<T> action) {
 // internal
 Context.current.SetEffect(UniTask.Run(() => {
  if (Context.current.HasExecuted(name)) { return; }
  await action();
  Context.current.SetExecuted(name);
 }));
}

there is a C# api that looks like this

[jolly token]

Hmmm right

Just wrap that think with class then it can be generic something

[undone coral]

^ something like that. SetExecuted would write to your save file

could be even simpler

skip the action if the save file shows it was executed

whaddya think?

this is if you're rerunning the whole function

in a language that cannot guarantee purity

= bugs

i would rather allow it to be impure but use a custom async executor

[jolly token]

The things that really irritating would be exception handling

[undone coral]

exceptions? who makes those

lol

[jolly token]

In general term lol
For example you give user a 500 diamond for initial draw
And you do the draw, but the game crashed

Hmm no you send server call and game crashed

Now the diamond is gone and your Task is deadlocked

[undone coral]

yeah

server call would be impure

i am not familiar enough with haskell

to say how it Deals

but there's an approach

[jolly token]

Monad?

[undone coral]

not

sure

[jolly token]

Only if C# let us overload >> without constraint

We could do monad with pretty >>=

[undone coral]

lol

that is essential

when i think language features, i think, shit we need an operator

[jolly token]

Like when C++ developer became so excited and made std::cout

That's probably why C# banned overloading >> without int param

beginning with C# 11, the type of the right-hand operand of an overloaded shift operator can be any.
OH shit dream comes true

[undone coral]

lol

[wet burrow]

Something is seriously wrong with my while loop. You can see the array has a length of 14 and it's supposed to destroy 2 objects and have a length of 12 by the end. But while it's reduced by two, the loop only runs once and only one of the objects is destroyed.
https://cdn.discordapp.com/attachments/377316629220032523/1021997202899480627/unknown.png
https://cdn.discordapp.com/attachments/377316629220032523/1021997203260178502/unknown.png
The only explanation I can think of is that one of these functions is somehow also removing an entry from the array, so it runs twice in one loop while only destroying one object.

[jolly token]

Does your object remove itself from list on OnDestroy or something

[wet burrow]

@jolly token the object is a Toggle (as in the regular UI component)

it says there

hang on I oughta double check

hang on it might be a ToggleGroup thing

[flat marsh]

Does by chance the second object have the same name as the first und you have collapse on in the console?

[wet burrow]

nope definitely not collapsed. I ended up fixing the problem (even if I didn't figure out the specifics of what was causing the bug in the first place), by declaring the list object and removing it before destroying it

[jolly token]

some editor flexible enough to copy-paste (not exactly, will always need small edits but) it into different projects and easy enough to use by dudes who cannot into programming.
Excel! 😄 https://github.com/cathei/BakingSheet

Not that complex but what Clash royale does
https://github.com/smlbiobot/cr-csv/blob/master/assets/csv_logic/tutorials_home.csv

[jolly token]

This also sounds like anti pattern my friend.
First, you are relying on static states (FindObjectsOfType*)
Second, it is still hard to refactor things and track references
Third, how would you handle dynamic object that instantiated?

I think he meant what he wrote- since it is suggested as replacement of singleton

[brisk pasture]

really, its unity is just a pain in the ass with this, since you do not control object construction directly for most things

and do not have just 1 entry point

being too strict about any approach is just asking for headaches

[jolly token]

That would be the root of the problem 😌

[brisk pasture]

but i often end up with a service locator, that lets me request objects registered to it by type, and lets me easily swap those implementations out

in literally everything else i do that is not Unity, this problem is fully avoided, since i control object construction and what order things are constructed in, so pretty easy to do dependencies first, and pass them down as constructor args

[maiden turtle]

use | like C++ streams

or rather ranges

[jolly token]

I wonder, maybe service locator is better pattern for Unity as they aligns with GetComponent series 🤔

[jolly token]

lol

C++ has it all

[maiden turtle]

meh, di is better

and it also could handle dynamic objects

[jolly token]

It’s not the problem with using API. You are getting your states from static states without context, that’s not flexible enough IMO

I’d prefer generic solution like DI really. With this I’ll have to write lots of ad-hoc code to handle state and context properly.

I guess you can just go with ECS pattern if you prefer

[orchid marsh]

di, referring to dependency injection?

[maiden turtle]

yes

aka passing dependencies on intitialization

you'd make an interface for invoking that event and just initialize the thing doers with a reference to the implementation when you create them

how you're going to pass that reference is your choice

you can do every object individually in a test

passing in a mock

you can use a di container

you can access a static singleton instance

di is more dynamic in this sense

but if you have a framework than it would likely use reflection and bind it automatically from its container

I instantiate everything manually

zenject does reflection to bind the dependencies

you'd have an initialization method, and it would bind the objects according to some rules that you define

but you still have the choice to initialize any object manually

I just have an initialization routine where I initialize everything explicitly

If I needed dynamic objects that have dependencies, i'd create a factory, explicitly

Might be more code and somewhat clunky, but the good thing is that there's no magic involved

[maiden turtle]

look at how zenject works, or check out di in aspnet core

essentially, its just an initialization method for you monobehaviours

and passing in the dependencies

[jolly token]

How is it advocating singleton? Using singleton with DI is compeletely fine because the fact that it is singleton is hidden, you can always refator or swap the implementation

[brisk pasture]

you are just being pandantic about it, its fine to have a few if they save you from other problems

[jolly token]

Your example is also a singleton as only single instance would exist. The problem is not singleton, the problem is “singleton pattern”

Btw my DI container impl does not use singleton anymore since I refactored it

[brisk pasture]

think you are way more concerned with what you call "good code" then getting stuff done then

like having 1 object as a locator or a di container that is singleton is fine

more then fine, since it provides 1 entry point where dependencies for runtime or test can easily be setup

[low mesa]

Hello everyone!
We are facing a rather odd issue, and we are unable to solve it so far.
Here is the situation:
We have a few packages imported into our game project, and one of them is
"collections"
2 of our components are relying on this package, but one of them is looking for 1.0.0, and the other one is looking for 1.2.0.

We don't have any problem until we try to build, no error msg nothing, but the build can not be done.
How can we segregate the 2 packages, or what other options we have?

We need both component to be working.

[jolly token]

That’s fine, debating for better design is a good thing

[brazen hamlet]

    private void Update()
    {
        if (Input.GetKeyDown(KeyCode.W))
        {
            reducedHP();
        }

        if (Input.GetKeyDown(KeyCode.E))
        {
            Noway();
        }
    }
    public void reducedHP()
    {
        healthpoints--;
        if (healthpoints < 3)
        {
            SecondPhaseStarts?.Invoke();
            FirstBoss.SecondPhaseStarts = null;
        }
    }

    void Noway()
    {
        healthpoints++;
        if (healthpoints > 5)
        {
            FirstBoss.SecondPhaseStarts = null;
            SecondPhaseStarts?.Invoke();
        }
    }``` 

On "SecondPhaseStarts.Invoke". basically when this object reach below 3 on its healthpoints, the other gameobject from another script that subscribe to this event has stop spawning. When it's more than 5 healthpoints, it respawns. It works! 

BUT, when i made the healthpoints below 3 again, it still spawns and didnt stop. pls advice? 😦

[undone coral]

hmm can you build an empty project?

it sounds like you have a complex project

2 of our components are relying on this package, but one of them is looking for 1.0.0, and the other one is looking for 1.2.0.
this doesn't really matter. they will just both use 1.2

it doesn't work like npm, this is a red herring

how do you know you aren't building? what is actually happening?

[undone coral]

this isn't really #archived-code-advanced, try #💻┃code-beginner !

[low mesa]

@undone coral
Thank you for the reply!
Yes indeed it's a complex project. Atm we trying to implement Singular (the analytic tool) and it causes this problem. Let me grab the msg we get during the build process.
(actually one of my co-worker making it, will ping you again in a bit)

[undone coral]

can you build the project?

[low mesa]

no. If the Singular package we have at hand is included, the build fails. (I'm still waiting for the report from my co-worker)

but if we don't have the Sing. in: we all good.

[undone coral]

okay so...

i'm sort of stating the obvious here but why don't you contact them

nobody has any experience with installing this obscure thing delivered as a unitypackage

[terse bobcat]

does anyone have an ide how to add a decent camera shake while running

[sturdy edge]

google "how to add head bob unity"

[terse bobcat]

bro you safed my life ❤️

[low mesa]

We did. Not much progress so far (2 days...)

This is the package problem that I have mentioned

• What went wrong:
  Execution failed for task ':launcher:checkReleaseDuplicateClasses'.

1 exception was raised by workers:
java.lang.RuntimeException: Duplicate class androidx.collection.ArrayMap found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.ArraySet found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.CircularArray found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.CircularIntArray found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.ContainerHelpers found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.LongSparseArray found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.LruCache found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.SimpleArrayMap found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
Duplicate class androidx.collection.SparseArrayCompat found in modules collection-1.0.0.jar (androidx.collection:collection:1.0.0) and jetified-collection-1.2.0.jar (collection-1.2.0.jar)
--

I was wondering, maybe we could just segregate the 2 packages with namespaces maybe? Or any other way to do that? Because as far as I can see the problem here is the fact that we have 2 of the "kinda" same packages.
If I could separate singular's stuff, from the others... could that work?

[undone coral]

okay

i mean you said collections package, i had no idea which package you meant, and here you clearly have an error message

and you're talking about an issue with the android build

it sounds very complex for you

you should wait for their response

[jolly token]

Question: Is there any deterministic/decimal type that support big exponent value? Precision is not that important.
i.e. The deterministic version of https://github.com/Razenpok/BreakInfinity.cs

[flint sage]

Bigint?

[jolly token]

It'd be too big and slow. I don't need that much of precision, I want more of incremental-game kinda number

They are not deterministic since they are using double under the hood, that’s why I’m finding something else

Because I need deterministic calculation across platform?

[sly grove]

Can you use two ints or longs. Basically scientific notation?

int baseValue;
int exponent;```

or maybe:

int hundredths;
int exponent;

Such that:

hundredths = 534;
exponent = 7;

Would indicate:
5.34 X 10^7

you'll have to write all your own arithmetic functions with this but it should work for what you need and be deterministic.

[compact ingot]

See if Decimal works for you. It’s much less prone to rounding errors at the cost of a reduced value range and 16 byte size.

[jolly token]

This is good, though I’ll need to implement some complex thing for multiplication etc. So I was finding if there is published solution first

[jolly token]

I’d love to use decimal, sadly I need less precision and more exponent

[compact ingot]

What exactly is your issue with a regular double?

best option if double doesn’t work is to just duplicate any other floating type implementation with a custom exponent size

[sly grove]

I think floating point is the problem in the first place

because the math is not deterministic across hardware

it needs to be deterministic fixed-point arithmetic. Although the use case hasn't been shared actually...

[compact ingot]

All I have taken away from reading about that once is : just don’t even try

[sly grove]

right - don't try to get deterministic floating point math. Do fixed point math instead!

[jolly token]

I’ll just say that I need to run input synced simulation (e.g. lockstep) with incremental stat number 🥲 I guess I’ll have to implement custom type

[haughty sun]

is Dear IMGUI used in the industry for unity?

[sly grove]

pretty sure Unity's own IMGUI (legacy ui) library is basically a C# port of Dear IMGUI

[haughty sun]

that shit giving me nightmares

I was making a imgui console for development but i realized the package doesnt work good as i supposed, don't wanna code it unity's own either...

[brisk pasture]

i would not say a port, but yeah its a IMGUI so heavily inspired by dear imgui

i really do not mind IMGUIs for tools

[timber flame]

Which one do you choose?
Find the shortest path from a source point to several target points (specific type of a tile)
a--> (b1,...,bn)
Output: a--> bm
1- Dijkstra
2- A* for each target point

[brisk pasture]

the 2 things are almost the same

just easier to influence A* into working in slightly different ways

[sly grove]

A* is just Djiskstra with an extra heuristic function

[brisk pasture]

also if 3d or working on the xz plane i would just let unity do it for you

[timber flame]

I know what it is

[brisk pasture]

its A* on a navmesh

[timber flame]

I want to know which one is faster you guess

Run A* for each target point or Dijkstra

[sly grove]

A* is faster if the hueristic is good

otherwise it's slower or equal to djikstra

[brisk pasture]

A* with the right hueristic will check less nodes

[timber flame]

Because that tile exists in many positions. So, there are many target points

[sly grove]

this is irrelevant

both algorithms find the shortest path to all nodes or specific node depending on how you write/configure it

[brisk pasture]

alot of the optimization for both of them will come down to proper data structures for the open and closed sets

and reducing the amount of nodes

[timber flame]

I think classical A* works with one target point!

[brisk pasture]

what are nodes in your case

[sly grove]

Both algorithms work with one target point

[brisk pasture]

grid tiles, a quad tree like thing or triangles?

[timber flame]

Simple grid

I don't know what you mean

Dijkstra's algorithm (/ˈdaɪkstrəz/ DYKE-strəz) is an algorithm for finding the shortest paths between nodes in a graph

So, it does not have one target point

finding the shortest paths between nodes

It creates a matrix [v,v]

[honest hull]

Hey so if I have a .asset file
how do I extract the mesh for it so I can modify it(change its index's from 16 bit to 32 bit) and save it back to the asset file?

[sly grove]

it does have one target point

it's literally the same algorithm as A*

with one fewer function call

I've written it in college (10 years ago but still)

it builds a matrix for memoization purposes

[timber flame]

If it finds the shortest path between all vertices, dude read

I have implemented it 15 years ago as well

[sly grove]

Only if you run it from every node in the graph and don't stop it after finding the node you're interested in.

[timber flame]

It is classical algorithm, right

see wiki

[sly grove]

hmm strange
how the wiki is showing it being used to find the shortest path between two particular points

[timber flame]

Dijkstra doesn't use heuristic

A* does

[sly grove]

exactly

that is the only difference

[brisk pasture]

heuristic is just the hcost function added

[timber flame]

I think you do not know the algorithm.
It should find the shortest path between all points because it uses the info

to go to the next

[sly grove]

I know the algorithm. If you let it keep running it will find the paths to all points

if you stop it when it finds the point of interest

then you're fine

The difference with A* is you need to start with a goal node in miond in order to have a hueristic function oriented toward that goal

that's the only difference

[timber flame]

yes, but A* classic heuristic is only for one target point

[sly grove]

yes

[timber flame]

it does not work with multiple at all

[sly grove]

right

well acutlaly

it DOES

but

the hueristic just becomes useless

that's all

you can let it keep running after finding the target node, just fine

[brisk pasture]

its also easy to just but it on a flag and disable that one added feature

[timber flame]

yes, useless

[sly grove]

I guess given what you're actually asking here

djikstra is most likely more efficient since a lot of work won't need to be repeated. But you could also just modify A* somewhat and have it re-use the costs calculated in previous runs just fine

modified A* would be fastest probably but simpler to just run djikstra until you see all your targets.

I am working on a similar project - tower defense game - where I precompute all the shortest paths up front

and have all the enemies just follow those precomputed paths

ultimately the performance ends up not mattering that much if you can precompute the paths and reuse them.

Which may or may not be applicable to your game

[timber flame]

1- Choose n closest target points (n-nearest search)
2- Sort them by distance
2- Run A* for the closest one
3- After reaching, remove that target point and repeat
I think it is OK

[timber flame]

Unfortunately, in my game, the roads are dynamic

[sly grove]

in my game too

I do the pathfinding recomputation when the player changes the grid

but that happens in a separate phase from the enemies moving around

not sure how yours works

[brisk pasture]

depending on the unit count and other factors, maybe want to look into more specialized approaches like what a rts game with high unit caps would do

[sly grove]

At some point on the spectrum from "finding shortest path from a to b" to "finding shortest path from a, b, c, and d to e, f, g, and h" I went from using classical Djikstra/A* to more or less just using Breadth-First-Search and doing lots of memoization and it went better.

[gray pulsar]

@timber flame you could also modify A* and/or Djikstra to pass in the heuristic and end condition as delegates, rather than having them hard coded as a target node + distance function to that node

not sure how bad that would be for performance, though

[timber flame]

I have implemented flood fill/BFS

but now, we want weighting/priority

[brisk pasture]

seen some approaches where instead of working the path backwards from destination like you do in A* or Dijkstra's, where it processes it all and saves directional vector information in each tile

[sly grove]

this is literally what my current project does haha

i call them signposts

the AI just follows them

[brisk pasture]

yeah supreme commander does a approach like this

[gray pulsar]

like flow field pathfinding

[timber flame]

Yes, creating them is heavy. Afterwards, it will be OK

Storing the direction

[brisk pasture]

why i mentioned it depends on your unit cap

its way more expensive for only a handful of things following

[sly grove]

yeah so you need to weigh your write/read ratio basically and decide if the cost to create them is worth it. In my game I have hundreds or thousands of enemies that all reuse the same paths so it's absolutely worth it

[brisk pasture]

but if you are getting into hundreds of path followers

[sly grove]

if your graph changes a lot, the cost to create them may be too high

[gray pulsar]

also depends on whether your units are navigating to the same place or different places. If everyone is going to the same destination from different origins, flow field can be really efficient

[brisk pasture]

really i just use pretty basic A*, but i made my data structures as effecient as possible, and reduced the amount of nodes as much as possible

also run it on its own thread

[sly grove]

It's still usable with multiple destinations, you basically just have overlapping fields, and the unit uses the one for the destination it cares about

as long as the number of units reusing the field is high it can be worth it

[brisk pasture]

yeah its good for the rts select like 100+ people send them to one area

using a mesh or quad tree really cuts down on the node count

over evenly spaced grid

and A* and Dijkstra's do not care what the data is, as long as you can get your neighbors, and the more convex space you can represent with 1 node the more you can rely on stuff like local avoidance while also reducing the amount of nodes that need to be considered even for getting costs for every tile

[umbral nimbus]

Hey guys, I'm trying to get this system of re-orienting a 2d isometric object to work. It works by containing 4 sprites of the same object in each cardinal orientation packed onto the same GameObject.

The problem is, the update function doesn't seem to work:

    public void setOrient(int orient)
    {
        tiles[this.orient].color = Color.clear;
        this.orient = orient;
        tiles[orient].color = Color.white;
    }

When I run this code the tile stays transparent

[chilly nymph]

Are you setting the Transparency / Alpha?

[umbral nimbus]

yes

[chilly nymph]

I don't see it.

[umbral nimbus]

Color.clear is (0,0,0,0)

Color.white is (1,1,1,1)

[chilly nymph]

Sorry I didn't know the enum supported lower cases.

Debug to check if you're on the right tile?

[umbral nimbus]

sure, hold on

yeah pretty sure that's the right tile

[chilly nymph]

I mean in code.

[umbral nimbus]

oh

[chilly nymph]

tiles[this.orient].color could be referencing a null?

[umbral nimbus]

ah. Pretty sure that isn't. Because it's referenced in the start() function

and it doesn't throw a NullReferenceException

[chilly nymph]

Drop a bp

[umbral nimbus]

bp? Sorry, I must be terribly new at this

[chilly nymph]

Breakpoint.

[umbral nimbus]

ah

ok

ok. So I just figured out how to debug mode. Thanks

hmm, I'll fiddle with it on my own for a while. Don't want to eat up too much of your time.

[grave vortex]

how can I replace the skybox material in a custom render feature?

I need to sort this out.

[bronze geyser]

Is there a way to declare a vertex on a mesh as an "anchor" and reposition the mesh relative to said anchor point?

I've way too many hours trying to position this procedurally generated mesh and my everything hurts

[lament salmon]

subtract the anchor vertex position from all vertex positions, no?

[bronze geyser]

Oh, that would make the anchor position 0,0,0

That makes sense

Excellent

Thank you ❤️

[craggy spear]

This is a code channel, for advanced code questions. Delete from here and ask in #💻┃unity-talk - though you'll be better off posting on the forums with that question

[regal olive]

Ok, thank you

[flint wraith]

Hello, have question, trying to implement creating of scriptable object with inheritance yet unity doesnt know what to do.
Anybody tried something like this? What am I doing wrong

Error happens when i create scriptable object and try to drag it to object with Test script

[fresh salmon]

What's the file name where class HelmetSO is declared?

[flint wraith]

@fresh salmon all files I have

[fresh salmon]

When you select the HelmetSO file in Unity, and look at the Inspector, do you see the code or not?

[flint wraith]

yeah,

[fresh salmon]

Setup looks good to me

[flint wraith]

yeah, i did initially put everything in Test script and split them after error into these files, but refresh should have handled that.
maybe some reset unity functionality to rebuild it?

[fresh salmon]

Yeah, it's weird that it says "None" on the SO instance file

Delete the SO (not the instance, the class itself) and re-create it to force a complete rebuild

[flint wraith]

Yeah, it fixed the problem, thanks hahahahah

[hot lintel]

I am trying to add the burst compiler to Unity 2019.4
I downloaded the "2019.4 Verified" package, and I get 5 errors. Each of them are the same but for different systems.
Basically it complains that I don't have XboxOne, XboxSeries and PS5 in my list of build targets.
Library\PackageCache\com.unity.burst@1.7.4\Editor\BurstAotCompiler.cs(259,47): error CS0117: 'BuildTarget' does not contain a definition for 'GameCoreXboxOne'
Does anyone know what version does not have this problem?
Because it's quite silly it's "2019.4 verified" yet requires build targets I am not actually going to build for

[wide loom]

When using INetworkSerializable is there any difference between doing it these 2 different ways (examples are different structs)?

 public void NetworkSerialize<T>(BufferSerializer<T> serializer) where T : IReaderWriter
        {
            serializer.SerializeValue(ref firstname);
            serializer.SerializeValue(ref lastname);
            serializer.SerializeValue(ref number);
            serializer.SerializeValue(ref heightScale);
            serializer.SerializeValue(ref widthScale);
            serializer.SerializeValue(ref speedScale);
        }

or this

public void NetworkSerialize<T>(BufferSerializer<T> serializer) where T : IReaderWriter
        {
            if (serializer.IsReader)
            {
                var reader = serializer.GetFastBufferReader();
                reader.ReadValueSafe(out name);
                reader.ReadValueSafe(out color);
                reader.ReadValueSafe(out logo);
                reader.ReadValueSafe(out players);
            }
            else
            {
                var writer = serializer.GetFastBufferWriter();
                writer.WriteValueSafe(name);
                writer.WriteValueSafe(color);
                writer.WriteValueSafe(logo);
                writer.WriteValueSafe(players);
            }
        }

The first seems cleaner and shorter, why would some examples do it the second way?

[hot lintel]

What if you don't wanna work with refs?

[wide loom]

Is there a reason you would not want to use refs?

[hot lintel]

A ref is not always desirable because of mutability.
Some times you want to keep things local, other times you want to work with references.

It really depends on your use case I'd say.

[gentle topaz]

good APIs leave the lower level options open for people who have very specific needs

[hot lintel]

☝️

There is also a lengthy explanation here on ref vs not ref https://jonskeet.uk/csharp/parameters.html

In most cases you'll never actually need to use ref

It has some nice use-cases, but it's rarely actually needed

[pulsar tundra]

Hi, is there a reason my Quaternion only produces half of the required rotation here?

[flint sage]

Why are you trying to construct a quaternion manually

[pulsar tundra]

Because every solution I've tried for my problem results in Gimbal lock, or undesired rotation

[digital helm]

Has anyone used babylonjs?

I'm trying to figure out how to convert the angles of the camera there into unity

Looks like Z there is facing the other direction

but can't seem to figure out how to convert the angles

any help would be greatly appreciated

[brisk pasture]

@digital helm would need to figure out what the up and forward axies are

then also if its left or right handed coord system

[sly grove]

this is not how you make a quaternion

what are you trying to do

[pulsar tundra]

rotate whole character around the tetherPoint

I've tried lots of diff things, AngleAxis makes the rotation bug when the Axis changes, using Dot makes the character spin like that when the Vectors are completely opposite

[sly grove]

rotate around in what way

that's pretty vague

what's the "tether point"? Is that where the rope touches the platform?

[pulsar tundra]

so the play should be able to hang parallel to the rope attaching it to the tether point which is where it intersects with the platform

so it needs to rotate on X by Vector3.right and Y by Vector3.forward

and then if I try to get the Angle from 2 Vectors, when I combine the rotation it's just completely wrong

[undone coral]

there might be an asset store asset for this. have you checked if Kinematic Controller has this kind of effect?

ref Quaternion currentRotation and constructing your own Quaternion suggests that there are

so many bugs

that some of them are interacting together

to produce positive (expected) effects

so it's going to be a lot of back and forth of us saying "well what if you fix X in your code" and you're going to say "no that breaks Y"

"i just want this to Z"

AngleAxis makes the rotation bug when the Axis changes, using Dot makes the character spin like that when the Vectors are completely opposite
this is because there are so many bugs, changing any one thing in isolation breaks some unintended interaction

[pulsar tundra]

alright np

ill figure it out im sure

[undone coral]

basically your code is this scene in the Simpsons right now @pulsar tundra

[pulsar tundra]

just been stuck on this for so long

[undone coral]

there are so many bugs they are in perfect balance

@pulsar tundra well you can always rewrite things

programming isn't VFX, you can't just tinker and layer on things until it works and then you ship it

it sounds like you want to have a 3d game where characters tether to a platform and swing around it

and you have a specific opinion of how the tethered versus untethered (free fall) behavior should be

@pulsar tundra if you want the tethering to behave naturalistically, you can use the physics engine to get this effect

would you describe the desired behavior as "naturalistic"?

[digital helm]

its y up and right handed system

not 100% sure how to check the forward axis tho

but i think its Z

[pulsar tundra]

yeh, literally everything else i.e gravity, collision etc works perfectly I just dont understand the rotation and how I'm gonna make it rotate correctly

[undone coral]

rotate whole character around the tetherPoint
what does that mean?

in the video you shared, you're showing a character moving about a tether point

i can see why you're using the word rotate

but there's rotation and rotation, you know what i mean?

are you trying to make the character move naturalistically by moving the character around the surface of a sphere centered on this tether point?

or do you want Physics to do that for you? @pulsar tundra

it isn't clear from your code or your words what you want

do you want to orient the player model a certain way?

the code snippet doesn't tell us, and you haven't told us

so we're confused

[pulsar tundra]

yes, the character should rotate around the surface of a sphere towards what I believe is its tangential velocity, perpendicular to the radius

[undone coral]

there's physics and then there's Physics

[pulsar tundra]

yeah I get that

[undone coral]

so why aren't you using Physics to do that?

[pulsar tundra]

well im using an Asset for the character controller like you said

[undone coral]

okay

[undone coral]

okay well here's how i would do this with code, without physics

the tether point is the center of my sphere, the length of the tether is the radius of the sphere - it looks like we're assuming the tether is of a fixed length / rigid, which is fine

i would draw a vector from the center to the player's current position. then, i would rotate that vector by a small amount: the angular velocity * Time.deltaTime, whose units work out to be a small amount of rotation.

that's pretty much it

the resulting vector is the new position

as soon as the tether is disconnected, i would transfer the angular velocity into Physics velocity

[pulsar tundra]

right I see

[undone coral]

i would not express angular velocity as tangential velocity

as soon as the tether is connected, i am assuming the tether is 100% rigid, so i will assume that the amount of linear momentum that gets turned into "angular" momentum (angular with respect to how i am simulating this, which is the rotation of a big sphere)

looks something like

angular velocity = Vector3.Cross(tether vector, velocity) / (length of tether squared)

this will correctly result in a lower angular velocity of this big rigid sphere i am rotating

if the tether is longer

is this helpful?

[pulsar tundra]

yes this helps a lot thankyou

[undone coral]

you got really into the weeds i think with the code you already have

so the gist is

when my tether connects, transfer all of my momentum into the rotation of a big giant sphere centered on my tether point

and we are solving for angular velocity in the equation

tangential velocity = (angular velocity) W cross (position of point on this sphere) R
tangential velocity = instantaneous particle velocity at moment of tethering V

[pulsar tundra]

okay that makes much more sense now

[undone coral]

i wouldn't describe this as physical

it is sort of naturalistic

it's really very gamey

but based on the video you shared that sounds like it's what you want 🙂

imo, you should actually place a rigidbody into the scene that models the sphere

whose mass is the player's mass

or whatever

and actually transfer the momentum

and this way you can still do it with Physics

[pulsar tundra]

okay yeah I'll give that a go too

[undone coral]

with the right collision layers it will work 🙂

you can also... do the actual tether

but it sounds like you don't want totally physical behavior

an actual rope would behave weirdly*

even this approach will behave surprisingly sometimes

like if you have a very small angle between your current velocity and the moment you connect the tether

in real life, the rope would bunch up

in this, you will suddenly go into a sphere moving very very slowly

[pulsar tundra]

yeah I have an idea for some things like at the top of the swing I cancel out Velocity so that the player falls with gravity

i can do that now

just tryna get this rotation part down coz im so confused

[icy aspen]

I have a problem, im making a game thats currently on webgl, i want to show touch controls only when the player is playing on mobile, but not when the player is on pc or something else. How can i do this?

[sly grove]

you can check if there's a touchscreen device available or not

I also think it would be a good idea to give the option to use touch or not when available because a lot of PCs also have touch screens these days.

[west scarab]

Using the manual nav mesh builder api: Is it fine to use many instance of NavMeshData and adding / removing them from the global nav mesh, or does this have performance implications? I couldn't find anything in the docs

[regal olive]

it works :)

[iron urchin]

I’ve got a question I hope one of you fine programmers can answer. In games that feature procedurally generated worlds, like Minecraft or No Man’s Sky, is the world completely generated at launch and held as data which is rendered only when the player reaches a certain proximity? Or is the creation and rendering happening simultaneously?

[regal olive]

second answer

[iron urchin]

Interesting. As a follow-up, would the first scenario be possible, specifically in an engine like Unity?

[regal olive]

Yeah sure, if the world is finite

The only general rule about what's possible or not is your hardware and the halting problem

[rugged radish]

both approaches are used, it depends on your goal
those 2 examples use the second approach

[iron urchin]

Can you provide an example that uses the first approach?

[lavish sail]

first one would take a lot of memory

way too much

[regal olive]

I would use the first approach to eg save it on disk if loading it from disk takes less time that computing it

[lavish sail]

uhh WorldBox I guess

maybe Terraria, not sure though

[rugged radish]

terraria generates and saves the whole world at start

[lavish sail]

terraria has a lot less to save because the world is 2d and finite

[regal olive]

You also need to generate it full if you need some sort of consistency between each part of the world

if some part depends on another then you kind of have no choice

[iron urchin]

So in theory one could generate the world on initialization, saving the mesh and object location data in arrays and sub arrays, storing those on disc, Then allow subroutines to access and alter that data at runtime?

[regal olive]

Yeah but be careful, disk is expansive

[lavish sail]

well if it's optimized enough

[rugged radish]

you probably don't want to save the mesh, but only the data, like heightmaps, biome maps, things like that
and generate the mesh at runtime

[iron urchin]

maybe I am miss using the term mesh. What I mean is the matrix data and world position locations. Is that not just an array of numbers that can be stored and accessed relatively inexpensively? I apologize I know I’m posting in the advanced code section, with admittedly not an advanced knowledge of practical coding.

[regal olive]

Depends what you mean by inexpensively, you can check your disk throughput if you wanna know precisely

[rugged radish]

mesh is basically the 3D model of your world

the terrain, specifically

in the context of my message

[sly grove]

No "infinite" procedural world type game would sanely use the first approach. it's a ton of wasted processing and storage for little gain.

There are lots of games that generate a relatively small procedural world at the start of a game session and use it for the rest of the game, something like Deep Rock Galactic would be an example of that

Or Factorio

But Minecraft and No Man's Sky are in a different category - since their worlds are "infinite"

[iron urchin]

I guess I was under the misconception that those games weren’t actually infinite. Just perceived infinite due to the sheer size.

[sly grove]

they aren't actually infinite but the limit is not that there is some world size that was pre-created, the limits are just the limits of the computer to represent numbers accurately.

like you have coordinates in a minecraft world on x, y and z. But the world uses floating point numbers to represent those coordinates, which have practical limits as you get too far away from the origin.

[iron urchin]

Ahh. So as the player travels farther, more world is generated, to the point that the data size can no longer be handled.

[sly grove]

something like that yeah

[frozen badger]

Minecraft is like, 5 times the size of the actual earth right

No way to store all that, really

[sly grove]

You run into both storage limits and floating point precision limits with minecraft

since the generated chunks are actually stored on disk

[frozen badger]

That's why the generator used to go haywire when you reached the higher numbers right

[sly grove]

Not sure I'm familiar with that particular phenomenon

[frozen badger]

since floating point precision falters as you reach certain number

[frozen badger]

When you got to extreme numbers it used to generate like this

[iron urchin]

Fascinating

[compact ingot]

with a 32 bit engine 16k x 16k meters is typically the limit of a game map that you can explore without any significant glitches due to float imprecision at human scale

[frozen badger]

That's.. a minecraft world is 60 million meters wide

[compact ingot]

to have a map of that size you need to have a moving origin

[frozen badger]

assuming one block is 1m

[sly grove]

yeah probably floating point precision issues

[compact ingot]

the minecraft generator failing is probably just an artefact of their perlin noise implementation that breaks down for large numbers but could be easily avoided by using a 64+ bit implementation

[frozen badger]

I wasn't aware perlin noise had a large number issue

isn't it all local

[compact ingot]

wdym local?

[frozen badger]

like, the value of an area is just dependent on the local values surrounding it

[compact ingot]

if you generate a 60mill unit map you sample that noise at x = 60mill and ofc it has float issues there, since you cant represent 60mill accurately in float

[frozen badger]

Yeah, that's what we were talking about

[compact ingot]

but you could just use a higher precision floating point type and be fine

or use multi resolution sampling

[iron urchin]

Local as in the range is connected to the neighboring vertices.

[frozen badger]

yeah, so having a four billion wide grid wouldn't make the perlin noise strange

like it's not inherent to the algorithm

[compact ingot]

well, mathemateically no, but every implementation will have this problem inherently