archived-code-advanced

[lapis shale]

thats what it looks like without the player animator

only idle

[sly grove]

Ok uyep see - assigning horizontal_move and vertical_move from Input.GetAxisRaw

Why would you delete that part?

Just leave it there

[lapis shale]

ok

[sly grove]

Add your animation stuff separately

[lapis shale]

so a new script?

[zenith ginkgo]

@compact ingot ok, now i have these

I removed GL.Clear and using WaitForEndOfFrame instead of return null

it no longer works when i press Play in editor now. it displays these straight away

[sly grove]

How did you conclude that?

1. Get your input from GetAxisRaw and store it in your variables (you have this part already)

2. Use those variables both to move the player, and as parameters for the animator

3. Please post questions like this in the appropriate channel.

[compact ingot]

#1 render texture and result texture you copy into should have the same size
#2 whats in line 81 now?

[zenith ginkgo]

I got it working mostly...

Yellz yea it works. It just takes like a second for it to render on start

Which is slightly annoying

https://mystb.in/ConfigInvestorPolls.csharp Current script

[compact ingot]

you could check if image.Apply(); is the culprit, if it is, nothing you can do about it

[zenith ginkgo]

Yea, I'll probably fade it in over a second or so

Good enuff for me

[lapis shale]

yeah a new script

it works now

using System.Collections;
using System.Collections.Generic;
using UnityEngine;

public class Player_Animation : MonoBehaviour
{

    public Animator player_animator;
    // Start is called before the first frame update
    void Start()
    {
        player_animator = GetComponent<Animator>();
    }

    // Update is called once per frame
    void Update()
    {
        player_animator.SetFloat("vertical", Input.GetAxisRaw("Vertical"));
        player_animator.SetFloat("horizontal", Input.GetAxisRaw("Horizontal"));
    }
}```

[compact ingot]

@zenith ginkgo you could also try using RenderTexture.GetTemporary. supposedly that is optimized for such throwaway use cases

[zenith ginkgo]

Where abouts?

[compact ingot]

instead of new RenderTexture() do RenderTexture.GetTemporary()

[zenith ginkgo]

Gotcha. I'll uise it, but it didnt speed anything up

Since what i need is a temporary RT for literally 1 frame

Thank you 🙂

[misty glade]

I have a NetworkManager singleton that I'm currently managing within Unity (leveraging the Update() loop to poll for networking events), keeping it loaded between scenes and so on. I am thinking of moving it to a pure static singleton without monobehaviour functionality so I don't need to establish references to it throughout my codebase. Thoughts?

[sly grove]

How will you poll for events?

[misty glade]

(in fact I'm thinking of moving more and more of my codebase away from MonoBehaviours unless it's specifically graphics/sound)

Not sure yet, but thinking that I'd do it in the same way that I am on the server - waiting and synchronous events

essentially

while (true)
  mysocket.SyncronizationEvent.WaitOne(1000);
  mysocket.HandleNetworkEvent(); 

I'm not sure if that's gonna be compatible with unity, actually

[sly grove]

You'll need to run this on a background thread then, and you'lll need a way to marshall data back into the main thread.

all of which is a good idea

[misty glade]

Yeah, I mean, I'd have to basically cleave off the network manager as a c# singleton and then give it access back to unity via other methods

[sly grove]

yep

[misty glade]

(or honestly, move all of the "engine" game logic and so on just into pure C# and just keep the display logic in unity)

it's pretty close to how my codebase is organized now, but my engine objects are still monobehaviour (and probably don't really need to be)

Aight. This is gonna be a big refactor but I think.. for the best. rolls up sleeves

[tough tulip]

why not a ddol? Networking in pure c# works (threading/tpl) but it usually gives more overhead when you need to synchronize the data with main thread which you would need most likely at some point (positional updates for ex)

[sly grove]

Well some code has to get the data into the main thread. Whether that's Sharping's code or some library code.

That overhead can't be avoided if you want to use that data to render something in Unity's main loop

[tough tulip]

Library code which uses networking cannot be avoided because most of them do use threading to keep the logics intact. Polling networking events using a new thread might not be always a good solution.
I have projects which have threaded network polling, and also some which have polling in main thread. I use polling in a bg thread only when I'm very certain that majority of my data doesn't need to be synchronized with Unity's gameobjects (voip)

[misty glade]

My game has limited updates that need any sync or client side prediction reconciliation (turn based strategy). I'm not opposed to a ddol approach but am increasing amount of code is unrelated to unity and I might even need to reuse it in non-unity clients (was toying with the idea of load testing or computer player agents without rendering needs)

Currently there's no logical difference in my game loop and a worker thread doing networking.. most of these managers have all the unity game loop methods commented out entirely (well, not even there, but you know what I mean)

I'm just feeling like the MB overhead might not be a cost I need to pay.. even the singleton framework approach requires some effort (ie, findobjectoftype and some thought on load order)

[small badge]

I would expect domain reload to be killing them automatically, unless you have it disabled?
The proper architectural answer is "all your asyncs should be using cancellation tokens that shut them down when the objects they depend on are destroyed", but I understand that that can be tough to add in after the fact.

[fathom flicker]

Is it possible to play a blend tree from code or make my own via code?

[hushed fable]

Should work like other states. You can't edit the state machine structure at runtime.

[fathom flicker]

If I do characterAnimator.Play("Move",0)
where "Move" is the name of the Blend Tree it will give an error saying that it could not be found during runtime.

[hushed fable]

Do other states work?

Does your case necessitate use of Play over parameters and transitions?

[fathom flicker]

Other states do work. I would prefer not to have to deal with making that web of animations.

[kindred tusk]

You can do a trigger transition from "any" which is pretty similar.

[fathom flicker]

That still ends up using parameters that are just checking the state no? I have a state machine for my character and I want that to handle playing animations. But just struggling with getting a blended animation for movement

[kindred tusk]

I don't know what you mean. It's just a normal trigger transition.

You'll need a parameter as the input for the blend tree anyway

[compact ether]

Could I get some help on figuring out how to approach this camera script to take both controller input, and mouse input? I'm sort of on a time crunch, and have to adapt what I have.
Horizontal2 and Vertical2 are both added inputs in the original Unity input manager. The Mouse X and Mouse Y are still there and intact.
My controller input from the joystick is working beautifully, so I'm trying to keep that as intact as possible.

    void Update()
    {
        //savedVec = Smash.savedVector;
        currentX += Input.GetAxis("Horizontal2") * sensX;
        currentY += Input.GetAxis("Vertical2") * -(sensY);
    }

    private void LateUpdate()
    {
        if (lookAt != null)
        {
            Vector3 dir = new Vector3(0, 0, -distance);
            Quaternion rotation = Quaternion.Euler(currentY + 12, currentX + initialOffset, 0);
            transform.position = lookAt.position + rotation * dir;
            transform.LookAt(lookAt);
        }
    }```

[untold moth]

You'd just add the mouse input to the current X/Y. If it doesn't move it's gonna be 0, so it won't contribute to the rotation. Otherwise both inputs contribute to the rotation.

[compact ether]

I genuinely thought it would break something, but by golly that did the trick, and there's no odd behavior. Thank you

[compact ether]

Although, I do have a follow-up question.
A user is claiming to be getting random camera input while playing, but I genuinely have no idea what could be causing it.
I have tested the game on multiple devices, and both PS4 and Xbox controller inputs.
I can't seem to run into any issue. This is the code for the camera controller input above.
The Horizontal2 and Vertical2 inputs take in the Joystick Axis assigned to the right stick. There's no alt buttons

Also, I just did a debug, and the only input detected is my current Xbox controller, so this has to somehow be code related.

[livid kraken]

Maybe his controller is broken and is submitting random inputs

[compact ether]

Welp, sadly my first review for my first self published Steam game is negative, because of a possible broken controller haha.
If anyone comes up with anything please let me know.

[livid kraken]

The only way to know is to ask the use for more info

[untold moth]

If you're in close contact with that specific user, maybe you could give him a debug build that would output the necessary data into a log file and they'll send it to you.

[plucky laurel]

i made an implicit contructor that accepts Action as single argument, however i can not supply a method directly without casting to Action first

Binding = Method; //method group error
Binding = (Action)Method; // works

is there any way around that?

ideally i wouldnt want to cast

[fresh salmon]

Should be working normally, unless your method has a different signature than Action

ie. no return value, parameterless

[plucky laurel]

nope doesnt

i guess the overload resolution for delegates happens on compiler level

special treatment

[fresh salmon]

Compiles:

Action Binding = Method;

void Method() { }

[plucky laurel]

Binding is not an action

ok sorry

Binding binding = Method;

where Binding is a class with implicit constructor

public static implicit operator Binding<T>(Action onEvent)
{
   return new Binding<T>(onEvent);
}

[long ivy]

How is the compiler supposed to infer what T is?

[untold moth]

Looks like a recursion to me

[plucky laurel]

from the declaration

i skipped in the example

[long ivy]

the declaration of what? T can be anything given your code so far

[plucky laurel]

the declaration of the variable to which it is being assigned

[fresh salmon]

Of the class probably, might be class Binding<T>

[plucky laurel]

Binding<SomeParam> binding = ...

[fresh salmon]

Yeah it is

[plucky laurel]

i just skipped it

its not possible from what i gathered

[fresh salmon]

Yep, getting the error too

Forcefully casting to Action makes it compile

Cannot overcome that, as the root of the error is deeply integrated in the language. IL-level

[plucky laurel]

seems so

[hallow cove]

hello

I wanna get data from my desktop to an app via wifi or a usb cable

the app being a standard apk

the problem is

I wanna send data like windows, say, spotify, any game

and I don't know how to display them there

[sly grove]

Wdym by "send data like windows"

[hallow cove]

sorry uhh

I mean like

I have an oculus quest 2

and I wanna have an app on the desktop and one on the oculus

the desktop one sends data to the oculus one via a usb or wifi

or just

I need to get data onto an android app from the desktop

I don't know how else to explain it-

like, username, avatar etc.

apps open

and so on

[sly grove]

Could serve up the data from a RESTful API running on your desktop over HTTP

USB would be different, lower level

[tough tulip]

since there are only two devices, a socket would be your best bet. you can create local socket for a duplex transport easily and would have the fastest transfer rate compares to other options.
For USB, you need to use adb in your pc to portforward to the needed port and connect your pc to localhost at that port hosted by Android. USB still has few limitations especially in Android (need authorized adb, developer mode, etc)

[hallow cove]

alright

thanks a lot!

I have no idea how I would do this lmfao-

[hallow cove]

so, I learnt how to make my own RESTful API

and I have no idea how to actually implement it to send data to my android from my pc

@sly grove

[sly grove]

now you have your Android call endpoints on your API to get the information it needs

[hallow cove]

uhhhhhhhhhhhhhhh

yesss-

I definitely know how to do that

[sly grove]

If you've made a RESTful API (amazing you learned how to do that in like an hour), the whole point of that API is that clients can make HTTP requests to it to transfer data.

[hallow cove]

I still don't understand it fully

wait

[sly grove]

so, your Android app needs to just make HTTP requests to your API

[hallow cove]

I think I get it-

omg is it really that easy-

but uhh

to transfer data, I'm gonna need to have the app both on windows and android

so, I can use #if and #endif statements to check for that, right?

[sly grove]

which app

[hallow cove]

I'm trying to make the app send like

data from the pc to the android

[sly grove]

No I get that.

[hallow cove]

and uhh

[sly grove]

Which side is Unity though

[hallow cove]

I need it to be the same

both-

both run natively-

that's my plan

so

it would work, right?

[sly grove]

yes

[hallow cove]

ok!

thank you so much!

[sly grove]

sure, but you're going to have to take special care

if your API server is also your Unity game

then you're going to have to know how to deal with multiple threads on the server

[hallow cove]

thank you btw~!!

[hallow cove]

ah, I see

[odd ridge]

I have my project split into its own assemblies. I need to access an imported projects namespace but it does not have an assembly definition to reference. How would I access the namespace from within my own assembly?

[hallow cove]

create an assembly reference file in the folder

just like an assembly

make a reference one

and use that

@odd ridge

[sly grove]

you would have to give it an asmdef

I've done this before with assets from tjhe asset store that didn't have asmdefs - just throw an asmdef in the root of the scripts folder for the asset

[iron pagoda]

Is this the correct way to make sure no other script can edit a public variable?

public bool playerIsGrounded { get { return isGrounded; } private set { ; } }
private bool isGrounded;

[stable spear]

the private set is risky because if you accidentally do playerIsGrounded = ... inside the class, it will silently do nothing

you can just omit a set altogether

[iron pagoda]

I don't expect complications within this script alone.
Is there another approach you would recommend?

https://gdl.space/odipugurer.cs

[stable spear]

it's good practice to avoid setting traps for yourself

[iron pagoda]

I agree. I just don't understand how get-set works yet.
if I understood you correctly, I can just

public bool playerIsGrounded { get { return isGrounded; } }

[stable spear]

yes

you can also just do:

public bool playerIsGrounded => isGrounded;

[iron pagoda]

Alright, great. Thank you. I'll see if I can understand this better over time.

hm, alright

[tough tulip]

public bool playerIsGrounded { get { return isGrounded; } private set { isGrounded = value; } }
private bool isGrounded;

this is what you're looking for^

[iron pagoda]

ah, so that's the correct syntax
but how is that different from the solution above?

[stable spear]

what's the point if it makes it mutable outside the class?

oh, misread

nevermind

[iron pagoda]

The script is a ground check, and I want it to be the only script to set the player's grounded state

[tough tulip]

with this, you can call both playerIsGrounded and isGrounded anywhere in your code and both will read the same value.
additionally you can serialize isGrounded or view it in Debug window if you're interested

[iron pagoda]

Noted

Thanks a lot. This takes some time to sink in, but I kept notes in comments.

[iron pagoda]

https://www.w3schools.com/cs/cs_properties.php
Reading on in the third* example, the site states that it does exactly the same, just with much less code, and no need for a variable in-between.
Do you agree that this can be sufficient in every way?

public bool playerIsGrounded { get { return isGrounded; } private set { isGrounded = value; } }
// vs
public bool isGrounded { get; set; }

[alpine adder]

hey all im currently receiving these errors in the shader Editor and have been stuck for like 3 days. our application does use assembly references and this is my gut of the root of the error

[small badge]

public bool isGrounded { get; private set; } = false will be a property that can be publicly read but can only be set within the containing class. It will not be serialised. If those are the behaviours you're looking for, then it's a reasonable choice.

[iron pagoda]

#archived-shaders <- have you tried this channel?

[alpine adder]

will try now

[sly grove]

There's no point in having two properties in this case.

tjm's solution is good

[iron pagoda]

Thank you for clearing that out.
I just want the main script to set the value of the public field

[small badge]

I assumed they were meant to be alternatives? There's a faint // vs comment in between them.

[sly grove]

oh you're right

you can also do it with serialization like this:

[field: SerializeField]
public int TotalHealth {get; private set; }```

that will:

• Show/be settable in the inspector
• Only be settable from within the class
• Be gettable from outside the class

[small badge]

Is the name of an auto-implemented property's backing field actually guaranteed to be consistent between compilations, though?

[iron pagoda]

awesome :)

[sly grove]

it doesn't matter really. The actual field name is something like <k_BackingField>TotalHealth

[iron pagoda]

[field: SerializeField] or can it be [SerializeField] ?

[sly grove]

has to be [field: SerializeField]

properties are not serializable but you can put the attribute on the auto implemented backing field this way

Oh I see what you're getting at @small badge

for serialization?

[small badge]

Yeah, I just wasn't sure whether that name was guaranteed consistent; obviously if it changed then any data serialised under the old name would be lost. Sounds like it is, though.

[sly grove]

Pretty sure it's consistent, yeah

[iron pagoda]

good stuff
while we're at it - is there a simpler way of displaying read-only values in the inspector, without using OnDrawGUI?

[sly grove]

I've used this for a couple years now without issue

Unfortunately not as far as I know. not in base Unity anyway

things like Odin will have that

[small badge]

Unity has a [ReadOnly] attribute now, I think. Possibly only in 2020 and later?

[sly grove]

Does it? I thought it was for the job system only

[small badge]

Yeah, you're right, I was misremembering seeing that in code.

[sly grove]

Yeah I got fooled by that too 😛

[iron pagoda]

Thanks a lot for the help, guys o/
Now I know 5 ways to do the same thing :D

well, almost the same things

[odd ridge]

@hallow cove @sly grove Thank you!

[arctic robin]

Hey, 2 things:

1. For visualizing an audioloop, I want to create dynamic 3D shapes that move with the music. Is a good solution to this to create e.g 64 cubes that I animate based on the music?

Like so

[quartz stratus]

Cool! So are you planning on assigning different cubes to different frequencies, then adjusting their Y scale based on dB? That seems like it could totally work, although I don't know how to read frequency/dB in Unity without doing some research

[arctic robin]

Something like that, yeah!

[arctic robin]

Oh also, here's the code for that I made at least:

        AudioSource src = GetComponent<AudioSource>();
        float[] samples = new float[src.clip.samples * src.clip.channels];
        src.clip.GetData(samples, 0);

        Debug.Log("Sample length = " + samples.Length);

        int squares = 32;
        int sampleSpace = samples.Length / squares;

        float[] visSamples = new float[squares];

        for (int i = 0; i < visSamples.Length; ++i)
        {
            Debug.Log("Getting data from point nr: " + (i * sampleSpace));
            visSamples[i] = samples[i*sampleSpace];
        }

Anyhow, is it too straining on the PC to create that many squares? I'm also not quite sure how to time the "sample playhead" together with the audio bars.

[sly grove]

Better would be to use a custom shader, GL lines, VFX Graph or a particle effect.

[arctic robin]

Interesting, got any places to check out?

All in all, I'm working towards two things here that I'm not sure how I'll do:

1. Creating a sequencer in Unity, which I am then able to match up to a loop. E.g sequence 4 kicks, when the kick plays it will animate something.
2. Creating the specific real-time audio visualization, either with real-time 3D objects or what Praetor mentioned.

[arctic robin]

Furthermore, as I want to create the visualizations during runtime, is there a way to extract several clips into one audio spectrum data file?

[hallow cove]

Np~!!

[muted widget]

Does anyone know what a CodeGen assembly is or how to decipher this error to go about resolving it? In Unity 2020.3.22f1
edit: I found the folder its referencing, but i honestly don't know where to go from here

Assembly 'Unity.Properties.Codegen' is a CodeGen assembly and cannot be Auto Referenced (Packages/com.unity.properties/Unity.Properties.Codegen/Unity.Properties.Codegen.asmdef)

[tropic lake]

this is a nice series on audio visualisation in unity
https://youtube.com/playlist?list=PL3POsQzaCw53p2tA6AWf7_AWgplskR0Vo

[sturdy geyser]

Anyone here a matrix math whiz?

[austere jewel]

Just ask the question

[hallow cove]

I have a simple question that I asked about yesterday

I have this RESTful API I made and I still don't really understand what the GET and POST do

like, let's say the api runs on my pc, and I wanna get the name of the pc user

[untold moth]

You made the api?🤔

[hallow cove]

I made whatever this is

it's gonna run on my pc and I need it to send the username of the windows user to an android

do I need to use the post function at all?

or just

get

[devout hare]

In APIs there isn't much difference, other than semantics. In theory and roughly speaking GET is for retrieving data and POST is for sending data.

[modest lintel]

You need Post to send something to the server. You need get to get something from a server

[devout hare]

Typically you'd GET user info and POST to change/create user info

[hallow cove]

aight

thanks!

[devout hare]

If you just want to retrieve the user then you only need the GET

[hallow cove]

also

I saw someone use a /:id at the end

I would like to know what that's all about

[devout hare]

It's for specifying which user's data they want

[hallow cove]

hmm

can I just ignore that for now and come back to it later or is that bad idea

[devout hare]

Do you have more than one user?

[modest lintel]

Do you want the name on the windows profile? Or the profile or whatever passes as a user in your game?

[hallow cove]

yup

[hallow cove]

and yes, I do want the windows username

and the avatar

[devout hare]

I mean if you just want the Windows username of the person playing the game, then you don't have more than one user

[hallow cove]

but this is more to send data from one pc and stuff

[devout hare]

or you don't need to specify the ID because you don't need it

[hallow cove]

they all use port 8080

if I request something on the api

I need it to return data from a paired pc or something

this is a pain I don't know how to use javascript----

[devout hare]

I'm not sure I understand what your setup is

[hallow cove]

so

I use node js to start a socket on port 8080

and then I use GET API requests in a unity app

and then here I need it to return the username

however

I need a way to tell the unity app what pc it corresponds to

does that make sense...??

I don't know I just learn how to write REALful APIs yesterday-

[devout hare]

Is the Node server running on the same computer as the Unity app?

[hallow cove]

no

the android app needs to get data from the pc so that's why

I saw apps using a pairing method

[modest lintel]

uh, you have an android app, a node server, a unity app, and a pc..?

[hallow cove]

where it gives you a number and asks you if it's correct

[devout hare]

Ok so the Node server can't get the Windows username of the Unity app

[hallow cove]

the unity and android app are the same

[devout hare]

it gets the username of the computer where the server is running

[hallow cove]

when I say the unity app, I mean the one on android

and I need an api on the pc

so the android can GET data from it

aka, the unity app

[devout hare]

not to mention that Android phones don't run on Windows

[hallow cove]

dsfuybvsyuvdf

yesss

so rn

[iron pagoda]

I'm extremely rusty when it comes to this, but from memory of web development, I am confident that you always use GET unless Custom User Input is required, for security purposes.

[hallow cove]

I have to write the API in Javascript

and have it send data when the unity app requests it

so

I don't know how to get the name and avatar of the windows user

[hallow cove]

I do also need to have the app open apps and stuff

but I need a way to tell the unity app what pc to connect to

that's my problem atm

[devout hare]

Name and avatar of the user who is running the server?

[hallow cove]

I need the api to send the username of the user and their microsoft pfp

to the unity app

I would do it like this for the username

pretty simple

[modest lintel]

Isn't this running ON android..?

[hallow cove]

wait wdym

[devout hare]

Again, the username of the person who is using the Unity app or the computer which is running the Node server? Which one?

[hallow cove]

I have an android app made in unity, the app requests data from an api running on the windows machine using GET requests

[hallow cove]

computer

computer which is running the node server

[devout hare]

Ok, great, finally

[hallow cove]

sorry I didn't even know you were confused about that-

sdvfusydf

[devout hare]

Yeah, I only asked 5 times, sorry

[hallow cove]

I have school and 6 hours of sleep daily I'm sORRy

sdfiugyudsv

[devout hare]

Then you don't need the /:id part because you don't have multiple users, you have only one user, the one that's running the server

[hallow cove]

but what if I have multiple users on the same network

if I have two android apps getting and posting data to the api, then they both control the same system

don't they?

[devout hare]

The server is running only under one user

[hallow cove]

I mean like, if I have two pcs, and two android apps

how does one android phone know which pc to use

I need to have them kinda "pair"

[devout hare]

They would have different addresses

[hallow cove]

uhhhh

how do I do that then-

atm I have one address and it's localhost:8080/vrtop

how do I make multiple addresses and how do I get them to show in the app

[devout hare]

You need the public IP of the computer, like 11.22.33.44, and then access with 11.22.33.44:8080/vrtop

Also note that setting up a server on your own computer isn't a stable solution, for something permanent you'd need an actual server

[hallow cove]

so

the android app is actually for the oculus

and I need to access data from ones personal computer

so, how can I get a list of all the pcs on the network that are running the api

if I can get that, I can get their ip addresses and "pair" them

[devout hare]

You can't

[hallow cove]

uhhhhhhhh

so then, how could I have the user select which pc to connect two from their oculus quest

[devout hare]

You'd have to hardcode the IP addresses into the Unity app

[hallow cove]

uhhhhhhh-

I could maybe make it so the app shows them their unique ip address

and have them input it in the quest

right?

[devout hare]

I don't know how Quest works, but why would the user have to input it manually?

[hallow cove]

it's literally android

[devout hare]

Or is Quest separate from the app?

[hallow cove]

just different input

I need to have the quest, aka, a literally android device, know what pc it has to receive data from

if I have two pcs running a node js server with the same address, and I request data from that address, it's gonna return both of them at the same time

so it wouldn't work

I know a lot of apps just make you put in your ip address

and connect your phone and pc that way

usually for wireless camera feed and stuff

[devout hare]

If that works for you then sure

[hallow cove]

now, I need to somehow make the api get data from the pc, but I have no idea how to use javascript

is there any way I can do that with two unity apps instead of one unity app and one js script???

like

an api running on another unity app

I am very tired please excuse my dumbness

basically, I wanna use unity to get data from the pc, and not javascript

and apparently I can make a unity app on the pc and have it serve as the api

[devout hare]

Again, if that works for you then go for it

[slate stratus]

would it be possible for me to change the serialized CAB dirs/CAB ids of an uncompressed assetbundle? my attempt at doing so utterly failed because of it refusing to load related bundle resources

the end result looked something like this

[hallow cove]

now I realize I should've probably asked this before making an api but uhh

I need to send stuff like

the whole window from a windows pc

let's say I open spotify

I need to get the window to be displayed on the android-

so umm

I just realized that I have no idea how fast restful apis actually are

can somebody tell me how fast they are?

[devout hare]

As fast as the outgoing internet connection

but streaming content from other software is ... well, let's just say I wouldn't start building stuff like that and I've done REST APIs for 10+ years

[sage radish]

REST APIs are just HTTP requests. They shouldn't be used for streaming video.

[devout hare]

there's no problem streaming video from a Node server but getting the video in the first place and everything else around it is a whole another ballpark

[hallow cove]

so

what do I use then

I somehow need to get a windows window to show up on the android screen

like

I don't need a full screen share

I just need the apps to show up as windows like they would in windows

but on an android

either via cable or wireless

let's say

I wanna make a custom desktop skin in unity

and I wanna send that data from the pc to and android phone

I wanna have the window show up like it would on pc

[sage radish]

@hallow cove I'll be honest, if you have to ask and you can't find solutions online yourself, this is probably beyond your capabilities. Like @devout hare hinted at, this is not a simple problem with a simple solution.

[hallow cove]

I just don't know what tools to use

I can code the project

I just don't know what tools I would use

it's definitely out of my capabilities

but that's the whole purpose of this project

[sage radish]

There are at least 4 steps here. 1. Find some way to record apps on Windows. 2. Encode that into a compressed video stream. 3. Stream that either over the network or over USB. 4. Decode the video stream on the Android client.

[hallow cove]

well----------

[sage radish]

Maybe start one step at a time. Just figure out a way to display an app window in the Unity editor. Ignore the streaming part for now.

[hallow cove]

I was thinking the same, I thought maybe I could get some gpu data that renders the windows, but then I wouldn't be able to interact with them most likely

[hallow cove]

that's the biggest challenge of this whole thing

[sage radish]

I don't think so. I think there are Unity assets that do that.

[hallow cove]

This can definitely be done with importing win32 specific functions into C# that allow you to capture a window. After you've done that you could craft a workspace similiar to ConsoleWorkspace that will allow you to render the results. Hooking back up input support to control the mouse pointer will require a little work, but is doable, too, also with win32 calls. I hope that points you in the right direction.

someone asked the exact thing I need

[sage radish]

(assuming you want to also have a Unity client on Windows)

[hallow cove]

yes

[sage radish]

@hallow cove This looks like it will take you 75% there
https://github.com/hecomi/UnityRemoteDesktopDuplication

If you're okay with nvidia GPU requirement on the Windows client

It doesn't say this works on Android, but I think it streams data from one Unity client to another

[hallow cove]

dang, so this doesn't work with amd

I'll see what I can do

[sage radish]

It uses this to get the desktop feed in Unity
https://github.com/hecomi/uDesktopDuplication

[hallow cove]

if this project works out I'm planning on pushing it heavily and making it open source so idk-

[sage radish]

You can maybe find another cross-platform video encoder library

[hallow cove]

yes

thank you so much!!

[sage radish]

No license on this though...

[hallow cove]

wdym?

[sage radish]

It doesn't specify a license, which tells you how you may use it in your own projects.

[hallow cove]

I see-

[sage radish]

Oh nevermind, it's in the readme. MIT license

[hallow cove]

it does

I was just about to say-

I was confused lol-

thank you so much

[marble rune]

how to fix

[wide knoll]

Posting in the advanced channel, what a rush
I'm refactoring parts of my project to add some multithreading and I'm looking for the most efficient possible way to throw an exception if you try to call certain functions from the wrong thread (or thread type). I was thinking of turning the relevant functions into thread-local delegates, and redirect to an error function on some threads, but I was wondering if there maybe was a built-in language feature/attribute that already does that (select different code path for a given method depending on thread). The key is not having to change existing calls, and also avoiding adding conditionals inside or outside the functions as they're used in some tight loops.

[marble rune]

oh making me mad reasding this

[wooden cedar]

I wouldn’t even use Unity for this, just a lightweight Android Java app and a Windows .NET app with a web socket.

But if you really needed it in Unity, that’s fine too. Just adds a layer between the parts.

[hallow cove]

I know how to use Unity well, so, I have more to gain by using it

[wooden cedar]

I’d argue you have less to gain by using it, if you already know it well, as you won’t learn as much

But, either way, I’d use a web socket and send a compressed image

Basically on Windows desktop, use this to access the base system. This is just a transparency example, but would probably work with get pixels now.

https://answers.unity.com/questions/869378/viewing-desktop-in-scene.html?_ga=2.4811376.274184869.1637031507-1019436638.1574873645

Then submit it to Android using something like this:
https://gist.github.com/danielbierwirth/0636650b005834204cb19ef5ae6ccedb

And pack the data using GZipStream and the png/jpg in a byte array

[marble rune]

no u dony cause u r alt account

[wooden cedar]

You are talking to yourself mate

[marble rune]

yes u r talking u r self

my dearest frnd

life saving vegetaion tool

[humble leaf]

Who are you talking to. Nobody here is having a conversation with you. 🤔

[wide knoll]

maybe it's literally an advanced code

[wooden cedar]

I’m not so sure in the advanced part

[heavy pebble]

Does anyone know how to convert a unity URP shader to the new 2D renderer? Google couldn't help me so far. It works perfectly in URP forward renderer but once I switch to 2D renderer it breaks. here is a small video showcasing the problem.

and the shader looks like this

[severe grove]

I'm building a reference database for all possible effects for abilities in my game. I am tying them to an enum so it is easy to select from a field in a scriptable object. How would you guys go about cleaning this up or making it more manageable?

[thorny goblet]

I think it's much tidier to use normal methods rather than anonymous ones. And this may be a very rare case where i'd use a delegate instead of an action to define the method signature, just to tidy up that declaration

[severe grove]

Alrighty. Thanks.

[thorny goblet]

no worries! Example

using System.Collections.Generic;

public enum GameEffect
{
    healValue,
    healPercent,
    addAbilityPointsTrait
}

public class Character { }

public static class DB_Effect
{
    public delegate void GameEffectMethod(float value, Character source, Character target, bool isRemoveEffect);

    public static Dictionary<GameEffect, GameEffectMethod> EffectMethods = new Dictionary<GameEffect, GameEffectMethod>{
        { GameEffect.healValue, HealMethod },
        { GameEffect.healPercent, HealPercent },
        { GameEffect.addAbilityPointsTrait, AddAbilityPointsTrait }
    };

    public static void HealMethod(float value, Character source, Character target, bool isRemoveEffect)
    {
    }

    public static void HealPercent(float value, Character source, Character target, bool isRemoveEffect)
    {
    }

    public static void AddAbilityPointsTrait(float value, Character source, Character target, bool isRemoveEffect)
    {
    }
}

[severe grove]

That does look much better. Thank you. 🙂

[muted widget]

I have an assembly that has "autoReference" set as true by default, but causes an error stating that because it is CodeGen assembly, cannot be written as "true". So i changed it to false and get this error instead now

(0,0): error System.IO.FileNotFoundException: 

{
    "name": "Unity.Properties.Codegen",
    "references": [
        "Unity.Properties"
    ],
    "optionalUnityReferences": [],
    "includePlatforms": [
        "Editor"
    ],
    "excludePlatforms": [],
    "allowUnsafeCode": false,
    "overrideReferences": false,
    "precompiledReferences": [],
    "autoReferenced": false,
    "defineConstraints": []
}

It looks like now it cant be found if it's set to false?

Removing the assembly entirely causes a wall of errors, is there a way to rewrite this assembly to avoid either option?

To test options I commented out "autoReference" line, but then the original error returns

Assembly 'Unity.Properties.Codegen' is a CodeGen assembly and cannot be Auto Referenced (Packages/com.unity.properties/Unity.Properties.Codegen/Unity.Properties.Codegen.asmdef)

If it's known it cant be autoReferenced, I'm not sure why it was written this way. Or because maybe in Unity 2018.3 it could be but not in 2020.3? Any help is super appreciated

[marble rune]

bruh why u pingt me! u pinged the all mighty busy man

[tropic lake]

you are off your rockers

[sage radish]

You have to reference it in your assemblies if you want to access it. You'll have to add an assembly definition to your code if you don't already have one.

[muted widget]

I'm not sure how to reference it in the assembly, i mean i see it in inspector and can compare it to how its written in visual studio, otherwise im not sure how to do that

[muted widget]

I was messing around with it to see visual changes but i noticed when i make a change it will apply them, but then when i go back to make another edit it will automatically revert to its default settings

[sage radish]

That's probably because you can't modify installed packages.

Have you checked if there's an update for this particular package in the package manager?

[muted widget]

Not since 2018

[muted widget]

This is the manifest for the folder of Unity.Properties.CodeGen, no updates for this since the date of 2018 written there

[sage radish]

Since it's been so long since it's been updated and it's still in preview, it leads me to believe it simply isn't supported anymore. What Unity version are you using with it?

[muted widget]

Originally 2018.3.8, now its in 2020.3.22

[sage radish]

Was it working with 2018.3.8 and stopped working after the upgrade?

[muted widget]

yupp, i mean i know its a drastic upgrade, but the other wall of errors was fixable, except this final one

[sage radish]

Well, whatever the fix is, you're going to have to find it yourself because it looks like Unity has stopped supporting this package.

CodeGen stuff in Unity packages have changed a bit in other packages since 2018, so however this package is doing it might not be supported by Unity 2020.

Even if you fix the reference and compile time errors, there's no guarantee it will work the same as before

[muted widget]

Would you be able to tell me what you know about CodeGen in particular?

[sage radish]

@muted widget Wait, I found a newer version of the package here. Last update 2020.
https://docs.unity3d.com/Packages/com.unity.properties@1.1/manual/index.html

Do you not see this version available in the Package Manager?

It won't always show as if an update is available, you have to expand the list of versions and pick the latest one yourself

[muted widget]

When i click on "show in package manager" it opens package manager but doesnt it show it to me

[muted widget]

It opens and just highlights the first thing on the list

(everything is as it was from 2018, even the ancient version of HDRP because thats even on version 10 now)

[sage radish]

You have to install it manually, confusingly through the "Add package from git URL..." option

and put in the full name of the package as the URL: com.unity.properties

[muted widget]

That was the next mission, ive seen if its not available, github is always the next option

[sage radish]

This isn't actually installing it from Git, this is just how you manually install hidden Unity packages in 2020.3

In 2021, they've separated it into its own menu item so it's less hidden and confusing

[muted widget]

Ohhh alright

Ah okay, that haha okay yes that did fix it, it triggered the install and then another wall of errors popped up but it was good because that was then fixed. I realize we get sent back to the normal errors we've gotten before with moving from 2018 to 2020 with references to Boo language which was deprecated in 2017

I see though, before when we've made attempts to make this leap into the newer engine for some reason we never came across the CodeGen/Unity.Properties error (would that be because of it being so old or in preview that its not guaranteed to always be problematic?), so far Boo not being in 2020 is the only issue with reference to TestUtil.cs and ABModelTests.cs

[jaunty kite]

Greetings to anyone down to read this! I really didn't want to do this, but this is already my last resort. The line in question was using UnityEngine.InputSystem; and the error is:
The type or namespace name 'InputSystem' does not exist in the namespace 'UnityEngine' (are you missing an assembly reference?) [Assembly-CSharp]
I'm currently on Unity 2021.2.2f1, and this issue just suddenly popped out into existence. And please know that i know this. i know what i should do to get it fixed. But here's the thing, as you can see on attached image:

• I have Visual Studio Code Editor package installed in my Unity project,
• I also have Input System package installed,
• Configured in Preferences to generate .csproj files based on packages I have, and
• OmniSharp found it, successfully loaded it, and added it.

Omnisharp still fails to find UnityEngine.UI and UnityEngine.InputSystem.
No errors in Unity console. I did the following stuff which me and my friends thought would fix the problem:

• Did stuff mentioned in https://code.visualstudio.com/docs/other/unity
• Downgraded, and Updated Visual Studio Code Editor package latest, 1.2.4
• Manually deleted all .csproj files and regenerated them
• Right click > Reimport All (because this deletes and recreates cache data)
• Manually deleted "library" folder
• Attempted to update VSCode and the C# Omnisharp extension (both up to date anyway)
• Installed Mono (which actually fixed my Intellisense/Omnisharp issue before in 2020.1b)

I also made a new, clean project with the same stuff installed, and the issue persists. And apparently, It's not only InputSystem that's affected. UnityEngine.UI, UnityEngine.TMPro, and UnityEngine.Rendering as well.

Any clue on what I should do now? Feel free to ping me anytime if you're down to help...

[small badge]

In my experience VS Code integration 1.2.4 doesn't work; references to code in packages aren't recognised, which sounds like exactly the problem you're having. The only thing that fixed it for me was downgrading to 1.2.3.

[sage radish]

(I read the rest of the message)

[small badge]

(By manually editing manifest.json, since PackageManager doesn't offer an option to downgrade.)

[jaunty kite]

Is that method any different from this one though

oh shit i pinged, mb

[small badge]

(And obviously then you'll need to regenerate solutions, yada yada yada.)

I haven't actually tried that option yet; if it lets you specify a version number it should work fine.

[jaunty kite]

It does; I've done this before where i even downgraded multiple times down to latest 1.1.x and 1.1.1 which, i have to mention: worked flawlessly on VSCode in Unity 2020.x

but here in 2021.2.2, even downgrading doesn't fix it :(

I was thinking, is there anything within Unity or Unity packages that needs Visual Studio Build Tools 2017? I may have updated to BT 2019 last week, but the thing is that this problem already existed 2 and a half weeks ago

[small badge]

Then I don't think I can help, sadly; even going over the changes from 1.2.3 to 1.2.4 with a fine-toothed comb I wasn't able to conclusively figure out what was causing the problem I had, so I can't really say how it might be affecting you. All I can offer is a data point that I got it working with Unity 2021.1, VS Code integration 1.2.3 and VS Code 1.62.2.

[jaunty kite]

The only difference we have in versions would be Unity itself... I'll just downgrade to 2021.1 for the meantime

[small badge]

I feel like I'd be surprised if it was a build tools issue? Failing to recognise dependencies seems like it would always be a csproj/solution generation issue, and I don't think anything in that process is reliant on the build tools. But I'm not an expert on the VS code solution pipeline so I can't say for sure.

[jaunty kite]

I was also hesitant with it being a problem with VS/VS BT since I also tried using VS 2019 as my external editor and it works flawlessly, with all that references and omnisharp and intellisense

but having chrome, vs2019, and unity at the same time feels too clunky for me so i wanted to fix vscode

either way, thanks @small badge! really appreciate your help :))

also you @sage radish 💕

[small badge]

In case it helps, I believe this bug report (and associated pull request) was the problem I was having with 1.2.4: https://github.com/Unity-Technologies/com.unity.ide.vscode/issues/4

Just in case it suggests other things to look at; for example, since the issue seems to be path related maybe check whether either the unity project or the package cache are located in a path containing uncommon characters.

[jaunty kite]

Oooh, that might just be the issue i'm having? I have assets from Asset Store, like Shapes by Freya and it's not throwing any kind of issue and Omnisharp acknowledges the package and it's scripts/references

Stuff's inside Library\PackageCache\com.unity.inputsystem@1.1.1\... while the .csproj files lead to Packages\com.unity.inputsystem\... which is completely different, and that path doesnt exist because there's nothing in Packages folder other than two json files

EDIT: 1.2.3 regenerates the .csproj files with the right path, but didn't replace the .csproj files; Downgrade works but I also had to delete all of the .csproj files from v1.2.4 so that v1.2.3 can generate freely

That was an oversight from me, my bad; Thanks again!

[regal olive]

Does PlayerPrefs immediately write to disk or does it store everything in some location in the program memory and write to disk when the application quits?

Perhaps you want to reset the scene's contents. Reloading in an additive context would mean to first unload it and then load it.

If it's a reset you could write custom logic for that.

[jaunty kite]

They go to the Registry.
Computer\HKEY_CURRENT_USER\Software\[CompanyName]\ProductName

https://docs.unity3d.com/ScriptReference/PlayerPrefs.html

Oh i didn't see the last part

The application saves quite a lot upon quitting, such as last resolution used, etc.

Here's some of them:

[regal olive]

Well it's not like Unity reveals how it implements PlayerPrefs since it's not written in C#

[regal olive]

These look like EditorPrefs entries

[compact ingot]

if you want control, write your own.

[regal olive]

welp that's the thing really

it'll probably not be of much use unless I am making a much bigger game

[compact ingot]

i'd not expect anything in unity that deals with the runtime platform (i.e. the OS) to be immediate

it almost certainly wont be immediate

[regal olive]

As a matter of opinion, me neither. It is much more sensible to hold stuff in memory, then write it to disk when the app exits normally or terminates (unless it's small enough or maybe crucial to store immediately?).

[compact ingot]

but whether it is deferred until application exit... you can just test that

[compact ingot]

depends on what it is, that doesn't hold as general principle...

[regal olive]

It'd be awesome if it happens interstitially

[compact ingot]

at that point you're probably better off controlling exactly when it happens and create a custom save system

[regal olive]

True, true.

[plush lion]

Hey, I have a math question

So here's my problem: I have a camera, and I have a formula that takes in position between {-0.5,-0.5} and {0.5, 0.5} and return a direction from the camera to a point on a picture that's 1 unit of distance away from the camera

{-0.5,-0.5} corresponding to the bottom left corner of the picture

And {0.5,0.5} the top right

The thing is I can't use camera.main.screentoworld, it just doesn't do what I want and is too dependent on the screen size

I need something that would behave the exact same on all computers

For my formulas, I'm using x * Mathf.Tan(fov.x / 2 * Mathf.Deg2Rad) * 2, y * Mathf.Tan(fov.x / 2 * Mathf.Deg2Rad) * 2
Fov is a vector with both the horizontal and vertical field of view of the camera that took the picture.

I'm not getting the exact expected size, though. Are my formulas wrong?

[plush lion]

Let me put it in a different way. How would you get the size of the base of an isosceles triangle, given the opposite angle and the height of the triangle?

Is it indeed tan (angle / 2) * height * 2 ?

[glass wagon]

dunno about the maths, but have experienced much irritation with the fov being odd in Unity.

[plush lion]

Yeah, but it looks like the fov I'm given is correct

I'm not given the fov of the camera of a scene, I'm given a photo and the fov of the camera that took the photo. From a different project.

[glass wagon]

is the fov value coming from one of Unity's cameras?

[glass wagon]

if you think you've got the maths right (I have no idea about this), then start figuring out ways to examine the accuracy of fov. It is weird, in Unity, for me, someone that thinks in terms of physical cameras. And is made much weirder by interactions with the options of aspect ratios you can and can't control (as these can impact the pixel aspect ratio, too), and the sensor size/shape simulations, and the lens simulations. Don't trust the fov values, first and foremost.

[plush lion]

So I just checked the code, the fov I get is not a camera.fov, it's computed in a completely independent way.

But it is originating from an unity camera

[glass wagon]

Ok, you can go hunt down how they presume to calculate FOV, and then figure out how that pertains to each and every device variation and aspect ratio you target, or... try thinking differently about how to achieve your desired insight by using ray casting. I'd go with ray casting!!!

[plush lion]

All of this is done to get a direction vector to give to a ray cast... ^^"

[glass wagon]

I understand, really, I do. Think the other way around, world to view origin, etc, then back out again. Sorry... the fov thingy in Unity has driven me mad, before, so I somewhat know what you're going through.

[plush lion]

Hahaha alright. I'm reviewing the code that compute the fov to see if I get mistakes from there (between us the dude that wrote that code has written many scripts of... questionable quality)

[glass wagon]

best of luck! Also, as a general rule of thumb, if there's a weird, counter-intuitive, non industry standard manner in which Unity could do something, and then further ways they could make it odd - that's the way they've chosen to do it. This is so consistently the case, across so much of Unity, that it's best to presume this is the source of problems whenever something misbehaves, to save yourself time and sanity querying your own code and methodologies.

[plush lion]

Alright, I'll take your word for it

Thanks for the help

[split kindle]

Hello, I got a non urgent question. I recently figured out I could use a combination of the [RequireComponent()] attribute and a [SerializeField,HideInInspector] decorated field of the same type that I assign in OnValidate() to get serialized references to other behaviour on a gameObject.
I like it for several reasons, it keeps the inspector clean, prevents designer from forgetting to set or missusing the field and garantees that the field is populated when an Awake() or Start() methods is called, regardless of execution order.
What am I missing here ? It seems like a fairly obvious pattern in retrospect but I've never seen it used. Is there a edge case I'm missing where these reference will be null ? (Excluding adding the component at runtime using AddComponent)

[sly grove]

Another loophole besides AddComponent is an instance of the component that was created in a scene or prefab before those attributes were added. If the prefab or scene is not opened again afterwards, the component may not be present and the serialized field may not be populated

But it seems pretty solid

It does require adding 3 attributes across a field and the class itself, as well as writing an OnValidate() method.

So that complexity may explain why you haven't seen it

[split kindle]

The interaction with prefabs does sound dangerous, so I'll hold off peppering production code with it, thanks for pointing it out. I don't mind the complexity too much, as it's much simpler than even the most basic custom inspector.

[lapis gale]

Hello. Can someone help me with one thing?
I'm making a game in the style of Timberman.
I created one script. In this script in the update method, I would like it to change the position in one of the axes by 200 units within 0.2s with each click. (I click, then it change position and I can click again)
It is an object that can move endlessly and moves in relation to itself.

[split kindle]

I'd use a Coroutine for that, lerping the object position over 0.2s (you can use "yield return null" to halt execution of the coroutine until next frame). Remember to have some kind of check to prevent that Coroutine to be started before the previous one is finished.

[lapis gale]

Okey I'll try that thx

[cedar ledge]

how to use hdrp for 2d platformer game?! help!!!

[lapis gale]

how to use post processing for 2D platformer game?! help!!!

[misty glade]

This is... not a good question

[cedar ledge]

how to ask good question?! help!!!

[misty glade]

http://catb.org/~esr/faqs/smart-questions.html

[cedar ledge]

how to open link on discord on windows 12

[lapis gale]

12!?

[misty glade]

apparently I missed the humor bus this morning .. my fault, carry on

[cedar ledge]

sorry i was bored while installing unity 2021.2 and vs 2022

[misty glade]

🙂

yah i fell for it hook line and sinker lol 🐟

[lapis gale]

are there any differences between vs 2019 and 2022?

[cedar ledge]

no they just wanted a bigger number

[lapis gale]

xD

[cedar ledge]

the vr template has oculus sdk and xr. is that just for multiplatform compatibility or is there a reason to use both?

https://docs.microsoft.com/en-us/visualstudio/ide/whats-new-visual-studio-2022?view=vs-2022

[lapis gale]

omg... microsoft is bored...

[cedar ledge]

64 bit, perf improvements, hot reload, intellicode, icons,

a couple small things here and there, really

biggest news is dotnet 6 and c# 10

[frozen imp]

#497872469911404564 if you want to discuss unrelated stuff.

[cedar ledge]

my b

[misty glade]

So I'm starting a new side project, an Idle game, and before I roll my own "big number" class (essentially a float and exponent), is there something that's already been done here? Should I just use System.Numerics.BigInteger?

(and maybe BigInteger isn't necessary since I don't need that precision)

Actually, it's definitely not the right approach since I don't want to bloat the network/storage space with this data structure that looks like it gets as large as needed, to hold the precision

[honest stirrup]

Hi! I've been working on procedural generation and I've gotten to here (see image).

How can I make more islands and make it look less like a mess.

    private void Start()
    {
        GenerateMap();
    }

    private void LateUpdate()
    {
        if (Input.GetKeyDown(KeyCode.Space))
            GenerateMap();
    }

    private void GenerateMap()
    {
        map = new int[width, height];
        decoMap = new int[width, height];
        print(map.ToString());
        RandomFillMap();

        for (int i = 0; i < smoothTimes; i++)
        {
            SmoothMap();
        }
        UpdateTilemap();
    }

    private void RandomFillMap()
    {
        if (useRandomSeed)
        {
            seed = UnityEngine.Random.Range(1, 1000).ToString();
        }

        rand = new System.Random(seed.GetHashCode());

        for (int x = 0; x < width; x++)
        {
            for (int y = 0; y < height; y++)
            {
                if (x == 0 || x == width - 1 || y == 0 || y == height - 1)
                {
                    map[x, y] = 1;
                }
                else
                {
                    map[x, y] = rand.Next(0, 100) < randomFillPercent ? 1 : 0;
                }
            }
        }
    }

    private void SmoothMap()
    {
        for (int x = 0; x < width; x++)
        {
            for (int y = 0; y < height; y++)
            {
                int neighbourWallTiles = GetSurroundingWallCount(x, y);

                if (neighbourWallTiles > 4)
                {
                    map[x, y] = 1;
                    decoMap[x, y] = 0;
                } else if (neighbourWallTiles < 4)
                {
                    map[x, y] = 0;
                    decoMap[x, y] = rand.Next(0, 100) < decoFillPercent ? 1 : 0;
                }
            }
        }
    }

    private int GetSurroundingWallCount(int gridX, int gridY)
    {
        int wallCount = 0;
        for (int neighbourX = gridX - 1; neighbourX <= gridX + 1; neighbourX++)
        {
            for (int neighbourY = gridY - 1; neighbourY <= gridY + 1; neighbourY++)
            {
                if (neighbourX >= 0 && neighbourX < width && neighbourY >= 0 && neighbourY < height)
                {
                    wallCount += map[neighbourX, neighbourY];
                } 
                else
                {
                    wallCount++;
                }
            }
        }

        return wallCount;
    }

    private void UpdateTilemap()
    {
        for (int x = 0; x < width; x++)
        {
            for (int y = 0; y < height; y++)
            {
                if (x == 0 || x == width - 1 || y == 0 || y == height - 1)
                {
                    waterTileMap.SetTile(new Vector3Int(x, y, 0), waterTile);
                }
                if (map[x,y] == 0)
                {
                    mainTileMap.SetTile(new Vector3Int(x, y, 0), grassTile);
                } else
                {
                    waterTileMap.SetTile(new Vector3Int(x, y, 0), waterTile);
                }

                if (decoMap[x, y] == 0)
                {
                    decoTileMap.SetTile(new Vector3Int(x, y, 0), null);
                }
                else
                {
                    decoTileMap.SetTile(new Vector3Int(x, y, 0), waterTile);
                }
            }
        }
    }

[devout hare]

The "bloat" is absolutely insignificant

[misty glade]

Google Perlin Noise

[misty glade]

Er, not at all?

[honest stirrup]

I'm using Cellular Automata

I just need to make it more islandly and not like a swamp

[devout hare]

How large numbers are you planning to use?

[misty glade]

The game is networked so... the clients/servers are going to be sending these values back and forth quite often

Arbitrarily large

1e900

(nonsensically large)

I don't want to send 900 bytes where 10 will do 🙂

[devout hare]

well 1e900 is obviously not 900 bytes but that's still definitely insignificant

[misty glade]

My thought right now was to roll a BigNumber class that overrides common operators like operator+, operator- and operator* (probably don't need divide in an idle game)

So again.. the game is networked so the cost on the server of serializing/deserializing strings of 900 digits to a big number and operating on it, serializing it again into a string and then serializing that string into a byte array and encrypting it and sending it to the client for operations every few seconds for dozens/hundreds of players is .. significant enough

I mean, it's just not the right solution 🙂

[modest lintel]

why not send binary

[misty glade]

i do send binary

or did you mean sending the data of the bigint as a byte array - i could do that, that looks smaller

but again, this is a lot of work to implement something that i don't need - I don't need the precision on the value

Also, this:

Although this example appears to modify the value of the existing object, this is not the case. BigInteger objects are immutable, which means that internally, the common language runtime actually creates a new BigInteger object and assigns it a value one greater than its previous value. This new object is then returned to the caller.

Note

The other numeric types in .NET are also immutable. However, because the BigInteger type has no upper or lower bounds, its values can grow extremely large and have a measurable impact on performance.

If I'm performing operations on these on the client in every Update() .. it's going to be a resource hog

well, maybe not, i suppose, since it's just 1 or 2 instances of this

[cedar ledge]

perlin noise

[honest stirrup]

Again, I'm using Cellular Automata

[drifting galleon]

perlin noise

[honest stirrup]

:/

[misty glade]

CA meaning what, you want to take the starting setup an use game-of-life style advancement? ie - some set of rules that determines what happens to each cell?

I mean, fundamentally what you're trying to do (and maybe what you're not understanding me/tech/mind say) is create coherent randomness

People who are pretty smart (ie, Ken Perlin) have done this and it's good, the only part you need to do is implement it, and there's hundreds+ of articles on that

essentially your RandomFillMap() should be entirely replaced with a perlin noise generator

[drifting galleon]

+1

[misty glade]

(and specifically, what you need to do is play with the output coefficient - ie, you're going to get a value from 0.0-1.0 for each tile, and you just determine if that value is a waterTile or grassTile in your setup)

your approach right now is to generate random noise and then smooth it some number of times

This is going to get you a very consistent and bland noise map

http://libnoise.sourceforge.net/glossary/index.html#coherentnoise

Smoothing "incoherent noise" means just a more boring/flat "incoherent noise"

So again... TLDR: scrap RandomFillMap() and implement perlin noise (and you won't even need SmoothMap after that)

it'll also be a lot faster since this is a hack:

        for (int i = 0; i < smoothTimes; i++)
        {
            SmoothMap();
        }

[inland cairn]

Is there any way to loop through hunderts of ints without lagging pc too much?

[misty glade]

looping through trillions of ints shouldn't lag your pc

[inland cairn]

And comparing them

[misty glade]

post your code - maybe you're doing something you aren't thinking is the problem?

also no problem

[inland cairn]

Better question is how can i cut mesh?

[misty glade]

    public static void SpeedTest()
    {
        int a = 1;
        int b = 2;
        DateTime startDT = DateTime.Now;
        TimeSpan ts = DateTime.Now.Subtract(startDT);
        for (int i = 0; i < 100_000_000; i++)
        {
            _ = (a < b);
            if (i % 100_000 == 0)
                Log.Debug("{i} iterations have taken {time}", i, DateTime.Now.Subtract(startDT));
        }
    }```

100m iterations (along with a datetime subtract and modulus operator - typically slow) in 1 sec

"cut mesh"?

[inland cairn]

yeah like you have cube and plane and you will slice the cube with the plane

[misty glade]

I don't have experience with that but I do know that it's a ... common? complex? problem, googling should yield a lot of info on it

I seem to remember someone posted an article in the last few weeks here about novel solutions to it

[inland cairn]

ok ill try it

[misty glade]

https://stackoverflow.com/questions/9709970/algorithm-or-software-for-slicing-a-mesh

[glass wagon]

is there a way to create a thread that's sure to not be on the same core as the Unity Main thread, or the Unity Audio thread?

[sly grove]

You generally can't control which core your threads run on. The CPU will schedule them as best it can

ALthough it does look like there is some kind of support for this maybe in C#? https://docs.microsoft.com/en-us/dotnet/api/system.diagnostics.processthread.processoraffinity?view=net-5.0#System_Diagnostics_ProcessThread_ProcessorAffinity

[thorny onyx]

I have a prefab which contains 2 MonoBehaviour scripts (script A and script B). The scriptA contains a reference to script B. When I instantiate the prefab, the Start method of the script B is called AFTER the start method of the script A. It's a problem because the script A calls script B method from its start method. Is it normal and how to solve this problem ? (I cannot use the Awake method)

[glass wagon]

do you know if each thread/core has its own registers?

[sly grove]

they do, yes of course

A CPU could hardly do its job if it had to share registers

although with HyperThreading who knows

I think from an application perspective, you can assume yes

[stable spear]

There is a way to set Awake/Start order, but it is heavy-handed. It would be better if you design your code so that the ordering doesn't matter.

[thorny onyx]

I can do that but I have to put "Initialize" method in the script and don't use start/awake methods... not very user friendly...

anyway, thanks for your help @stable spear 🙂

[stable spear]

Generally there's cleaner ways to do initialization than have custom initialize functions, depending on what you're doing. But without seeing your code, I can't offer any specific recommendations.

[sly grove]

Separate your initialization code into Awake and Start.

Do "self init" stuff in Awake, and code that reaches out to other objects in Start

why can't you use Awake? (just noticed that part)

There's also the OnEnable() method which will run before Start() as well

[thorny onyx]

I can't use Awake because I have to separate my initialization code

as you just explained

the code in start must be in start because it's not self init

[stable spear]

But what exactly are you initializing between scripts?

Much of the time, dependencies between scripts can be untangled and removed.

[sly grove]

don't understand how this precludes the use of Awake. Sounds like you're doing some self init in Start that should be in Awake()

[thorny onyx]

I have simplified the use case but there are others scripts that are used in script B start function

[queen plover]

<@&502884371011731486>

in multiple channels

[humble leaf]

!ban 473126488728666112 scam

[thorn flint]

ddwi16#9308 was banned

[misty glade]

I'm writing a BigNumber struct with some custom functionality, and overloading a bunch of operators. My operator + is a bit of a disaster, I think. I'm not sure how I would go about profiling this, though (in terms of memory use).. Any suggestions on starting points or google terms?

        public static BigDecimal operator +(BigDecimal left, BigDecimal right)
        {
            if (!left.IsValid) throw new ArgumentOutOfRangeException("left", left.GetInvalidReason());
            if (!right.IsValid) throw new ArgumentOutOfRangeException("right", right.GetInvalidReason());

            if (left.Exponent == right.Exponent)
                return new BigDecimal(left.Mantissa + right.Mantissa, left.Exponent);
         
            //Treat left as the smaller item, swap if necessary.
            if (left.Exponent > right.Exponent)
            {
                (left, right) = (right, left); 
            }

            while (left.Exponent < right.Exponent)
            {
                left.Mantissa /= 1000;
                left.Exponent += 3;
            }
            decimal newMantissa = right.Mantissa + left.Mantissa;
            BigDecimal ret = new BigDecimal(newMantissa, right.Exponent);
            ret.Adjust();
            return ret; 
        }```

(this implementation of "big decimal" is geared towards forcing it to "multiples of 3" exponents.. so valid numbers are from 0-999 for the mantissa and 0/3/6/9... for the exponent)

[stable spear]

you can use right.Exponent - left.Exponent to compute exactly how many times you need to shift left down. The fastest way to do it would be to operate on the 128-bit internal representation of decimal. But I'm not sure you want to get into that mess.

[misty glade]

I tried setting a breakpoint before/after 100 million operations and snapshotting memory before/after, and it looked clean

Yeah, I definitely don't want to be touching internal representation and bitshifting unless that's the only way to optimize

[stable spear]

there's no memory allocation, its just inefficient math-wise

well, there is memory allocation at the end with new BigDecimal

[misty glade]

I'm just not comfortable with my skills currently to do that

[sly grove]

only on the stack

[misty glade]

I also have an alloc on the swap tuple operation, I think

[stable spear]

is BigDecimal a struct?

[misty glade]

yeah

[sly grove]

Nothing in this code does any heap allocation except the exception raising

[stable spear]

oh ok. then yea, no heap allocation

[sly grove]

which is good 😛

[misty glade]

OK.

I was literally typing "is that bad"? 🙂

OK, maybe I can look at the CPU performance then.. I've actually never used the CPU profiler

Is it worth doing this now?

[sly grove]

(left, right) = (right, left); ```

[misty glade]

nor not really

[sly grove]

this is crazy to me

that works?

[misty glade]

Aye - I stole that, lemme cite the source

// https://tearth.dev/posts/performance-of-the-different-ways-to-swap-two-values/

haha great image 😐

doesn't exactly inspire confidence

[long ivy]

works great until it explodes due to some edge case I thought about but didn't address -- that just means it'll fit in with the rest of my code

[sly grove]

I'm blown away, would have guess the XOR thing was the fastest but I guess not

it's just a parlor trick after all

[misty glade]

(this article goes super deep on the performance implications of that operation, btw, which i mostly skipped over)

I was briefly considering the unsafe pointer swap and then decided against it

what's the xor thing?

(I don't have operator xor defined on this struct, fwiw)

operator^ is it? i don't think i've ever really used it.. i haven't used any sort of bitwise flag enum that I can recall, but i can't think of a common usecase for xor

[stable spear]

X86 has an XCHG instruction for exchanging values. Not sure if the compiler can make use of that, but it is likely to be faster than a bunch of XOR.

[misty glade]

interesting

what are these numbers on the left? 844, 988, 21

number of ops?

[stable spear]

% of samples taken at that point

I'm assuming. The fact that there are so many means that line must be taking a long time

what is IsValid ?

a property?

[misty glade]

yeah - pretty light, actually:

        private bool IsValid
        {
            get
            {
                return (IsValidMantissa && IsValidExponent);
            }
        }
        private bool IsValidMantissa
        {
            get
            {
                if (Mantissa < 0)
                    return false;
                return true;
            }
        }

        private bool IsValidExponent
        {
            get
            {
                if (Exponent < 0)
                    return false;
                if (Exponent % 3 != 0)
                    return false;
                return true;
            }
        }

[stable spear]

lol, that's light??

[misty glade]

actually IsValidExponent has the modulus in there

i forgot about that

hm... I could theoretically get rid of it since the struct itself can't be constructed invalid-ly

(i hope)

[stable spear]

Do you treat the fields Mantissa / Exponent as immutable after construction?

[modest lintel]

I wonder if the %3 would be faster or doing the old trick of taking the sum of digits and checking if that is divisible by 3

[misty glade]

sort of - public get, private set, and they're only set via the operators

[stable spear]

there's no "sum of digits" instruction that could possibly be faster

[sly grove]

Can probably slightly improve performance by removing branching too, e.g:

        private bool IsValidMantissa
        {
            get
            {
                return Mantissa >= 0;
            }
        }```

[misty glade]

https://pastebin.com/YGAxJ49J

[stable spear]

plus checking if that is divisible by three would still require %3

[misty glade]

here's where it's at now - critiques welcome

[modest lintel]

On a much much smaller number

[stable spear]

X86-64 operates on 32-bits or 64-bits at a time

there might be circuitry to handle 8-bit or 16-bits, not sure, but it won't be significantly faster

[misty glade]

how can I .. uhm... dive into the hot path? I'm having trouble reading this cpu profiling thingy.. although I suppose that's not really a code/unity question and more of a tooling thing

Like - what's this mean..?

[sage radish]

Doesn't it use an array to store all the bits, like BigInteger?

[misty glade]

isvalidmantissa appears to be taking a big chunk of time?

[stable spear]

does what?

[misty glade]

Mine doesn't - internally I just store a decimal and int

[stable spear]

Can you arrange to do the validity checks only once, upon construction?

[misty glade]

yep, although the operator constructs as a convenience for addition with ints/decimals

see lines uh... 173ish

public static BigDecimal operator +(BigDecimal left, decimal right) => left + new BigDecimal(right);

[stable spear]

that's needed anyway

right could be negative

[misty glade]

I mean, obviously I could remove that and force the caller to ctor their own BigDecimal

ie - place the slow stuff (the modulus operator, i guess?) in the ctor and then force "consumer-sharping" to ctor and manage the right operand BigDecimal

[stable spear]

btw, do you actually need performant code or are you just testing profiling?

[misty glade]

I may need it - I'm anticipating that the server is going to be doing thousands-millions of these operations per second, depending on the number of connected clients

[stable spear]

the right way to do this is probably to mimic the same representation as decimal, but with more bits. And maybe native code if needed

[misty glade]

there's workarounds but i was hoping to put some thought in now and not have it be a refactoring headache later

[stable spear]

what is it going to be used to represent?

[misty glade]

idle game soft currency

ie, "energy"

[sly grove]

If you really want to get into extreme optimization you'll need to dive into unsafe code at some point

[misty glade]

😐

[stable spear]

decimal's range isn't large enough for this?

[misty glade]

no, we'll be doing the traditional genre exponential growth

the only linear progress will be the exponent

ie - generator A makes "1e1" energy per 30 sec, generator B makes "1e4" energy per 30 sec, etc... generator Z makes "1e900" energy per 30 sec

essentially nonsensically large numbers

[stable spear]

how do you compare BigDecimal's ?

[misty glade]

operator== overload and epsilon

        public bool Equals(BigDecimal right)
        {
            if (!right.IsValid)
                throw new ArgumentOutOfRangeException();

            if (Exponent != right.Exponent) 
                return false;

            return (Math.Abs(right.Mantissa - Mantissa) < BigDecimal.Epsilon);
        }

[stable spear]

there are multiple representations since decimal has its own exponent

[misty glade]

(probably also will need to be optimized)

[stable spear]

Mantissa = 100, Exponent = 0 is the same as Mantissa = 1, Exponent = 2

[misty glade]

exponent = 2 is illegal in my struct

[stable spear]

fine, 1000, 0 and 1, 3

[misty glade]

1000 is also illegal 🙂

[stable spear]

why?

[misty glade]

there is only one combination of mantissa + exponent for every number in R

[stable spear]

it passes your checks

that's true normally, but there's two exponents here

the one internal to decimal, and the external one

[misty glade]

it passes my checks (which I should add a check for, thanks) but should be impossible - Adjust() fixes it if the number is >= 1000

(and Adjust() is called in the ctor)

        private void Adjust()
        {
            while (Mantissa >= 1000)
            {
                Mantissa /= 1000m;
                Exponent += 3;
            }
        }

[stable spear]

lol

[misty glade]

i know.. i'm not sure that this struct is on the right path 😐

it works but.. man it feels stinky

I did come across this repo: https://github.com/Bakaender/IdleParts/blob/master/BigDouble.cs but this is mostly the same as mine, with slightly different behaviour

[stable spear]

funny thing is that decimal has 22 unused bits in its internal representation

they could have easily extended the range to a huge amount

[misty glade]

OK well I just did approx 300b allocs and 100b operator+ and it took 3.7 sec so.. i'm going to optimize later if that's not good enough

really? that's weird

where's the latest decimal source? i wanna poke in there too

https://github.com/dotnet?WT.mc_id=dotnet-35129-website here somewhere?

[stable spear]

https://referencesource.microsoft.com/#mscorlib/system/decimal.cs

[misty glade]

ah, thanks, that's what I saw too but I saw it was for framework 4.8

[stable spear]

https://referencesource.microsoft.com/#mscorlib/system/decimal.cs,134

[misty glade]

yeah, that's weird

Bits 0-15 (the lower word) of the flags field are
// unused and must be zero;

Wonder what for... digs into code some more

I'm assuming future expansion room for flags of some sort

[drowsy estuary]

Is there something that already exists for checking frustum vs box intersection? I'm surprised I havent found anything readily available so far for something that has to be common to about every 3d game out there.
(NOTE: will be used in performance-critical code)

I have a suspicion I can just use projection matrix for this but I'd rather not try to guess my way through this

[stable spear]

@drowsy estuary https://docs.unity3d.com/ScriptReference/GeometryUtility.TestPlanesAABB.html this maybe?

[drowsy estuary]

Ohhh right I knew I had seen something like that before! Thanks.
Do you happen to know if this is Burst-compatible?

Nevermind, Plane uses the old Vector3 and such

Cant check the source either since it disappears off in some external C++ call.
Oh well

Anyway, thanks

[tepid dove]

how do bullet hell creators manage bullet movement? right now i have each bullet starting their respective Move() coroutines when they're spawned (which may or may not start more coroutines subsequently) but i can't imagine that being super performant for hundreds or thousands of bullets. should i be using async/await? or just updating their position/rotation through Update? or something else

[drowsy estuary]

Regarding frustum vs box culling:
I found this which is stated to occasionally result in "false positives" (as in; says it does intersect while it actually does not) for the sake of a more simple implementation. Is this a common compromise to make for games?
https://www.gamedev.net/forums/topic/512123-fast--and-correct-frustum---aabb-intersection/

[stable spear]

particle system + custom shader is one possibility

[drowsy estuary]

As soon as you have to deal with 1000+ or even just 100+ of of something, consider DOTS/ECS if you really want to avoid performance problems later on.

Otherwise, I'd advice just letting Physics do the work for you and give each bullet a proper velocity and let the Physics engine simulate the motion for you.

[tepid dove]

either way i'm gonna have to refactor a bunch 💤

[hushed fable]

Well first I would start by actually figuring out how likely it is that you can get away with the Unity way. How many projectiles do you actually end up with and what sort of headroom is there?
If that fails, starting by moving projectiles to your own collections and running logic on them linearly by yourself and doing DrawMesh* calls yourself should do.

You don't need to move to ECS (and probably shouldn't until they consistently have versions available) to take advantage of some of the ECS concepts and apply them yourself

[drowsy estuary]

@tepid dove If rendering becomes the primary cause for performance problems, check the following things:
Plan A: Check if draw-calls for your bullets are properly batched. If not, take steps to ensure they are.
Plan B: Use Graphics.DrawMeshInstances(...). This is blazingly fast if all you want is just blast 1000 of identical objects onto the screen.

(But if physics and/or motion updates is the primary cause for performance problem, check what I said a few minutes earlier.)

async/await is something you generally don't resort to in Unity.

Primarily because a lot of Unity API is not accessible on other threads.

If you do want to go multi-thread, Jobs/DOTS/ECS is the way. However, this is really different to the MonoBehaviour approach and requires you to basically re-learn what you know about (game) programming.

[tepid dove]

yeah idk much about jobs/dots/ecs to begin with, all i know is that it's good for handling large amounts of objects but has a really different workflow? (also seemingly unfinished for an indefinite time, thanks unity)

[drowsy estuary]

Remember to use the profiler often! It will tell you exactly where code is slow

[tepid dove]

anyways thanks for the ideas

[drowsy estuary]

Super different workflow, yes.

Jobs is the best way to get some of the DOTS advantages but without having to learn so much new things.
Jobs is already fully functional and no longer in the preview stage.

The only catch is that you have to figure out how to 'pack' what you want to calculate/execute into a Job but somehow not have to rely on the Unity API calls that are restricted to the main thread.
Generally this means only using pure data types such as Vector3.

If you use the special Burst-compatible data types such as float3, you can have your Jobs 'Burst-compiled'. This is basically a free 25x performance increase.

But for now I would advice:

• Use Physics for motion updates.
• Use draw call batching or mesh instancing for rendering.

[tepid dove]

wait, if i'm using the physics engine for handling movement, does that mean all my bullets need a rigidbody? right now i have no rb on them

[drowsy estuary]

Yes.
You also don't have to do collision detection yourself anymore. You can now use Colliders for this.

[tepid dove]

oh i've been using physics.overlapcircle to handle collisions

[drowsy estuary]

If you need to execute certain code upon collisions, Unity has special API for this.

[drowsy estuary]

Lol I guess that also works.
I often use Raycast for bullet intersections which also doesnt require colliders (although the object the bullet will hit does need a collider).

Up to you to decide how to approach collision handling.

[tepid dove]

alright cool

[cedar ledge]

how do i choose more modern .net versions

it doesn't give me the same options as here

also, i have the vs solution turned on, but i cant access the build settings inside vs

[hushed fable]

Modern is the only option you have in recent versions of Unity afaik

[cedar ledge]

wdym

so .net standard 2.1 is .net 4.7.2 or whatever?

in that case, why cant i see the assembly properties?

[hushed fable]

.NET Framework is .NET 4.8

https://docs.unity3d.com/2021.2/Documentation/Manual/dotnetProfileSupport.html

What build settings do you need in VS?

[cedar ledge]

Assembly-CSharp -> Properties

i found the setting

it wasn't set to 4.8, actually

nice

C#10, baby

[sly grove]

wow it's almost like writing Java now

[cedar ledge]

https://forum.unity.com/threads/unity-future-net-development-status.1092205/

atm unity supports all of csharp 8 and a subset of csharp 9

[hushed fable]

@cedar ledge No reaction gifs please.

[misty glade]

FWIW i rather enjoyed that particular one 🙂

[cedar ledge]

1984

[misty glade]

re: conversation a few hours ago @stable spear @sly grove .. bigdecimal implementation is complete and somewhat tested.. I spent more time than I'd like to admit writing the "big number to english" methods using the Modified Conway-Wechsler system (https://googology.wikia.org/wiki/-illion)

these numbers work up to 10e3000 or so

i mean, the english conversion part

"novemnonagintanongentillion"

[stable spear]

what will people be doing with all this money

[misty glade]

heh

our game's soft resources are energy (joules) and matter (quarks) so I'm sure we can explain away the nonsense of the absurdly large numbers

[sly grove]

A clicker/idle game I assume

[misty glade]

a bit - less clicker, though

[stable spear]

even in the real universe, the total amount of energy goes up to no more than 10^80

[misty glade]

a third of the game is the idle component that I'm building now, another third is the 1v1 battle component thingy you've likely seen, and the last third is the collection mechanic

[sly grove]

good thing we're not limited by the real universe

[misty glade]

technobabble handwavey quantum tunneling discover new universes atomic folding blah blah blah

[cedar ledge]

that's number of atoms

[misty glade]

i'm not a content creator i just write bad code

[stable spear]

I figure each atom has no more than 1 joule

[misty glade]

the nice part of my nonsense i made today is that it's really a small struct.. basically:

public struct BigDecimal : IEquatable<BigDecimal>
{
        public decimal Mantissa { get; private set; }
        public int Exponent { get; private set; }
.. snip a whole lotta nonsense ..
}

i think 22 bytes uncompressed

actually I don't know about that, i don't know what the overhead of a struct is

[stable spear]

20 bytes exactly

[misty glade]

16 for the decimal and 4 for the int included? = 40 total?

or you're saying there is no overhead

[stable spear]

16 + 4 = 20 yes

[misty glade]

ah, cool

[stable spear]

there's no overhead if you're just storing it on the stack (e.g. as a local variable)

[misty glade]

several bytes of which are unused flags on decimals 😐 but i can live with that

[stable spear]

Oh, there might be padding if decimal had an alignment requirement beyond 4

Presumably it doesn't

[misty glade]

my data structure packs down to 4 bytes with MessagePack

i thought i'd have to give up a lot in terms of network size but that's pretty good.. messagepack is black magic

[stable spear]

uh. is that assuming all the numbers are the same?

[sly grove]

I think MessagePack does some stuff like - if your int is small it shoves your 32 bit int into maybe 5 bits, and then shoves other data right next to that, etc..

so if your numbers get larger they may not pack down to 4 bytes

it is very slick though

[misty glade]

Ah, you're right dox, it does get a little bigger as more non-zero bytes get used

looks like it caps out around 33 bytes

And of course, now that I watch this I realize that I probably should have just gone with a float since it gives me more than enough precision and I could have faked the equality part for values i care about

ah well i'm done mucking with it for today

@sly grove yeah.. like I said.. black magic. https://github.com/neuecc/MessagePack-CSharp/blob/master/src/MessagePack/Internal/UnsafeMemory.cs

I have virtually no idea how this works but... I am glad that it does

[stable spear]

consider also double if you need more

[misty glade]

Yeah.. probably not, tbh. I think most of the time I only need 4-6 sig figs. Decimal was attractive so I could do native equality with decimal.operator= for small values

[stable spear]

why would you need to do equality?

[misty glade]

at small levels - ie, before they've gone postal with exponential, the next generator will cost a human-readable amount, ie, "10 joules"

so the number the user sees will be ticking up every time the generator "cycles" (ie - every 10-20 sec)

it would be weird if the user had a float joules of 9.9999999 even though the game says it pays out "1" per cycle or whatever

I was just hoping to avoid any of that float nonsense

in reality it's not that big a deal - i could have rolled my own operator= and just hidden epsilon issues away from the client entirely.. but.. i started down this path this morning without having thought of that

tbh it might not be that big of a refactor, since really all that's exposed is a mantissa (deci-float-ish-thing) and exponent (int)

[stable spear]

eh?

you didn't answer the question.

even with decimal, equality is going to be too unstable for a meaningful comparison between different calculated values

with a float/double you can always round before printing

[iron kettle]

(already posted this in General Code) Does anyone know the best way to access/alter the axis speeds for a Cinemachine FreeLook camera? I'm trying to make it so the camera will only move when the right mouse-button is pressed

[clear steppe]

Hey everyone I come here seeking knowledge.

I've recently re-downloaded my project from perforce and the only problems that I am getting are the following:

It's a third party plugin that I just updated and it doesn't seem to want to detect the other file which is living in the same project (abeit in the Editor subfolder)

Any ideas if the assembly references are causing this? I've been pretty stumped and my lack of experience with dlls and assembly references is showing here. I'd appreciate any insight on the matter. Thanks guys!

This is what my solution explorer looks like. I actually had never seen those -firstpass projects before. and I don't know what do different projects mean and how those might affect compilation.

[flint sage]

Looks like some is in an asmdef while the rest is in the default assemblies

[clear steppe]

I have no idea how to move those over to other assemblies. What's the difference between an asmdef and default assemblies ?

I've looked online extensively but I can't seem to find the answers I'm looking for.

[cedar ledge]

it's public, not internal. idk why it's complaining

[austere jewel]

Assembly definitions are files that manually define assemblies in a project https://docs.unity3d.com/Manual/ScriptCompilationAssemblyDefinitionFiles.html
They make for great separation of concerns, manually defining how assemblies are referencing each other

[cedar ledge]

improves compilation time in some cases

[austere jewel]

Editor folders under assembly definitions do nothing, only the asmdefs define what assemblies are used when they're a part of the folder structure

It's hard to know what's wrong without knowing where those two files are in your project in relation to the assemblies

It seems though that you have one file in an asmdef relying on another in the editor assembly. Scripts in assembly definitions cannot rely on scripts in the default assemblies

[clear steppe]

What would you suggest I do to fix this? Move everything to the same assemblies?

[glass wagon]

this is annoying.

[austere jewel]

I'm unsure how you got yourself into this. You could remove their assembly definitions, and that would fix it.
The "proper" way is to learn assembly definitions and reference things properly, but that's a lot more work.

[clear steppe]

Yeah right now the project won't move forward too much into the future, so occam's razor is good now.

How would I remove their assembly definitions? What does that imply? Is that set up in unity or in vs? Is it some sort of file?

[austere jewel]

search for t:AssemblyDefinitionAsset in the project

[clear steppe]

Any case, I'll do what you suggest.

[austere jewel]

and find the one that is relevant

then just delete it

[clear steppe]

Hmm... My vs project can't find t:AssemblyDefinitionAsset in the solution.

Should I just straight up look for .asmdef files in the project and delete them?

[austere jewel]

Not VS, Unity.

[clear steppe]

OH

IT FREAKING WORKED

THANK YOU

Oh my, i had been stuck with this for weeks. It was until a moment ago that it occured to me to ask here.

Thank you so, so much.

[mental zenith]

Damn looking at advanced code scares me...

[cedar ledge]

just blend in

hey guys. quaternions are so easy to work with, right? 😉

[flint sage]

Sure

The math behind them not so much

[safe whale]

Hello,
I'm using the function newMesh.CombineMeshes() to combine a mesh that is over Int16.MaxValue (65536+ vertices).
To do so, I'm setting the newMesh.IndexFormat to UnityEngine.Rendering.IndexFormat.UInt32 but once the newMesh.CombineMeshes() occurs, event if the format is well set and that the mesh has 69k vertices, it only combine to 65536 vertices

I'm using CombineMeshes() with only 1 CombineInstance, could it comes from there ?

[flint sage]

Guessing combine mehses doesn't support uint32

[small badge]

https://forum.unity.com/threads/solved-mesh-combine-and-65k-limitation.541496/ claims that it does, as long as the target mesh's index format is UInt32 (which it sounds like it is in this case).

[spiral kelp]

I'm trying to download a .hdr image at runtime to use for a skybox. regular jpgs and png's work fine with the built in texture downloader method, and hdr works fine with the shader/material when imported at editor time. but whenever I set it at runtime it gives the dreaded question mark texture. what am I missing?

[plucky laurel]

the hdri importer is most likely part of the editor

meaning its not available in build

[small badge]

UnityWebRequestTexture only supports JPGs and PNGs containing colour data. For other formats you would need to either use a third party library, or download it as raw texture data and load it with Texture2D.LoadRawTextureData.

[safe whale]

Yes but I think that it's not supporting more than 65k for each CombineInstance in the array. I tried to duplicate my CombineInstance, and with my 69K vertices mesh, it combine to a 131072 vertices one. So it goes over Int16.MaxValue easily, but for one CombineInstance it doesn't, cause 131072/2 = 65536

[spiral kelp]

Thanks, missed the jpg and png only bit. trying with the raw data loader now

[small badge]

Alternatively ImageConversion.LoadImage might be able to load EXRs? The documentation is ambiguous.

[plucky laurel]

it may load it but pretty sure the sphere etc wrapping is done in editor only

[spiral kelp]

the builtin panoramic skybox shader handles the wrapping fine (regardless of texture passed to it), just trying to work out the download part 😅

wellll at least something is loading now! probably on the right track, thanks for the help!

[fallen karma]

Is it to be expected that GetComponentsInChildren does not work with generic interfaces? i.e. GetComponentsInChildren<IMyInterface<SomeType>>(). For me it returns nothing even though it should

[mint sleet]

      public Action<BucketMarker> OnMarkerClicked;
      private void Start()
        {
            OnMarkerClicked += FocusToMarkersPosition;
        }
``` Hey, I wonder if I need to unsubscribe the Action in OnDisable function like we do in "delegates"?

[fallen karma]

Action<T> is a delegate, if you need to unsusubscribe to clean up dangling references you should

[mint sleet]

but in unityevent we do not need. right?

[plucky laurel]

you need to same way

UnityEvent is just an object

[cyan vector]

How do I make a checkpoint gameobject inactive when the scene is reloaded from a save if it has already been activated?

[devout hare]

myCheckPointGameObject.SetActive(false);

[plucky laurel]

your objects have to check what "stage" they are in, i.e. if you have somewhere a flag "LoadingFromSave" set then in OnEnable the object has to handle it accordingly instead of doing normal activation routine

or whatever the analogous reasoning would be in your architecture

[quartz stratus]

Have you already implemented a save system and if so what type are you using?

[cyan vector]

I'm using this save system: https://github.com/IntoTheDev/Save-System-for-Unity
And my current script is here: https://paste.myst.rs/sxo9a4jw

[tough tulip]

I wasn't aware that it didn't work. But nonetheless you can write your own helper method to iterate over all child Transforms recursively. In each transform do GetComponents for MonoBehaviour, and then try to typecast it to your generic interface. If the typecast passes, simply return the component

[cloud terrace]

Does anyone know of scripts I can use (even if they're not free) for culling cubic voxels that aren't seen by the camera before I send them to the renderer?
for example: a script that takes in a list (or an [x,y,z] array) of cube positions on a grid and the camera's position, FOV, angle and outputs a list of cubes minus the occlusion culled ones

[sly grove]

how are you rendering them? I assume they don't all have their own MeshRenderers?

[cloud terrace]

a surface shader and Graphics.DrawMeshInstancedIndirect

[sly grove]

I think you'd have to do the frustum culling on the GPU in this case

Looks like some people have done this... https://forum.unity.com/threads/gpu-frustum-culling-is-working-now-but-how-to-write-lod-occlusion-culling.702596/
Though I'm not exactly sure what technique(s) were used

Seems like they're using a separate compute shader to do the frustum culling.

[cedar ledge]

Is this moire effect from the graphic or is it a photo of a screen?

[cedar ledge]

There's a way to get interfaces, but not like that

FindObjectsOfType<MonoBehaviour>().OfType<IMyInterface>().ToArray()

Then you just need to filter it for children

[sly grove]

Its better if your interface has a non-generic parent interface which you can use

Or I think you can do:

public interface Example : IMyInterface<SomeConcreteType>```

and have your class implement that

and then search for that 🤔

[cedar ledge]

I'm not at my computer but maybe FindObjectsOfType<MonoBehaviour>().Where(i => (transform.children.Contains(i.transform))). OfType<IMyInterface>().ToArray()

[cedar ledge]

I don't think the generic typing is the problem

It's that getcomponents doesn't find interfaces at all