archived-code-general

[clever lagoon]

oh.. not at me.. nvm

[left violet]

no I think he was refering to calling methods

[hard viper]

literally last week, a feature in Unity was broken because they had a string that said “ScrollTextSeachBar” vs “ScrollTextSearchBar”. Which made a whole range of UI features totally unusable because some dumbass used a string, and a separate dumbass made a spelling error.

I make a non-generic abstract parent when I want concrete methods to be inheritted by all children, and maintain internal variables . Interface otherwise.

my spawnpoint has a ton of machinery, so I saw it was greatly beneficial to make an abstract base class

[left violet]

hahah

[clever lagoon]

@left violet agree with Loup 100%.. my rule of thumb: compiler errors are better than runtime errors.

[left violet]

I was not saying anything against his suggestion 😄

Clearly it is better practice to not use Strings

[clever lagoon]

sorry, didn't mean to imply otherwise.. was just giving a reason WHY

[untold siren]

How do I check what layer my navmesh agent is currently walking on? I've got two so far:

• Walkable
• Wall
  however it seems to only always return walkable?

[latent latch]

why is your agent walking on walls

[untold siren]

to simulate the AI wallrunning

which is why I need to be able to check the layer / area the navmesh is currently walking on, so I can rotate the mesh

[latent latch]

probably can do it with the navmesh raycast but I'd expect there's another method

[molten radish]

what do u like more rider or vs code?

for c#

i kinda prefer rider

[untold siren]

rider 100%

[molten radish]

ye its a lot faster with compiling maybe bc its live... and also auto complete seems to be faster too

rider auto saves right?

[untold siren]

mhm

[molten radish]

ah thats why

ive seen there has been a unity script editor... what happened to it?

or is this just vc?

[simple egret]

This one got discontinued. Your mainstream options are now Visual Studio, Visual Studio Code, Rider

[molten radish]

alr... yeah i think u dont even call them mainstream seen a graph where other was at 0%

for used programs for unity

[simple egret]

Some still want to suffer, we've seen a few use Notepad

[heady iris]

I saw a student using Word once

[molten radish]

xD

[heady iris]

(in a class I was TAing)

[untold siren]

my lecturer used notepad once

bro actually has a phd in "metaverse technology" too

[molten radish]

best ide has typing function

[simple egret]

I mean how do you even compile code in Word, that thing is stored in XML

Let's do it in Excel instead. It has tabs, one sheet per file

[molten radish]

._.

[untold siren]

I mean powerpoint is turing complete

;)

[molten radish]

btw. when using shaders would u rather do it in unity or in blender

[untold siren]

still stuck on this..

        private void CheckLayer()
        {
            if (_navMeshAgent.SamplePathPosition(NavMesh.AllAreas, 0.1f, out var hit)) return;
            Debug.Log((hit.mask & _wallMask) != 0 ? "On Wall" : "On Floor");
        }

it always prints "On Floor"

[heady iris]

Blender has a completely different rendering system. You must create shaders and materials in Unity.

[molten radish]

materials are often imported tho

[quaint rock]

shaders dont carry over though

also the materials rarely come over fully correct, generally just the names of them and the slots and where they apply comes over

[heady iris]

Some info can make it over, yes

[quaint rock]

then you edit those in unity to have the correct shader and textures

[heady iris]

It can work if you use the Principled BSDF shader in blender and directly collect image nodes to its inputs. But arbitrary materials aren’t going to survive the FBX export

[latent latch]

Yeah, I've given up importing anything but general mesh datas

[molten radish]

alr. well ill learn how to do shaders in unity then

[heady iris]

I just create materials in Unity and then assign them in the model importer

[latent latch]

Shaderlab is dumb easy if you like visual scripting

[molten radish]

i kinda dispice visual scripting

[heady iris]

You mean shader graph?

[latent latch]

shader graph right

[heady iris]

Shaderlab is the text based format

[heady iris]

I do too, but visual editing works way better for shaders

I use both styles depending on what I’m doing

[lean sail]

I dont like visual scripting either, but shader graph is really good. Visually let's you even see what happens in intermediate steps

[latent latch]

majority of special shaders you'd make is just copy pasted code anyway

[heady iris]

Writing HLSL works better if you need to do really elaborate stuff

[latent latch]

no reason to reinvent the wheel

[static matrix]

I am trying to get a system going where a creature's legs go the the three closest points (basically just the second point in a linerenderer) but I am running into some issues
The script is kinda lengthy so I will put it in pastebin but I am not totally sure what isn't working here

[quaint rock]

you will have a easier time with shadergraph since it does alot of the work for you. you need to implement more yourself if using hlsl and shaderlab with urp

[static matrix]

https://pastebin.com/VmgfYTf4

[latent latch]

hlsl is also just a bunch of unity methods and macros

[molten radish]

mhm mhm noted.. but visual coding for other functions are often a mess...

[static matrix]

im getting an issue where two legs are going to the exact same spot (same coords not just visually) and some just go through walls

examples:

unsure what the issue is here

[zinc parrot]

I ended up doing this in a kind of hacky way, by somewhat abusing serialized static nested structs inside an editor window

[deft dagger]

hey guys, i need help with unity 2d, i have jump code that works fine, but i want to make the character jump only when is in the floor collider so i made something like this :
void OnTriggerEnter2D(Collider2D other) {
if (other.CompareTag("Floor")) {
JUMP;
}

why doesnt it work ?

[static matrix]

A:
include ```
before and after you enter your code
and B:
what tf does "JUMP" mean
like what is your actual code

[deft dagger]

i just wrote JUMP as a short term for the jump code i wrote

so i dont have to type that much 😅

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

public class jump : MonoBehaviour
{
public float velocity = 1f;
public Rigidbody2D rb2D;
void Start()
{
// rb2D = GetComponent<Rigidbody2D>();
}
void OnTriggerStay2D(Collider2D other){
if (Input.GetKeyDown(KeyCode.UpArrow) && other.CompareTag("Floor"))
{
Debug.Log("Gibtu");
rb2D.velocity = Vector2.up * velocity;
}
}

void Update()
{
}

}

this is the code

[static matrix]

why do you have rb2D's assigner commented out?
its also transform.up not Vector2.up
and idk how "ontriggerstay" is evaluating but it is a lot better to use an overlapbox with layers

IsOnGround = Physics2D.OverlapBox(transform.position - new Vector3(0, 0.8f), new Vector2(0.5f*transform.localScale.x, 0.21f), 0, Ground);
        if(IsOnGround && Input.GetKeyDown(KeyCode.Space)){
            IsOnGround = false;
            body.AddForce(new Vector2(0, JumpPower*15));
            transform.localScale = new Vector3(1.5f, 0.5f);
            
        }

like this
(Ground) is a layermask

[deft dagger]

i see, thanks man i will try this

[static matrix]

anyways, anyone have any idea whats up with my code?
is there some glaring issue I am not seeing

[vagrant cargo]

How do I stop player from clipping into walls/other colliders?

[static matrix]

hmm
I was going to say continuous collision but you already have that
maybe the colliders just don't line up right?
you could always make it so that object are not moved when pushed, but I don't think that is what you want

[vagrant cargo]

It also happens slightly with the wall so I don't think pushing is the problem (though it does exaggerate it)

[molten radish]

well this is prob whats happening. fixed update isnt fast enough for the player speed

[vagrant cargo]

My code is in Update() not fixedupdate

[molten radish]

that doesnt matter update isnt simply fast enough for its speed

[vagrant cargo]

I see, does that video have a way to fix it?

[static matrix]

this is why physics code is usually in fixed update
but the collision is set to continuous doesn't that usually fix that issue?

[molten radish]

no sadly not... he says he covers it in another video

[molten radish]

but this should fix it

[vagrant cargo]

Ok I'll try it

[molten radish]

did it work?

[heady iris]

Continuous collision detection means that the rigidbody actually sweeps out a volume in space, rather than just checking if its new position is occupied

It's more expensive, but also more accurate

Thera are a few different modes you can use. https://docs.unity3d.com/Manual/ContinuousCollisionDetection.html

(well, two of the modes sweep out a volume in space)

[molten radish]

sweeps out?

[vagrant cargo]

I had to change the interpolation to none, now it's significantly better but still a bit there

[heady iris]

Interpolation just makes the physics system change the position of the object each frame

instead of only in physics updates

it shouldn't be affecting the actual behavior

[molten radish]

hmm tried Continuous collision still clips through

[vagrant cargo]

It is for some reason

[vagrant cargo]

Oh it's possible it's just a visual thing if that's what you mean

[static matrix]

blue light filter

anyways, still unsure what is causing my issue
genuinely stumped

[static matrix]

this

[molten radish]

i think theres still a way to fix it maybe with apply force

[molten radish]

yes that works

[terse quarry]

I've implemented a pause button in my space simulation by setting the timeScale to 0 when its clicked. When you click on a planet in my game, I want something to appear, but if the timeScale is 0 will this prevent it from appearing?

or does it just stop update()

[static matrix]

eh as long as there is no lerping it should work I think????
Idk

[heady iris]

Setting the time scale to zero won't stop Update from being called.

deltaTime will simply be zero.

[molten radish]

also at a high enough speed it would

[heady iris]

show us your code.

[vagrant cargo]

using AppCore;

using UnityEngine;

namespace Game.PlayerComponents {
    public class PlayerMovement : MonoBehaviour{
        [SerializeField] private float movementSpeed = 5f;
        
        protected Vector2 _currentMovement;
        protected Rigidbody2D _rigidbody2D;
        
        // Unity functions
        private void OnEnable() {
            App.Instance.inputManager.OnMovement += OnMovement;

        }

        private void OnDisable() {
            App.Instance.inputManager.OnMovement -= OnMovement;
        }

        private void Awake() {
            _rigidbody2D = GetComponent<Rigidbody2D>();
        }

        private void FixedUpdate() {
            MovePlayer();
        }
        
        // Overridable functions
        private void OnMovement(Vector2 movementInput) {
            _currentMovement = movementInput;
            _currentMovement.Normalize();
        }
        
        protected virtual void MovePlayer() {
            _rigidbody2D.position += _currentMovement * (movementSpeed * Time.deltaTime);
        }
    }
} ```

[static matrix]

what is protected and virtual for I have never really used those keywords

[molten radish]

as i said i would do
cuz acceleration = speed/time
and force=accelaration*bodymass
Vector3 force = (movementspeed / Time.deltaTime) * _rigidbody2D.mass;
_rigidbody2D.AddForceAtPosition(force,_currentMovement.normalized);

[vagrant cargo]

how would force be a vector3? isn't it just a float?

[molten radish]

well it needs vector3 variables um... and intellij says so

[vagrant cargo]

Why should I use addforceatposition instead of just addforce

[molten radish]

uh u can use the other too

[static matrix]

force is vector3 because the
wait

huh????

[molten radish]

ye prob better if u use just addforce

if u dont want rotations

rotation would be kinda funnie tho

[vagrant cargo]

I feel like adding force just in general isn't the best way to do it in my case because I don't want the velocity to be changing around I just need it to move

[static matrix]

you could lerp it if you want that smooth feel

Lerp my beloved

[vagrant cargo]

It's not that the movement isn't smooth it's that it keeps glitching in/out of other colliders

[static matrix]

just change the layers so it doesn't look like it
we love rugsweeping

[molten radish]

well there doesnt need to be accelaration

[vagrant cargo]

That's true I could just do that

[molten radish]

and doesnt accelaration happen anyways if u multiple by deltatime

[vagrant cargo]

No because before I was just changing the position manually, not changing the velocity

[molten radish]

i mean like

[static matrix]

make the wall appear on top because then it would look just like it was squishing a lil

like change the walls render prio to 1 and keep the player at 0

[molten radish]

it would still clip through if it goes bullet speed

[static matrix]

not if continuous detection is on

or it shouldn't if cont is on

[dapper schooner]

Has anyone ever run into this issue? I tried deleting the Library, Cache and Removing and Reinstalling the package. Updating Visual Studio. Rebuilding the Solution and nothing seems to get rid of this error.
Library\PackageCache\com.unity.ugui@1.0.0\Runtime\UI\Core\Image.cs(1,1): error CS1056: Unexpected character '

Also worth noting that the Image.cs file will not open in VS. It opens in Notepad (which is empty when it opens) because VS says it does not support the file type.

[molten radish]

lemme check

[molten radish]

nope

goes through

[static matrix]

well this is 3d

[molten radish]

oh

maybe 2d it aint

well i dont think my method actually changes really the speed tho?

[molten radish]

why would velocity be changing?

[latent latch]

here's the secret

raycast yourself

[molten radish]

i still dont understand

[latent latch]

use as many raycast methods you need if you're clipping

[vagrant cargo]

That's what addforce does no?

[vagrant cargo]

As in send a raycast and don't move if you're x distance from a wall?

[molten radish]

no addforce can be constant,,, the force can be constant

so the speed would be too

[static matrix]

how "optimized" is raycasting
like I feel like I would get a big perf hit if I send out like 20 raycasts but will I?

[latent latch]

well, performance hit vs clipping

choose 1

[molten radish]

can someone explain raycast i looked it up but it doesnt fit in this context

really

[latent latch]

you clip through walls because the velocity is greater enough that your gameobject should be behind it between an update

so what interpolate does is it raycasts ahead an update the length of which your character would end up with that velocity in thenext update

[molten radish]

ah

seems like a lot of calculating

before actually doing stuff

[latent latch]

one raycast is fine, but when you've more complex meshes then that's a lot of extra calculating

[molten radish]

yeah so it wont really work with super fast stuff

[latent latch]

fps games will interpolate bullets via raycast (assuming they are not hit-scan / instantaneous registering bullets)

but those are relatively cheap

[molten radish]

they do?

but doesnt ur laptop just die

[latent latch]

it's not that expensive. You're just doing an extra operation for them, but stuff like vehicles it can be a little more problematic

[vagrant cargo]

I don't need something super fast it's in 2D so that's fine, do you mean adding extra raycasts in addition to the interpolating ones?

[latent latch]

yeah, unity does ok, but if it's not enough then add your own custom logic for checking

[vagrant cargo]

Ok how would a general outline of that work

[molten radish]

arent there options... like isnt there some kind of thing that just doesnt work on ms and is just constant what does force do for example why does it work

force looks at the other force right? and thats infinity prob

[latent latch]

if the position from velocity is greater then the wall detected by the raycast, just postion yourself to the closest point near the wall

[molten radish]

btw how hard is it to port 3d games to android

do u need a lot of optimiziation? or can it run quite fine

ive heard game companies loosing detail from their original works

like when looking at concept art vs finished product

[leaden ice]

"port" is very different from building a game for mobile from the ground up

https://blog.unity.com/games/optimize-your-mobile-game-performance-tips-on-profiling-memory-and-code-architecture-from

[molten radish]

ill have a look at the blog thanks

[severe sable]

is there ANY tutorials or demos or even vague information on how to structure vehicle creator games? Like stormworks, robocraft, terratech ect

I can't find anything and I've been searching forever

[jagged snow]

For some reason the raycast hit is detecting the parent object rather than the transform that it hit. I attached the collider to the head transform but teh debug.log is saying that its parent. I'm certain that the collider on every other object related is disabled.

Im not sure how to fix this

[leaden ice]

How do you know that

Did you know that RaycastHit has multiple properties on it

[jagged snow]

            {
                Debug.Log(hit.transform);```

[leaden ice]

yeah because 4you're using .transform

did you read the documentation on that?

Maybe you actually want .collider

https://docs.unity3d.com/ScriptReference/RaycastHit-transform.html

The Transform of the rigidbody or collider that was hit.
Rigidbody first if it exists

[jagged snow]

oh so raycast.hit.transform always calls the parent?

ah

[severe sable]

I've been working on systems for this for ages now and I'm close but it feels so god damn messy and I really need something to go off of

[jagged snow]

thank you

[leaden ice]

Big hint: Structure your craft as a tree (in data structure terms)

Everything follows from there

[severe sable]

Thats what I've tried my best to do 😭

[leaden ice]

Well are you well versed in graph/tree data structures and algorithms?

[severe sable]

I feel like I'm just overcomplicating things

[leaden ice]

What issues are you having

[severe sable]

I feel my structure sucks since I am having issues with passing things between classes and gameobjects acting as containers but then also a container class attached to them, which the gameobject is only there as a visual container in the inspector

[leaden ice]

Sounds like you're having trouble separating your GameObjects and Components from your actual data structure of your game

[severe sable]

I have all the data setup and everything for the blocks, but I can't wrap my head around finalizing the connections. I even have the methods to add blocks to a ShipContainer

[leaden ice]

I recommend a pretty strict separation. Use GameObjects and Components as more or less the presentation layer for a "pure" data structure under the hood

[severe sable]

Presentation layer? Let me show what I have

That sounds like what I have done

Like this is what I need right?

[leaden ice]

Have you ever heard of MVC?

[severe sable]

model view controller

I am a bit familiar

[leaden ice]

In an MVC model your Unity MonoBehaviours (aka components) would be the view/controller layer, aka the presentation.

But the model itself would be a separate internal data structure

[severe sable]

Interestingly enough that was the setup I had prior to a bunch of changes I did today

I just found myself lost in classes with little code in them and it felt like I must be doing something incorrectly

is that not the case?

[leaden ice]

Small classes are a good thing

[sick timber]

yo can someone help me with an input system thing, I'm trying to turn my sprite towards the direction it's going but all the tutorials I've seen use axis inputs and I can't because the arrow keys will be doing a seperate thing than the wasd.

[devout radish]

have different actions for the arrow keys and wasd

[sick timber]

yeah. It's a two player game and player 1 uses wasd

[devout radish]

So create two bindings, one for player one and two

Movement P1
Movement P2

[sick timber]

yeah I have them. My problem is how to turn player 1 in the direction its goin

[devout radish]

Atan2 is a function that can be used to make an object look in the direction it's moving

Or if you want to be able to control the rotation specifically

You can use other keys

like Q and E

[sick timber]

k imma try atan2 thanks

[hard viper]

i worry unity refreshtile is slow when going through all my rules in a row. Assuming a given ruletile can be expressed as only using 3 rules (rule, not rule, and no rule), do you think it would make sense to:

1. In editor, serialize a 256 long array of all possible neighbor combinations, and the answers.
2. Ingame, on refresh, evaluate all the rules on the 8 neighboring spots, make an int key, and then go access the array of answers.

does this sound smart, or dumb?

[chilly surge]

Answer to pretty much all performance questions: profile/benchmark first.

But yes generally precomputing a lookup table and doing simple lookup afterwards is faster, it's just whether it's worth spending your dev time on it or not.

[lunar python]

Hi, I am struggling whether to pick Inheritance with Abstract Class or Interface for Controller's of Player's, Enemies' and Objects. I would like to return Controller by using GetComponent

I know that I can return derived Controller using Inheritance with abstract class, can I do it with interfaces? Thanks

[cosmic rain]

Sure, you can cast it to whatever type you expect it to be.

[errant raven]

Does i5 4th gen, 4 gb ram, GT 610 support Unity 3D?

[cinder juniper]

does anyone have a good first person movement script???

[somber nacelle]

don't crosspost. and we don't do handouts here. go watch one of the hundreds of tutorials to learn how to make it yourself

[errant raven]

. Anyone?

[somber nacelle]

that's not a code question. you can also just look at the system requirements on the docs

[cinder juniper]

can u suggest a good tutorial?

which teaches the best movements

i need to iclude rigidbody

[somber nacelle]

hope this helps

[cinder juniper]

bro, i searched yt but i cant find good ones with rigidbody

[cinder juniper]

this just redirects u to google -_-

[cinder juniper]

they use the other thing

[somber nacelle]

yes because it is a link to the google homepage which you can utilize to search for tutorials that cover the concepts you want to learn about

[dawn nebula]

What do you guys think of this technique?
https://www.youtube.com/watch?v=kVhuIHXoSIw&ab_channel=PitiIT

Tl;dw in cases where you want a GameObject to have more than one tag, you create Tags as a instance of a scriptable object, and then give the gameobject a component with a list of tags. Check the contents of the list for the Tag SO you're interested in.

[west lotus]

Why SO? This can easily just be a flags enum

[lean sail]

An enum is definitely fine for this. I would only use scriptable object if you want to somewhat enforce a system where things can only be labelled as certain things.
For example in my game I have character factions. I store the information for what a character is aggressive/friendly towards in an SO but the actual faction itself is just an enum. That saves me from possible errors of trying to assign the same list to every AI

[latent latch]

bitwise enum is best enum

Unity tags into bitwise enums when

[cosmic rain]

There's already layers for that😛

[latent latch]

yeah, but after you grab the colliders you want you can also grab tags directly from the hit gameobject without any GetComponent calls. I was thinking though that I could probably just make a static lookup for the tags which then points to additional bitwise enums

[frigid bone]

Anyone have any good tips for optimizing a grenade sending damage to alot of objects at once?
Im currently doing physics overlapsphere and then getcomponent and sending details through that.
I cant use broadcast message cause i need to send 4 variables to the function. (Unless there is a way around that)

Thought of one way , by using a new class with those 4 variables. But if anyone have any other tips , id be happy with that 😛

[lean sail]

Use the profiler to see what your actual issue is. Also I don't know what you think broadcast message is gonna do but you dont need it.
Also cant really help without seeing code

[frigid bone]

It works , i was just asking if anyone had any other way of doing it more optimaly , especially since getcomponent can be taxing

[lean sail]

Cannot suggest something "more optimally" if we dont see what you're doing in the first place

There should be basically 0 lag unless you're doing this with like 100s of objects

The profiler will tell you if it's actually GetComponent that is your issue, or if it is something else.

[main shuttle]

Personally I don't see how you can change much if you want to remain in standard Unity OOP.
As far as I know, you can't do jobs for this because it's all managed code and jobs don't work with managed code, and I don't know how you would go around that, but perhaps you can look at that.
Otherwise you need to get the components like you are doing.
You could do this way more efficient in DOTS/ECS if you really want a lot of AOE damage, but you would need to change everything, and like bawsi said, not sure if it really is an issue currently.

[frigid bone]

Yeah ill check the profiler and such if lag does occur. Im just being mindful of all the things that could contribute to lag. Make sure all the bricks in the house are clean from the start and all that.

I might have an over exaggerated fear of getcomponent

[main shuttle]

Normally you do that at the end of your development, just do a couple of iterations of fixing the top 3 biggest performance problems.
Don't prematurely optimize, you will spend a huge amount of time on it for little gain at the moment.

[latent latch]

Well, can refer to what I've mentioned up a few lines ;p

[lean sail]

You can get away with a LOT more than you think. Try experimenting to get rid of your fear. call GetComponent on something 10 times per frame, 100 times, 1000 times. See what actually becomes an issue

You can even add a for loop that does nothing except iterate 1 million times per frame and still have >100 fps

[frigid bone]

Yeah i really should. I have gone out of my way to eliminate getcomponent from every update function and only call it when needed.
Maybe it was a problem back in the day with unity and not anymore , cause my teacher told me like 12 years ago that you should avoid getcomponent where you can

[errant raven]

Most of the time the sys req don't help much

[latent latch]

at most, if you're going to use overlap you do want non-alloc because gc is the bigger problem

[errant raven]

Just asking if it'll be fine on a i5 4th gen

[lean sail]

Theres a difference of getting rid of it because you can cache something, and just avoiding it because you're unsure of if itll matter

[main shuttle]

Well, you should cache your components, if it's the same component each frame. That would just be wasteful not to do.

[frigid bone]

In this case , its just an overlapsphere , send the values of the explosion to 30-100 units in an area

So i dont need to cache them

[main shuttle]

Exactly, but grabbing the main camera component each frame is not a good idea, just cache that one.

[lean sail]

Well yea those you cant cache because itll be unique everytime

[latent latch]

you can also minimize the type lookup by reducing the components on the object that you're checking

[frigid bone]

😎 👌

[deft timber]

if you are grabbing components from the same object, it is better practice to do it in Awake

[frigid bone]

But yeah , i should experiment more. Cheers guys 🙂

[fervent furnace]

just assign it in inspector?

[main shuttle]

Not always possible with prefabs

[lean sail]

Doubt anyone would be able to directly answer this, still not a code question. If it fits the requirements listed on the docs then you can probably open unity. No guarantee you would be able to actually get decent performance and make something

[frigid bone]

in this case , agent , animator and rigidbodies can all the assigned in the inspector , but i did it like this to remove clutter in the inspector window

[latent latch]

I always assign in awake/start/onvalidate if possible

so many times my prefabs have broken serialization and I sat there wondering what the hell they were binding to

I rather the compiler yell at my that they are assigning incorrectly than realizing later they weren't assigned at all

[frigid bone]

Fair

I try to serialize as much as i can without too much clutter

I want them all to be about this size

I could even remove some of that , but its good to know for debugging

[vapid condor]

i made an asset that can help you organise inspectors, see #1194282415917629603

[frigid bone]

Ohh nice , ill have a look

[sly crag]

Hey guys, I'm trying to build my test version of the game for ios, I finished the process in Unity and I've got the Xcode project, How do I build an IPA and make it public so everyone that I want can download it, I've heard there's a special store for a test version of apps, anyone knows what is it?, and how can I build and publish on it? or there is a way without using any store?

[deft timber]

this is a code related channel

[sly crag]

there's no build related one!

[deft timber]

sure, this is still a code related channel

#💻┃unity-talk

#📱┃mobile

[sly crag]

ok thx

[static matrix]

I could try and switch from trying to get the closest points to just raycasting for where it will try to place the legs but that seems like it would be inconsistent

[hard viper]

@chilly surge Ty for the input. In my case, my refreshTile is super expensive because I have many tiles, with many rules, and my rules actually check multiple tilemaps for each evaluation. So my RefreshTile is a large chunk of my load time. Like 1/4 last I checked

[static matrix]

I think maybe I start with a simpler creature that wouldn't require any of this or require pathing to make work, and would help improve the game

[chilly surge]

If you identified that indeed being a performance issue then sure yeah, precomputing a lookup table is indeed a common technique. It's the classic trading memory for speed.

[hard viper]

256 shorts x 100 SOs is probably insignificant, right?

[round violet]

content :
a,b,c are floats
someVector is a Vector3

is it faster to do A or B ? or both are to close to be taken in account ?
Does the answer change if only having a ?
A:

someVector + new Vector3(a,b,c)

B:

new Vector(someVector.x + a,someVector.y + b,someVector.z + c);

[hard viper]

I’m aware this feels like premature optimization, but I think it will be worth it in this case. Since Unity’s Ruletiles aren’t really optimal. They are evaluated like if else if else if else instead of a switch

[tender patio]

Here is my WeaponSway code but I want to increase the Y rotation by 180

How can I do that?

[chilly surge]

A 256 array of shorts is like, half a KB give or take (or maybe 1 KB, not too sure if things are padded)

So yes that sounds fine to me.

[hard viper]

B is slightly faster probably, but not by much

[round violet]

funny, I would say A

[hexed pecan]

Can't you just add 180 to the angle you use for rotationY?

[hard viper]

i would not do B tbh, because it is messier.

[hard viper]

A creates 2 new Vector3. B only makes one.

[tender patio]

I can't

it's not working

[hexed pecan]

What did you try?

Also you could just do Quaternion.Euler(-mouseY, mouseX, 0) instead of two AngleAxis

Are those mouse values deltas or accumulated 🤔

[round violet]

also another question, why can't you change a transform position x,y,z coords independently like you would do with a normal vector3

[hexed pecan]

You can. Get the position vector, modify it, then assign it back

[tender patio]

I tried this but didn't worked.

Then GPT told me to do this

and now it's working

[hexed pecan]

I knew it's GPT..

[round violet]

what I meant was doing this : ```
transform.position.x = somevalue

[tender patio]

No this code is not written by gpt

I just asked the question

[hexed pecan]

https://unity.huh.how/compiler-errors/cs1612

[heady iris]

The "gotcha" is that transform.position is a property. It calls a method that returns the current position.

If transform.position was a field, this would be completely fine.

[round violet]

ty, this website also looks great, going to dig in

[heady iris]

you wouldn't be modifying a copy

[hexed pecan]

That's what I said too, add 180 to the angle, not to the quaternion

[hexed pecan]

Yeah it's pretty dope, made by vertx

[round violet]

so its not a ref

[tender patio]

yeah thank you!

[fervent furnace]

i think +new Vector3() or add them individual are the same thing
fetch six floats from ram to cache to registers
three add
load three floats back to ram
probably need some benchmark or directly read the assembly code

[buoyant crane]

A Vector3 is a struct and a struct is a value type. It’s sort of like changing a copy of an int, and expecting it to change the original.

int a = 0;
b = a; // b = transform.position.x
b = 3;```
A will still be 0 in this case

[heady iris]

Vector3 is a value type, yes.

Note that this doesn't depend on whether position is a property or a field. This is entirely a characteristic of Vector3

[fervent furnace]

int a=10;
public get_a(){
  return a;
}
get_a()=100;

[heady iris]

The really insidious problem happens when you get a copy of a reference type.

For example, Renderer.materials

That returns a new array containing the renderer's materials.

Modifying that array accomplishes nothing.

But it's not a compile error, because Material[] is a reference type.

This also causes friction when working with ParticleSystem

ParticleSystem.MainModule is a struct -- but it's full of properties that update the original particle system when set.

https://docs.unity3d.com/ScriptReference/ParticleSystem-main.html

[buoyant crane]

Here’s the specific error if you want to read more about it https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/compiler-messages/cs1612

[heady iris]

So you do this weird song-and-dance:

ParticleSystem.MainModule = system.main;
main.startDelay = 10f;
// you don't assign it back to system.main !!

You can't do system.main.startDelay = 10f; because the compiler sees you trying to set a value on a copy of a value type and assumes that it would be useless

😵‍💫

[round violet]

what is the difference between a property and a field ?
I thought "field" name was used for variables that are visible in the inspector.

is this a csharp or unity based name convention ?

[heady iris]

These are C# concepts.

Fields and properties are both members.

A member is anything that's part of a class or struct.

public class Foo {
  public int bar; // field on Foo
  public int Baz => 3; // property on Foo
  public int Buz() { return 5; } // method on Foo
}

[round violet]

what is =>

[late lion]

Shorthand for { get { return 3 }}

Expression body

[heady iris]

=> is the lambda expression operator

[fervent furnace]

lambda expression

[late lion]

Works for methods too:

public int Buz() => return 5;

[heady iris]

It appears in a few places.

[round violet]

so this is why Baz is a property, because it isnt returning self as a ref ?

[heady iris]

It's a property because it's...a property.

public int Foo {
  get {
    return 3;
  }
}

this is also a property

[round violet]

im trying to memorize the "difference" between a property and a field

[heady iris]

A field is basically just a variable, with nothing else going on.

[fervent furnace]

property is method

[heady iris]

A property is made up of accessors (get and set) that call methods

https://learn.microsoft.com/en-us/dotnet/csharp/programming-guide/classes-and-structs/using-properties

[round violet]

so as soon as you add some get and set to a var it becomes a property ?

[heady iris]

I wouldn't say you "add" them

fields and properties are two completely different things

[round violet]

public int Foo {
  get {
    return Foo;
  }
}

what would this be ?

[heady iris]

a stack overflow error :p

It is a property.

Its get accessor returns Foo

[round violet]

why ?

[fervent furnace]

they are method so

int Foo(){
  return Foo();
}

[heady iris]

well, how does Foo decide what to return?

it reads Foo

and how does Foo decide what to return?

it reads Foo

[...]

[round violet]

but when not setting any get, isnt this the "default" behaviour ?

[heady iris]

what do you mean "not setting any get"?

you explicitly gave the Foo property a get accessor

[round violet]

when you just write sometype somename

[round violet]

ah okay i see

[heady iris]

int x;

This declares a field named x that stores an int

int x {
  get {
    return 123;
  }
}

This declares a property named x that only has a get accessor. that accessor returns 123

[leaden ice]

Circled thing just calls itelf

hence infinite recursion

[heady iris]

https://learn.microsoft.com/en-us/dotnet/csharp/programming-guide/classes-and-structs/properties

you can read about properties in the C# docs

https://learn.microsoft.com/en-us/dotnet/csharp/programming-guide/classes-and-structs/members

and about all of the various kinds of members here

[round violet]

so this is like making a readonly int var initialized at 123

[late lion]

Though there are similarities to how fields and properties are declared in code, that's just for consistency. A property isn't a "field with a get added to it", though it may look like it when you read the code. If the language designers wanted to, they could have made property declarations look completely different from field declarations, but that would be annoying to look at.

[heady iris]

It has a similar outcome, at least.

[round violet]

wait, the default set behaviour is still there ?

[heady iris]

No. There is no set accessor at all here.

[round violet]

since you only "overrided" the get ?

oh

[round violet]

i'll read this when going back home

[heady iris]

[System.Serializable]
public struct Vital
{
    public float value;
    public float maxValue;
    public float Percentage {
        readonly get => UnityEngine.Mathf.InverseLerp(0, maxValue, value);
        set => this.value = UnityEngine.Mathf.Lerp(0, maxValue, value);
    }
}

[round violet]

ty

[heady iris]

an example of a property

[late lion]

readonly get is going to be confusing without explanation

[heady iris]

the readonly modifier is there to promise that using Percentage's getter will have no effects

you can ignore that

I forget why I needed that.

[fervent furnace]

think them as method, they are just method

int a
public int A{
  get=>a;
  set=>a=value;
}
vs
int get_a(){
  return a;
}
void set_a(int value){
  a=value;
}
```if you dont declare them then they dont exist

[late lion]

Java has no properties, so most Java codebases are filled with get_x set_x methods.

[heady iris]

When I read Percentage, I use InverseLerp to calculate how far I am between 0 and maxValue

[round violet]

what is get and set called ?

[heady iris]

When I set Percentage, I use Lerp to calculate a value between 0 and maxValue

[late lion]

getter and setter methods

[heady iris]

if you're talking about the C# code, they're accessors

a property has one or more accessors in it; they're how you interact with the property

public int What {
    set => maxValue = 123;
}

[round violet]

so as soon as you set one of these, you loose the other default behaviour ?
is i add get, the default set would be lost and i would have to add it myslelf ?

[late lion]

There are no default getters and setters

[heady iris]

you can even do something like this, if you want

[heady iris]

You might be thinking about how you can do this..

        public int Automatic { get; set; }

If you don't provide an implementation, C# can automatically generate something for you

[heady iris]

this will create a field for you and read/write it

public int Automatic { get; private set; }

a slightly more interesting application

this means that the setter is private, even though the getter is public

[round violet]

i feel my brain starting to learn to much, but this is very intersting

[heady iris]

If you implement get, you can't have set implemented automatically, and vice-versa

(because the compiler can't be sure what you did in the other accessor)

[round violet]

ty everyone for answering my questions :)

[heady iris]

this would set maxValue to 123 whenever you assigned something to What

completely ignoring the value you tried to assign

i've never done this one :p

[round violet]

thats very intersting

that would avoid me creating a private and public var just to get it from outside

[round violet]

so What would never be updated ?

[fervent furnace]

public void set_a(int value){
  a=123;
}

[round violet]

i missread sorry

[heady iris]

Well there's the thing

There's nothing storing an int here at all!

We've literally just declared that we have a property called What

and that you can set it

public int Pointless {
  get { return 0; }
  set { } 
}

This does nothing and stores nothing.

It just returns 0 when you get it

There is no field at all.

[hard viper]

at that rate, just use a const

[heady iris]

yes, this is just for illustrative purposes

[round violet]

also, inside the set accessor, can you "get" the value we are trying to assign ?

example:

int a {
   set=>x*2

}

a = 4
// a is now 8

[hard viper]

value

[heady iris]

yeah, value

[hard viper]

is the name of the keyword

[fervent furnace]

using System;

public class HelloWorld{
    public static int a=10;
    public static int A{get=>a;set=>a=123;}
    public static void Main(string[] args){
        A=100000;
        Console.WriteLine (a);
    }
}

[round violet]

okay ty

[heady iris]

it's a special keyword that's only a keyword sometimes

[heady iris]

Notice how I had to write this.value up here in the setter

because value is the value that's being passed to the setter

[hard viper]

is there a reason when I make a ruletile that it won’t serialize a [SerializeField] private field from a generic base class?

[heady iris]

Can't think of one.

[round violet]

also, in csharp can you do something like the ... in lua ?

let me write an example

[late lion]

Is the class [Serializable]?

[heady iris]

for arbitrary numbers of arguments?

[round violet]

yes

like a Print

[heady iris]

public int LotsOfArgs(params int[] argsHere) { } 

[round violet]

so "params" is the keyword ?

[heady iris]

correct

[late lion]

With an array of some type

[heady iris]

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/keywords/method-parameters#params-modifier

[late lion]

And it must always be the last argument

[round violet]

only array, can't make it a list ?

[late lion]

Only array

[round violet]

okay

[heady iris]

It must be a single-dimendsional array.

[fervent furnace]

it is called variadic arguments

[latent latch]

usually if I find myself in a situation using params I eventually go back and pass by a struct

[round violet]

also, i discovered that a same method can accept different type of args and numbers

like Instantiate

what is the name of this and whta is the synthax ?

[latent latch]

overloading?

[round violet]

sorry it was Instantiate

overloading isnt for the non-obligatory args ?

[heady iris]

That's called overloading, yes.

[round violet]

okay

can you make one yourself ?

[heady iris]

The correct method is chosen based on the types of the arguments you provide.

Sure.

[fervent furnace]

same method name but different signature

[latent latch]

overloading isn't the word I've would of used naming this concept

[heady iris]

Unity isn't "special"; it isn't breaking any C# rules here

[round violet]

where can I find the synthax to make one ?

[fervent furnace]

https://learn.microsoft.com/en-us/dotnet/standard/design-guidelines/member-overloading

[round violet]

im sorry, i still dont understand what is a "signature" of a method

[round violet]

well the synthax is very simple :)

VS does all the hard work

[fervent furnace]

https://stackoverflow.com/questions/72788932/what-is-a-function-signature

[latent latch]

there's also operator overloading which allows us to do Vector3 + Vector3

[round violet]

could you explain a bit more ?

[fervent furnace]

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/operators/operator-overloading

[latent latch]

adding objects together is ambiguous, but if you define how the operator is handled between the objects then you can do that

[heady iris]

It allows you to define new ways to use existing operators.

public static Vital operator +(Vital left, float right)
{
    left.value += right;
    return left;
}

This lets me add a Vital and a float

producing a new Vital that I return

[round violet]

so a signature is like a unique identifier for all methods ?

[heady iris]

A function's signature is the type and number of arguments it receives, as well as the type it returns.

[latent latch]

prototype is the cooler word

[heady iris]

The combination of method name and signature must be unique.

[round violet]

i dont see the use case :/

[fervent furnace]

they can be identifier eg Java native interface

[heady iris]

Vital health;

public void Hurt(float amount) {
  health -= amount;
}

[round violet]

yeah ?

[heady iris]

health isn't just a float. it's a struct that includes the maximum possible value, and also has a property to get the current percentage

[fervent furnace]

vec3 + vec3 is an example

[heady iris]

I could have written a method to reduce the Vital's value instead

health.ChangeBy(-amount);

but I like the operator

Unity uses operator overloading to enable all of these operators.

[round violet]

its only for vectors ?

or is this a illustration image

[fervent furnace]

btw treat it as methods again

[latent latch]

Better example are matrix operations

[fervent furnace]

public static vec3 operator+(vec3 a,vec3 b)
vs 
vec3 add(vec3 a,vec3 b)<----you will see this if you coding in some language that not support operator overloading

[round violet]

where does the result of a+b sent ?

[fervent furnace]

the return value

[round violet]

okay

so you dont write {} or anything inside ?

[fervent furnace]

just example

[round violet]

okay

[fervent furnace]

the method body are the same eg a.x+=b.x.... then return a

[round violet]

lots of intersting stuff

i saw this a dotnet doc

public string? Day;

what is the ? for

[fervent furnace]

nullable

[round violet]

so Day can be a string or null ?

[fervent furnace]

a wrapper

public nullable<T>{
  public bool exists;
  public T theActualValue;
}

since value type cant be null you sometime need it to represent "no such value" (or use try get pattern)

[round violet]

i dont understand

[latent latch]

you can't ever compare a struct to see if it's set or not

[round violet]

what is a nullable ? if you cant set the var to null ?

[latent latch]

once you declare a struct, it's initialized with the values

don't really like this pattern anyway. It's better to just have a bool inside of the struct and see if it's set or not if you wanted

[fervent furnace]

eg
if i have a function that returns a int, how can i have an unique value that indicates fail?
no way to do this. (if it returns a class then i can return null)
here you can return nullable to make it looks like null that indicates fail

[round violet]

[fervent furnace]

treat it as an additional bool for checking that done by compiler magic

[round violet]

so it doesnt permit you to set to null a var ?

[fervent furnace]

value type cant be null, only reference and pointer type can be

[round violet]

okay

[latent latch]

Vector3 vector; //actually dont need to new it
can be interpreted as
Vector3 vector = Vector3.zero;
because the values are always defaulted to zero anyway

[heady iris]

In some languages, there is no null.

Instead, you have to explicitly handle something being "missing"

By default, C# does not behave like this.

You can freely assign null to any reference-typed variable.

This means that when you're passed a Widget, it could be an actual widget, or it could be null

[heady iris]

You can tell the compiler to behave more like this.

In that case, you are forbidden from assigning null to a Widget variable.

The upside is that, if you have a Widget, you know it's a valid reference to a widget

When you turn this mode on, you can use ? to explicitly say "this can be null"

Widget? is either a reference to a widget or a null

[heady iris]

You're thinking of nullable value types

[heady iris]

this was a nullable reference type

using ? with a value type makes it possible for the value type to actually be null (which is normally nonsense)

so:

• nullable reference type: used to allow a reference type to be null. you only need this if you actually tell the compiler to be strict about things being non-null
• nullable value types: used to allow a value type to be null. you always need this if you want to be able to have a null value type

I've never really used nullable value types, though.

[heady iris]

By default, Widget really means "Widget OR null"

which can be a surprise for people coming from more strict languages (:

[round violet]

👍

[rigid island]

IEnumerator CountDownItem()
{
    Application.targetFrameRate = -1;
    var randomIndex = Random.Range(0, items.Length);
    var index = randomIndex;
    var timeincrease = 0f;
    stopWatch.Start();

    while (timeincrease < 0.3f)
    {
        image.sprite = items[index];
        index = (index + 1) % items.Length;
        timeincrease += Time.unscaledDeltaTime;
        yield return new WaitForSeconds(timeincrease);
    }
    image.sprite = items[index];
    stopWatch.Stop();
    Ready();
}```
Is there a way to make this work also on low FPS? 
When I test in 10fps the timer ends too quick less than 1 sec thought time.detlaTime was framerate independent

[latent latch]

have ya tried fixed delta

[rigid island]

not yet, let me try that rq

[latent latch]

ill be real. A lot of stuff that I assume would be delta.time has not worked better than fixed delta and I'm honestly not sure anymore lmao

[heady iris]

if you're using unscaledDeltaTime, then you'll want WaitForSecondsRealtime

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

[mellow sigil]

Adding any kind of deltatime here doesn't make much sense, it's just the duration of the previous frame

[rigid island]

to be fair that was testing I also had regular time.deltaTime and did the same

[heady iris]

I am confused about the objective, yes

are you trying to make the loop take longer and longer over time?

[hard viper]

Time.deltaTime is framerate dependent

[heady iris]

That's going to be framerate-dependent because the amount you add per frame will vary

[latent latch]

oh right it's wait for seconds

[heady iris]

It's going to be similar to the wrong-lerp situation

imagine youre game is running at a trillion FPS

[rigid island]

I think my whole logic is prob wrong, I should prob explain my usecase

[heady iris]

timeIncrease will go up very very slowly, and the loop will run many many times

[latent latch]

waiting for seconds with frame time doesnt make much sense yeah

[heady iris]

obviously it'll also run those iterations faster

[rigid island]

Im trying to make this basically

[heady iris]

but you'll wait for a different total amount of time than if you had a low framerate

[rigid island]

picking random shape and make it visually appealing

[heady iris]

I would just write a function that turns a duration into an index

for example

public int FrameAt(float time) {
  return Mathf.FloorToInt(Mathf.Sqrt(time));
}

[hard viper]

the logic in your code is wrong

[mellow sigil]

It should be timeincrease += someConstant instead of deltatime

[heady iris]

[0,1] -> 0
[1,4] -> 1
[4,9] -> 2
etc.

calcaulate the index every frame and display the relevant shape

If you try to wait for a varying amount of time, it's going to be incredibly easy to be framerate-dependent

[hard viper]

I would do:
while (true)
increment index;
display shape at index;
yield return new WaitForSecondsRealtime(delay);

[rigid island]

wait I think this just fixed it..

[heady iris]

you know how you can approximate an integral with rectangles?

you're doing that

the width of the rectangles is depending on the framerate

[rigid island]

I thought about this initially but got confused quick with the time thing

[heady iris]

time would be the amount of time since you started

so, float startTime = Time.time; at the top of the coroutine, then just pass in Time.time - startTime

I would call this a "closed-form" solution, since it directly tells you which frame you're on at a given time

[hard viper]

I would probably store nextUpdateTime, to make it really framerate independent.
nextTime = startTime + first delay;
…
every update, nextTime += next delay

[heady iris]

instead of it being a product of exactly how long you've delayed each iteration of that loop

[hard viper]

then WaitForSecondsRealtime(nextTime - Time.time);

this way innaccuracies do not accumulate

basically, store the time at which the event happen, not the delta time

this usually makes a lot of math a lot easier

[heady iris]

yeah, kind of like how I do automatic weapons

float delay;
float cycleTime = 0.001f;

void Update() {
  delay -= Time.deltaTime;
  while (delay < 0) {
    Fire();
    delay += cycleTime;
  }
}

This correctly handles the fire rate being higher than the framerate

[hard viper]

i usually use this whenever buffered inputs get involved

store the time at which button was last pressed. Then to check jump, compare the time last pressed vs current time

it makes math and logic much simpler

[rigid island]

Trying to figure out all this to fit my coroutine

also this seems to be giving somewhat consistent results through diff fps |
timeincrease += 0.1f;

[hard viper]

nextUpdate = Time.unscaledTime + delay;
while true {
yield return new WaitForSecondsRealtime(nextUpdate - Time.unscaledTime);
do the thing;
nextUpdate += desiredDelay;
}

that’s it

[rigid island]

is nextupdate a local var?

[hard viper]

yes

[rigid island]

but while true instead? how does it stop itself?

[hard viper]

you could make it non local. Does not matter

[hard viper]

you could put other logic to know when to break

[fervent furnace]

wait until next timestamp

[hard viper]

but you only break if you want it to stop forever

[rigid island]

also what is Time.time + delay; delay

[hard viper]

that is the value of ingame time at the next time you do the update

but it’s a float, so we just need to know if our current time is beyond that timestamp

[heady iris]

the point of Loup's code is to account for the difference between the time you wanted and the time you got

[hard viper]

this way the schedule is preserved

[heady iris]

Suppose you're trying to wait for 10ms, but it actually takes 11ms

If you want to do something every 10ms, you shouldn't wait for another 10ms at this point.

You should wait for 9ms.

[hard viper]

exactly

[rigid island]

Ok I will try this out thanks! hopefuly my brain can process this rn xD

is this how online slot machines really do their logic

dang

[hard viper]

it’s not rocket science tbh. you just calculate the time you actually want something to happen, and check if you passed that time yet

[heady iris]

note that no amount of fiddling will make this idea work: as long as you wait for timeIncrease seconds, you will be framerate dependent

the faster the game is running, the more times you'll have to run that loop to reach 0.3

you'll wait for 0.01, 0.02, 0.03, 0.04, ..., 0.28, 0.29 seconds

vs. a low framerate, which will wait for 0.1, 0.2 seonds

[hard viper]

if you want extra safety, you want to check that (nextTime > Time.time) before you go to next loop iteration. This is in case you actually need to skip a whole evaluation

[heady iris]

(well, and 0.3 seconds)

[rigid island]

what about using a constant like Nitku suggested

[hard viper]

but I would assert this, because if this is not true, then that means you’re actually needing to skip, or are super falling behind

[rigid island]

seems its giving me roughly same timing for both fps changes

[fervent furnace]

timer.time+=dt;
while(timer.time>=timer.wait){
  timer.time-=timer.wait;
  timer.wait+=some more delay if you want
}
```calculate next timestamp is computational faster, idea of this is the same as Fen's one

[rigid island]

doesn't have to be pin point accurate if its 0.1 off or somehing

[heady iris]

You'll have errors if a frame takes longer than the delay.

[hard viper]

the code I wrote is the most robust method

if you run the game on two different machines, they will desync

[rigid island]

Ok I will try that out, I'm trying to digest all the math xD

[heady iris]

I would, personally, just write a closed-form solution and not try to delay for varying amounts of time

[rigid island]

how would that work ?

[hard viper]

right now, simple is better, fen. the man is confused

[rigid island]

very xD

[hard viper]

he means tabulating the delays

which is an extra layer of calculation

[rigid island]

ah ye my monkee brain can't handle that rn

[heady iris]

X axis is time. Y axis is which index you're at.

Plug in the current time to get a number.

[hard viper]

i want a gif of a man panicking at a blackboard right now

[heady iris]

If the number differs from the last frame's number, you've advanced one or more steps; play a sound or whatever

[rigid island]

[fervent furnace]

this should be the best or simplest way indeed, store the start time and input the dt into that function output the index for you

[heady iris]

while (true) {
  int index = GetIndex(Time.time - startTime);
  yield return null;
}

This is completely framerate independent. It will produce exactly the correct index at any given time, no matter what the framerate is.

No funny tricks required.

You just have to choose an appropriate implementation of GetIndex

The example above is the 0.7th root of the time

You can also take the logarithm of the time to get a differently-shaped curve

You could even make an AnimationCurve if you wanted to

[rigid island]

Evaluate curve would go in GetIndex?

[heady iris]

Right.

The steeper the curve, the faster the index changes

you could even remember the last frame's index

while (true) {
  int newIndex = GetIndex(Time.time - startTime);

  while (index < newIndex) {
    // make a noise idk
    ++index;
  }

  yield return null;
}

so if you went from index 0 to index 3 in one frame, you could still know you took 3 steps

and thus play 3 sounds, or spawn 3 particles, or whatever

[fervent furnace]

some curve with second derivative <0 eg concave downward for ease out effect

[heady iris]

(fixed; i had the wrong logic in the loop)

[heady iris]

this is more-or-less exactly what my shooting logic does #archived-code-general message

both of them correctly deal with unusual situations: many events per frame, 0 events per frame, wildly varying numbers of events per frame

[hard viper]

ok, so here, black is a timeline, with black ticks for the Time.time when you actually have your frame. The red-blue bars at top correspond to what color we want at a given time. The cyan dotted lines are the actual times where we want to change. The black ticks with red/blue stars are frames where we change color.

In the method I described, nextUpdate is the value of the next cyan dotted line. This is the next time where we actually want to change.

[rigid island]

Ahh I see, that makes a bit more sense to me

[hard viper]

Every time we change, we calculate the next one. We also wait for an amount of time between now (the black tick we are at) and the next cyan line (nextUpdate).

So once the cyan line passes, we try to change on the very next frame (black tick)

[rigid island]

Ok I will these outs and hopefully get it going lol
thanks all for the very informative responses !

[hard viper]

just as a heads up for this, you want a contingency plan if your FPS is super low, and you miss a whole block. eg there are no frames when we’re scheduled to be blue

i would do this by just making:
if (nextTime > Time.time) yield return new WaitFor…

so if you fell behind, you don’t actually wait, and skip straight to the next step

[rigid island]

without a yield wont the loop not end ?

[fervent furnace]

it should be nextTime>=t.t?

[hard viper]

right, fixed

[hard viper]

it will keep looping, calculating the next timestamp, until the next time is actually in the future

[heady iris]

if your delay is very low, you might need to run the loop multiple times per frame

[heady iris]

same idea

[fervent furnace]

while(true){
  if(nextT>=T.t)wait nextT-T.t //you need to wait
  do your stuff
  nextT+=delay;
}

[heady iris]

If you unconditionally wait until the next frame, you will be framerate-dependent

because you will do less work than you should when the framerate is too high

[swift falcon]

is it a good idea to use try{} in the actual final game code and not just debug?

[hard viper]

this does mean other things, like in that case, you actually displayed blue for zero frames. But that is more an issue of you tuning for super low framerate

[heady iris]

Assuming you're not causing exceptions for silly reasons, yes.

[hard viper]

some exceptions are meant to be caught. some are not.

[heady iris]

Exceptions are unusual, but manageable, problems. It's reasonable to handle them in your code.

[knotty sun]

not really. no. lots of unnecessary overhead which your playtesting should have taken out the need for

[leaden ice]

depends on the circumstance. Putting it around a network or file access is good. Putting it to prevent a NullReferenceException or ArrayArgumentoutOfRange exception is bad

[heady iris]

for example, I catch exceptions coming from Json.NET and handle the failure to parse

[hard viper]

if you have a corrupted save, you want to throw a CorruptedSaveException. then catch it to display an error message, while cancelling the level load.

[swift falcon]

in my case im still trying to access a deleted object, so it would be easier to use try{} in this case

[heady iris]

No, you should just fix your code.

[knotty sun]

no

[heady iris]

You can test if the object is destroyed before trying to use it.

[hard viper]

fix your shit

[swift falcon]

thats what im trying to figure out

[heady iris]

any Unity object can be compared to null

if you get "true", it's an invalid object

[swift falcon]

if i check if this.gameObject != null it says that the if is doing the error

[heady iris]

that means this is a destroyed unity object

[swift falcon]

yea but i get an error

[knotty sun]

you should never use try catch as an option for 'I cant be arsed to fix this'

[heady iris]

Show us your code. I can't guess what you're trying to do.

[analog axle]

then this is null

if this.gameObject is throwing an error, doesnt that mean this is null

[heady iris]

It sounds like you're calling a method on a MonoBehaviour that's been destroyed

[analog axle]

and has no gameObject?

[swift falcon]

ignore the current if, i was just searching online alternatives

[heady iris]

well, it's normally not really possible for this to be null

you'd get an error trying to call the method in the first place

[analog axle]

thats what I assume but whats the other possibility

[heady iris]

But if this is a unity object, it will equal null when it's destroyed

[knotty sun]

?? ??????????????????

[heady iris]

and you'll get an error if you try to access any Unity property

[analog axle]

so yeah the gameobject is destroyed then

[heady iris]

this is very weird code

[leaden ice]

if (!this)```
or
```cs
if (this != null)```

[heady iris]

I should point that gameObject cannot be null on a valid MonoBehaviour

[knotty sun]

never use ? or ?? with Unity Objects

[heady iris]

every MonoBehaviour must be attached to a game object

[hard viper]

exceptions are basically almost exclusively intended in final product for things that can forseeably happen after it is in a customer’s hands.
Like seatbelts in a car. Car should not crash, but it can, and then throw a seatbelt exception.

What you have is more like an assertion, which is checks at the factory. Eg checking if the car has an engine. Because a car without an engine shouldn’t even reach a driver.

[heady iris]

hang on aren't you just doing this

Destroy(gameObject, 1);

I guess you're turning off the box collider first.

[swift falcon]

why?

[hard viper]

if you are using try catch to handle an exception that routinely happens in your code, it is like letting your car get totalled whenever you turn it on, and then trying to get it fixed daily

[heady iris]

• why is this async?
• ?? only cares if the reference is literally null; Unity objects equal null when destroyed, which you'll miss entirely
• it's impossible for this.gameObject == null if this != null

it's just..really weird looking

[hard viper]

exceptions should only happen when something goes very wrong

[swift falcon]

i wonder

[heady iris]

if you want to destroy an object after a few seconds, I'd just write a coroutine

[knotty sun]

no

[swift falcon]

async because waiting would freeze the game otherwise

[hard viper]

what

[analog axle]

coroutiine

[analog axle]

async isnt for something like this

[heady iris]

unity coroutines make this really easy, yes

[hard viper]

Co-routine. routine that runs co to your routines.

[south violet]

Can I somehow control what instanceId values will have my objects? I tried to do that with Random.InitState(), but it didnt work for these, even if these objects are created after I use Random.InitState()

[knotty sun]

!this is meaningless. this.gameObject != null is usefull

[leaden ice]

this.gameObject gives a MissingReferenceException if this was destroyed

[swift falcon]

alright, removed async and no errors

final ver

[heady iris]

this is correct for a non-MonoBehaviour

[knotty sun]

this cannot be destroyed, otherwise it would not be running

[leaden ice]

not true

[primal wind]

Task.Delay won't do anything without async

[heady iris]

(well, !this is meaningful for any unity object)

[swift falcon]

it does

[leaden ice]

very easy for it to have been destroyed.

Destroy(myComponent);
yield return null;
myComponent.SomeFunction();```
inside `SomeFunction`, `this` now refers to a destroyed object.

[swift falcon]

!this.gameObject ?

[knotty sun]

fair, I had not envisaged such stupidity

[heady iris]

indeed. this isn't literally null, but it's a destroyed unity object, so it will equal null

[analog axle]

[[current topic]] is very humorous.

[heady iris]

just write a coroutine

[swift falcon]

im not familiar with that

[analog axle]

https://docs.unity3d.com/Manual/Coroutines.html

[heady iris]

IEnumerator DestroySelf() {
  yield return new WaitForSeconds(4f);
  boxCollider.enabled = false;
  yield return new WaitForSeconds(1f);
  Destroy(gameObject);
}

done

[analog axle]

#archived-code-general message

[swift falcon]

ty

[heady iris]

If the monobehaviour dies, the coroutine stops, so you don't have to reason about that

[analog axle]

and then u do the ol' StartCoroutine(blah());

[hard viper]

I find this too easy for my use case. Could you perhaps add a couple of loops and variables? Maybe muck it up a bit?

[heady iris]

I'll add a blockchain AI

gimme a sec

[analog axle]

add 5 counters to count the frames

[hard viper]

last time I asked an AI to code, I asked it to write hello world using only emancipated code using the Underground Railroad pattern

ChatGPT refused because it makes light of slavery. So I asked it to tell me a story in the tone of my racist uncle, who only speaks in C# code, and really wants to talk about ….

[swift falcon]

ok works

[hard viper]

“Sure thing, bud”

[swift falcon]

whats the benefit in using corrutines

[analog axle]

👍

[hard viper]

AI is easily swindled

[heady iris]

Coroutines let you write code that's executed over several frames without having to think about correctly using async

[analog axle]

it was actually designed for your usecase

[hard viper]

easy to time relative to specific parts of Unity’s main execution of event functions

[heady iris]

you write an enumerator. Unity asks the enumerator for new values every frame (by default)

[swift falcon]

good

[heady iris]

So you aren't just running code..whenever

[primal wind]

can't wait for awaitable

[heady iris]

I can wait

[audience laughing]

[rigid island]

Ok I think I missed out something.. how Do I slow this thing down again lol

 IEnumerator ImageRoulette()
 {
     //Application.targetFrameRate = 10;
     var randomIndex = Random.Range(0, items.Length);
     var index = randomIndex;
     while (true)
     {
         if (nextTime >= Time.time) yield return new WaitForSecondsRealtime(nextTime - Time.time);
         image.sprite = items[index];
         index = (index + 1) % items.Length;
         Debug.Log("Something Happening");
         nextTime += delay;
     }
 }```
The good news it looks the same speed across diff framerates

[fervent furnace]

isnt your delay should grow to have an ease-out effect

[rigid island]

ahh ok so delay should not be constant ?

makes sense

so delay can be AnimationCurve.Evaluate ?

[heady iris]

If your delay starts to vary based on the current time, then you're headed straight back to framerate-dependent town

because the amount of time you wait -- and thus the time until you change the delay again -- will now depend on the exact time

Although, actually...

As long as you're never actually using the current time to decide how long to wait for, you'll be fine

If you did, then the exact amount of time the frames took would matter

But if you just adjust delay by a constant amount each time, it'll be fine.

(or adjust it based on index, or anything else that is not the current time)

[onyx marten]

where could i get newest release for facepunch.steamworks? the latest on reddit has some issues critical to me
are there alternatives to facepunch?

[rigid island]

so I cannot use curve because of Time.deltaTime right? a bit confused on that part. How would I evaluate without?

[crimson spoke]

hello humans

[leaden ice]

https://wiki.facepunch.com/steamworks/Installing_For_Unity

[heady iris]

sum up delay each loop to keep track of how far into the animation you are

[onyx marten]

as i said in my message, the newest release there is outdated-

[heady iris]

you said it has "issues"

we can't really help you based on that

[heady iris]

imagine doing this without it being a coroutine at all

[fervent furnace]

there is not many can do to deal with low frame rate indeed since this is animation
in extremely case delay=whole animation time you sprite just change one and no animation at all, the current implement is compensated for low frame rate (dont wait if nexttime>=Time.time)

[onyx marten]

its a bug which was fixed after the date of that release

[heady iris]

just writing a loop that prints out what time each transition should happen at

this loop wouldn't care about in-game time at all. it'd just go...

float duration = 0;
while (duration < 3) {
  Debug.Log(duration);
  float delay = GetDelay(duration);
  duration += delay;
}

[rigid island]

what would GetDelay do?

[heady iris]

get the current delay. you could also just modify a delay value

float duration = 0;
float delay = 0.1f;
while (duration < 3) {
  Debug.Log(duration);
  delay += 0.1f;
  duration += delay;
}

You asked about using an animation curve though

[heady iris]

hence doing this

We want to calculate each delay the exact same way no matter what the framerate is

[rigid island]

ohh so basically what I have now but without the yield?

[heady iris]

Exactly.

[rigid island]

I put this in and it was working somewha cs while (true) { if (nextTime >= Time.time) yield return new WaitForSecondsRealtime(nextTime - Time.time); image.sprite = items[index]; index = (index + 1) % items.Length; Debug.Log("Something Happening"); nextTime += delay; delay += 0.001f; }

[heady iris]

You'd just add a yield that makes you wait until Time.time + startTime > duration, meaning you're free to run the loop again

As long as delay isn't calculated based on the current time, this will be framerate independent

It's a bog-standard loop that counts up towards a target.

You simply pause that loop whenever it gets ahead of the current time.

[onyx marten]

the old facepunch steamworks that is in the build had issues with withmetada for community items, which as i can see in github was fixed later on, is the only way to just build it myself?

[rigid island]

confused a bit on this part , how would I yield that ? how to get startTime ?

[heady iris]

oh, that formula was bogus

Time.time > startTime + duration

startTime would just be the time when you started the coroutine

[heady iris]

you're already doing this, really

it's just phrased differently

[rigid island]

ahh ok

[heady iris]

Having duration separate makes it easier if you need to calculate a delay based on your progress

like if you want to evaluate an animation curve

[rocky jackal]

if i have an object with a script that saves the render settings it will take the ones from whatever scene it is in ? even if its loaded additivley ?

[analog axle]

what's the reason why burst doesn't allow you to typeof() a physicscollider ecs component???

[leaden ice]

it will take the ones from whatever scene it is in
What kind of settings are we talking about? I guess it would depend on the details of the script.

[rocky jackal]

just some basic things, i want to take the environment settings from whatever scene is loaded additivley and apply them to my main scene

[rigid island]

Works good!
I think I will leave it here until I figure out the AnimationCurve part for smoothing, but seems like now the results are pretty much the same across multiple fps
Thank you very much for your help & patience @heady iris @hard viper @fervent furnace
here is code I used (ofc will fix the magic numbers later :p)
https://hatebin.com/trgnzyiedq

[leaden ice]

it will take the ones from the active scene. Active scene has a very specific definition and can be changed with https://docs.unity3d.com/ScriptReference/SceneManagement.SceneManager.SetActiveScene.html

Also see https://docs.unity3d.com/ScriptReference/SceneManagement.SceneManager.GetActiveScene.html

[rocky jackal]

thank you

[cerulean mist]

        while(elapsedTime < duration)
        {
            elapsedTime += Time.deltaTime;
            curveValue = _speedOverTime.Evaluate(elapsedTime / duration);

            speedFactor *= curveValue;
            yield return null;
        }```

Quick question, i'm using a coroutine to evaluate a curve (a curve that will always be 0 to 1), but the duration may vary (for example the duration might be 5 seconds). How to I evaluate the curve over duration even though the curve is only length 1?

[heady iris]

that code looks fine to me

if the animation curve ranges from t=0 to t=1, at least

[cerulean mist]

it's not working as expected. the length of the curve and duration are different values

oh wait i think speed factor might be throwing me off lol

[heady iris]

this will set speedFactor to zero as soon as the curve evaluates to 0

and then it'll never come back

[cerulean mist]

That's fine, that's what I want

I want the speed of an object overtime to decrease based on a curve

[heady iris]

oh, 0 to 1, not 0 or 1

well, this will still be wrong

it's framerate dependent

if your framerate is very high, you'll multiply speedFactor by curveValue more often

[cerulean mist]

ahh i see

so curve value needs to be multiplied by Time.deltaTime

[heady iris]

that'll make you slow down even faster

You'd want to change speedFactor less when deltaTime is small

I'm not sure what a valid way to do that would be

I guess you could do something like

speedFactor -= speedFactor * Time.deltaTime * curveValue;

but that's still not really valid

notably it'll make speedFactor go negative if deltaTime is greater than 1

which is very unlikely, but it tells us that this is mathematically invalid

[leaden ice]

This problem is why the physics engine uses a fixed time step

[heady iris]

(is this even physics-related?)

Ah, I bet this is going to be a lot like https://unity.huh.how/lerp/wrong-lerp

some kind of exponential function will do the trick

[leaden ice]

Whatever it is it's a simulation with a first order variable (position) and based on a changing rate over time (speed) and to make that stable you need a fixed time step

[heady iris]

Yeah, this is identical to:

speedFactor = Mathf.Lerp(speedFactor, 0, 1 - curveValue);

So the same fix will apply

speedFactor = Mathf.Lerp(speedFactor, 0, 1 - Mathf.Pow(1 - curveValue, Time.deltaTime));

[cerulean mist]

Oh, that does the trick. Thanks!

[heady iris]

The page here explains roughly what's going on

Normally, one long frame will incorrectly move you further than two short frames

[cerulean mist]

This is a very handy website, thank you for linking this 🙌

[heady iris]

because you do a BIG step with the original large value

The exponential function fixes that by telling you how much you'd slow down if you had an very large number of very short frames

it's like continuously compounding interest

the magic e constant and all that fun stuff

[heady iris]

it's very handy (:

[leaden ice]

read the editor logs

[swift falcon]

Where do I find them?

[hidden flicker]

This code will snap the secondary object to seemingly completely random positions. This is intended to lock the distance between two objects.
https://hastebin.com/share/eqebaxewem.csharp

[leaden ice]

rootObj.position - obj.position is an offset/direciton vector

you are trying to use it as a position
What are you trying to do exactly?

[heady iris]

!logs

[tawny elk]

📝 Logs

Documentation

Editor logs

Windows: %LOCALAPPDATA%\Unity\Editor\Editor.log
MacOS: ~/Library/Logs/Unity/Editor.log
Linux: ~/.config/unity3d/Editor.log

Unity Hub

Windows: %UserProfile%\AppData\Roaming\UnityHub\logs
Mac: ~/Library/Application support/UnityHub/logs
Linux: ~/.config/UnityHub/logs

[hidden flicker]

Lock the distance between two objects. If a childobj is more than maxDistance away, prevent it from moving further

[leaden ice]

Simple:

obj.position = Vector3.MoveTowards(rootObj.position, obj.position, maxDistance);```

Or if they're all direct children of the "root object" you can do:

obj.localPosition = Vector3.ClampMagnitude(obj.localPosition, maxDistance);```

[hidden flicker]

Thank you, solved

[cerulean mist]

    private IEnumerator EvaluateSpeedOverTime()
    {
        float elapsedTime = 0.0f;
        float duration = _useLifetime ? _lifeTime : _speedCurveEvaluationTime;
        float curveValue = _speedOverTime.Evaluate(0.0f);

        while(elapsedTime < duration)
        {
            elapsedTime += Time.deltaTime;
            curveValue = _speedOverTime.Evaluate(elapsedTime / duration);

            speedFactor = Mathf.Lerp(speedFactor, 0, 1 - Mathf.Pow(1 - curveValue, Time.deltaTime));
            yield return null;
        }

        speedFactor = _speedOverTime.Evaluate(1);

        yield return null;
    }```

Actually I don't know if this is working correctly, it starts lerping correctly and then suddenly snaps to _speedOverTime.Evaluate(1);

[heady iris]

well, yeah, you..told it to do that at the end

[heady iris]

The point of this is to gradually reduce speedFactor based on the current value of curveValue

It's a continuous process.

The lower curveValue is, the faster speedFactor goes down

[cerulean mist]

well yes but isn't that last line makign sure it doesnt overshoot?

[heady iris]

no, it's just setting speedFactor to whatever value the curve has at t=1

but the curve doesn't represent a speed at all

it's a rate of change of your speed

My understanding was that you wanted to make speedFactor go down based on the value of the curve.

If you just want to set the speedFactor to equal the curve's value, just do that normally

[hidden flicker]

How would I prevent my player from losing speed when they enter a slope?

[heady iris]

otherwise, I don't understand what you want, and you need to explain it

[hidden flicker]

Here is the worst diagram ever made, for reference

[cerulean mist]

No you're right I think we're on the same page. I understand its a rate of change of speed. The reason I put it there is because the ending curve value will be 0

I out it there because I wanted to make sure it didn't overshoot

[heady iris]

I'm not sure what you mean by "overshoot"

as in, the speed goes negative?

[cerulean mist]

correct

[heady iris]

that's impossible

the code moves towards a speed of 0

[rigid island]

ok spoke too soon.
This works well on the first run, but in consecutive runs it becomes shorter in wait.
I'm definitely overlooking something obvious .

[heady iris]

each frame, it lerps from the current speed towards 0

the amount it lerps depends on the curve value and on deltaTime

[heady iris]

you forgot to reset something

[rigid island]

yeah thought it was the nextTime but that made it worse xD

[heady iris]

You can compute a rotation between the old normal vector and the new normal vector

Quaternion.FromToRotation(oldNormal, newNormal);

then multiply your velocity by that to rotate it

[hidden flicker]

I understand what you're saying, but that's not quite the issue

[cerulean mist]

I see, my mistake

[hidden flicker]

My move direction does get rotated when I enter a slope

It's a sudden jolt that results from essentially banging your knee temporarily on the slope

[heady iris]

right, and this would redirect your velocity so that it keeps its magnitude, but is now pointing uphill

Although, I'm not sure how I'd do this with a rigidbody. I am imagining a character controller here.

I guess you'd just need to remember what your velocity was before you banged into the slope

[hidden flicker]

Alr, thank you

[round violet]

Can you add an offset to a additive scene ?

or get all objects from this scene

[heady iris]

A simpler option would be to just notice when your velocity goes down suddenly and set its magnitude back to what it was before.

I'd try that first!

[leaden ice]

https://docs.unity3d.com/ScriptReference/SceneManagement.Scene.GetRootGameObjects.html

[round violet]

thanks man

[rigid island]

Ah I just needed nextTime = Time.time in the Coroutine start

its .5ms off but whatevs

[round violet]

how do i get my "scene reference" so I can get the rooted objects from a specific scene ?

[leaden ice]

From various methods in https://docs.unity3d.com/ScriptReference/SceneManagement.SceneManager.html