archived-code-advanced

[shell grove]

yield return new WaitUntil(() => Master.time / (notes[btn].Frequency / 44000f) == 0f );

that kinda works

but it doesn't

If only I was good at math

[sly grove]

Why check for equality

Just do a > or < check

[polar ermine]

i dunno where this belongs on the code difficulty scale but.
i need to display some "status cards" in a grid, based on what status the player has accrued. (poisoned, flying, injured, etc).
a status card consists of some tooltip text, and a sprite.
the grid starts empty, but i need to add and remove such a card from the grid at any point

my question is, i suppose, what's the best infrastructure to set this up? will i need a scriptable object for each possible status? how do i reference those dynamically to instantiate them into the grid in a way to easily create and apply new statuses as i develop?

[cursive horizon]

depending on how they are used, i might just have a hashset of status enums somewhere that tracks 'active statuses', and in your UI you read it and display the appropriate icons (which would be a prefab added to a parent with a grid layout group, most likely)

if you want to use SOs you can do that too though, same sort of idea

[polar ermine]

i already am keeping track of a few statuses as enums, i do just need a way to associate a status with a sprite, and have some sort of reference to all existing status associations to instantiate them into the grid.

i suppose im having trouble nailing down the specifics of that setup, as i'd prefer a flexible solution that involves as little boilerplate set up for each time i want to add a new status the game might use.

[cursive horizon]

Well it depends if you want to mix the presentation with the logic. If you don't mind that, I would say just make an SO for each status instead of an enum and include presentation details there (like the sprite, display name, etc). Then you can have another SO which holds the 'currently available statuses' and put whichever ones you want active in the game into there, and assign that reference to whatever needs access to the statuses.

If you want to keep presentation and logic split, I would keep the statuses as enums and just make a 'StatusDisplay' class, and then make an SO with a dictionary of status->StatusDisplay which you reference in your view component. Unity doesn't serialize dictionaries out of the box so that approach is easier if you have Odin, but you can work around that limitation if you want to.

I tend to like the second approach, but I also use Odin which makes it pretty simple. One advantage there is that if you want to sometimes display statuses differently (like maybe there's a 'status glossary' which uses fancier sprites and has a more in depth description). Associating the view data with the status directly can get confusing if you have multiple similar presentations of the same value

[polar ermine]

no odin for me, but i ought to look into it.
having a dictionary of statuses isn't too bad, i think. though my experience is limited, so lmk if my brainstorming is flawed.
a scriptable object (statuses?) that hold a number of statuses (say a serializable status class) which in there I can hold references to each statuses individual sprites and information?
something like (pseudo code)

public class Statuses: ScriptableObject{
  public Status[] status;
}
...
[Serializable]
public class Status{
  public string name;
  public string desc;
  public Sprite sprite;
}

[cursive horizon]

yep, that would be the first approach i proposed

[polar ermine]

my thought here is that in the inspector i can just add a new entry into the statuses and apply their things once, then reference them elsewhere.

[cursive horizon]

right, and to add a new status you jut have to create a new SO instance and add it to the list

which fits you 'minimal boilerplate' goal

for the status list, I have a class like:

public class ConfigSet<T> : ScriptableObject {
  public T[] items;
}

and then i extend that to create lists of SOs which i can plug and play, cause this pattern ends up being really common

also it has some helpers on it like GetRandom

[polar ermine]

hm, alright, awesome, that's super useful
and to reference a particular status, im gonna simply need to have a reference to that Statuses SO on my player, right?

[cursive horizon]

i donno what your game is, but that seems reasonable

one of the nice things about SOs is that you can reference them from a few places if you need to, just don't go too crazy because they are sorta 'globals'

but if you had the player who needs to know about them, and then a ui panel or something, they could just both reference your Statuses list and that's handy

[polar ermine]

for sure for sure.
it's a card game, and each card can have certain statuses applied to them. i'm already using SO's to generate card information and visuals and instantiate multiple cards based off of a single card template prefab, so it's pretty cool that all i need to do is just add one reference to the template's script

[cursive horizon]

Ah, well I would expect statuses in a card game to potentially have very different effects, so I wouldn't try to encapsulate them. I'd probably have some systems whose job it is to handle the effects for whatever status and keep the statuses a 'dumb' enums. In that case, you might not actually need a list of 'all enums' referenced anywhere, because each system would only care about the one(s) it needs to interact with (and can just check if they are active on the player and do nothing if not)

depends on the game and what sort of stuff is happening though

and that approach does require more boilerplate because each status is handled explicitly (but in my experience, that's actually what I want)

so your damage system is actually what would query and apply logic for the 'double damage' status

the advantage there is that you keep all of your 'damage logic' in one place, but it can get complex if you have a lot of different statuses

still, having that complexity be explicit is better than hiding it in a bunch of objects with OnApply or whatever (imo)

[polar ermine]

yeah; im probably currently doing it pretty suboptimally, but for the one status i have built up it's simply an enum (flying or grounded), and im already using that enum elsewhere to determine what happens if they get attacked depending on which effect is applied to them

i just wanted a way to link that to a visual for the player, but also i can definitely see it getting more out of hand as i scale the mechanics, as it becomes a case of checking whether or not a status is applied and applying logic based on that for each status

[cursive horizon]

yeah, it depends on how many statuses you will have, how similar their effects are, how many mechanics your game has that they can interact with, etc

but honestly that sounds fine to me...if you have a lot of statuses and they might be interacting with each other because they all affect 'damage', you might want to force yourself to make all of that explicit and in one place

so if you already have that in place and just want to map the statuses to visuals, the dictionary approach sounds good to me

there are definitely free extensions that will serialize it for you without odin but i can't speak to them

i feel like i'm leading you in circles 😛

[polar ermine]

lmao

it's fine, i definitely got a good starting point to at least try

im relatively new to the whole unity thing, so i don't mind doing things myself until i'm comfortable enough to use something like odin to just save time

[cursive horizon]

that was how i felt for a few years, but once I finally did buy it I wished I had much earlier

especially for a card game where you want to be able to 'configure' complex cards, i can't imagine building tooling for that in the editor without odin's ability to serialize interfaces

well i can imagine it, it just sounds like a big waste of time

But yeah, it's always good to run into the problems that tools solve before you reach for them. Just know that odin is dope and will be there for you 😄

[polar ermine]

haha, fair enough. i'll give it a look sooner rather than later. currently there isn't too much complexity that i've needed any third party editor tools, but i know that can change quick

odin: noted ✅

[cursive horizon]

e.g. in the card game i'm working on, my cards are just an array of ICardComponents which odin kindly serializes for me:

you can do that in unity now with [SerializeReference] but I'm not sure what gotchas it comes with

[polar ermine]

ooh, actually... i think i need something for something similar.

i have a lot of card components in cards that don't need them and it's gonna be a mess when i expand it further

[cursive horizon]

yeah, eventually you end up with a lot of 'potential complexity' and it's hard to tell how much is active on any given card

[polar ermine]

like, i have a card ability that heals; i don't need components in there to tell me how it affects the card's movement status

or what sort of damage it does

healing damage

[cursive horizon]

yeah, and it's a similar thing for statuses imo

you could try to encapsulate them and have all the possible effects on each status, but I think this way is simpler

i spent a bunch of time trying to work around the default serialization ages ago, so doing this is great

that said, it's still useful to split your 'editor config' from your 'runtime' config, so you can kinda do both

like the 'card editor' might want to have more context for each possible thing that a card can do and show/hide sections, basically just be smarter than 'dropdown with every option in it'

but ultimately i do like the runtime version being 'pure data component' based for something like this

[polar ermine]

hm, yeah, i get it a bit better now. definitely sounds useful

i've been trying to simplify it with editor scripts but it wasn't worth the effort.

[cursive horizon]

Totally. For me Odin bridges that gap where I wish the editor could handle more, but not enough for it to be worth setting up anything custom. And then when you do want something custom, you can use the odin pieces to whip it up really quick.

I swear I'm not an Odin salesperson 😅

[polar ermine]

well you're selling me on it!

[cursive horizon]

well apparently it's half off right now...

[polar ermine]

oh shoot

[cursive horizon]

ey you got 20 days!

[polar ermine]

what a bargain

hopefully enough to get my paycheck in but it's a pretty good deal anyhow

[fierce holly]

i took a slightly different approach, but what you guys suggested definitely helped me find new solutions and ideas. I was pretty stuck on this, so thanks!

don't mind the layer mask, i put it to "water" for testing purposes because i didn't know how to add other layer masks immediately

[fierce holly]

i did make use of the layer mask as you suggested!

[undone coral]

looks good

[undone coral]

there must be dozens of people making card games

the best thing, in this exact case and nothing else, is to do it in code with a giant switch statement

because expressing string formatting is easiest in code

you should have a Tooltip class with a static string[] Tooltips(Entity entity) { ... } method that generates all the tooltips

trust me, i thought about this a lot, and in my own card game, this was the best way

everything else you could do, you'll wind up reinventing string formatting

specifically the part where it's easy in code to just... get a thing

whereas in a string field, it's hard

you'll do a lot of

if (entity.HasStatus(Status.BLAH)) { 
  tooltips.Add($"You are suffering from {entity.GetValue(Values.BLAH)} blah"))
 }

you don't want to create a DSL for entity.GetValue or whatever

that's just reinventing C#

don't do that

"what if i have a sprite icon i want to show" return it in the string in a way that's compatible with textmesh pro

all this stuff you want to do with text, do it in code

my hot take 🎤

[polar ermine]

I actually do have a tooltip class/system that I was using for something else, and I was able to use it for this as well.

Though I worry more about my logic system, cause it’s going to get really messy if I don’t learn how to split things up better.

[undone coral]

i don't think you'll earn anything by "splitting" this up

there is exactly one UI element that needs the tooltip information

that's some tooltip element

you can make a single static method that deals with generating the tooltips for anything

everything else, you'll just wind up repeating yourself for no gain

there's no reason to have a subclass of tooltip, and then make singleton tooltip instances, etc. etc.

it's just a ton of ceremony for no gain

you don't want the card DSL to be aware of how it affects the tooltips

you don't want to encode that there

it's too unwieldy

just put it all in one place

a lot of card games have way too many tooltips

because people try to be so pure about it

you will inevitably want to customize the tooltips everywhere they appear, so a generic method won't do much for you

[polar ermine]

Yeah I’m using a static tooltip class courtesy of a great YouTube vid I found. It’s been useful thus far.

[undone coral]

okay great

unless your game needs to support card design at runtime

it's almost always better to have just encoded card behavior as C# code the way XMage does

[polar ermine]

The only card design at runtime I do is setting the sprite and a few details, based on which card is drawn. I’ve one card prefab template and I use SO details on instantiating to fill out the details.

Maybe not perfect but it’s given me the least headache once I set it up

[undone coral]

i would suggest something of the form

class Fireball : AbstractBaseCard {
 override IEnumerator Play() {
  var selectTarget = new SelectTarget(TargetSelection.ALL);
  // yields when the player
  // selects the target
  yield return selectTarget;
  var target = selectTarget.target;
  // plays the damage animation
  // deals the damage
  // it's a coroutine
  yield return Game.instance.DealDamage(5, target);
 }
}

@polar ermine

[polar ermine]

Thanks! Saved that for reference. Currently I’m making individual ‘actions’ for any given card, and the attack method reads the action parameter and uses the necessary variables like damage and damage type. An extensible class would be better

[undone coral]

it's just much easier to express what's going on using code

because card games are already algorithmic

[polar ermine]

Very true. Literally everything beyond the chosen course of action is behind the scenes calculations lol.

The same could be said about a lot but eh

[cursive horizon]

how does that interact with stuff like a status which boosts player damage by 50%?

maybe that question doesn't really make sense with how e.g. magic models everything

where that would be an artifact which adds +x to cards instead, probably? does magic do any relative calculations?

[undone coral]

IEnumerator DealDamage(int baseDamage, Entity target) {
  var damageBonuses = this.allCardsInPlay.SelectMany(entity => entity.Enchantments).OfType<AttributeModifierEnchantment>().Sum(modifier => modifier.Value(this));
 var totalDamage = baseDamage + damageBonuses;
 target.hp -= totalDamage;
 yield return StartCoroutine(DamageAnimation(...));
 if (target.hp < 0) {
  MarkAsDestroyed(target);
 }
}

[cursive horizon]

what if i want to show the player ahead of time how much damage a card will do?

i guess you can have a CalculateDamage method and call that and pass that result in

[undone coral]

yeah for everything that you need

like, just write the code to do it

don't make a meta coding blah blah lisp system thing UNLESS you need an editor the end user will interact with

because a normal person can't write C#

[cursive horizon]

that isn't really what i would do instead

[undone coral]

well both are totally good

i use a DSL

[cursive horizon]

different solutions for different contexts

[undone coral]

but that's because the end user does make stuff

yeah

[cursive horizon]

'just write the code' isn't really architecture though 😛

where i write the code matters!

[undone coral]

lol

[cursive horizon]

i mostly agree with you

[heady bane]

how do you get asset bundle load from file to work?

so I have a folder of asset bundles set up and ready to go

and the examples code from here:
https://docs.unity3d.com/Manual/AssetBundles-Workflow.html

to load in asset bundles

I can confirm the asset bundles are in the app.

is this because im running the project in the editor and need to use a different directory than the streaming assets directory shown in the example?

[minor scaffold]

Hey guys, I have a quick question. I'm writing an editor script. When I instantiate an object it loses its prefab reference. As you can see in the photo, if I create an object in the scene by drag and drop from prefab to the scene, the object's color is blue and it still inherits from the prefab. But when I instantiate it via script, it becomes "A clone". Is there a way to instantiate the object and not lose the reference of its prefab.

I need this because im using PrefabUtility.GetCorrespondingObjectFromSource(TObject componentOrGameObject)

[long ivy]

https://docs.unity3d.com/ScriptReference/PrefabUtility.InstantiatePrefab.html should do it, right?

[minor scaffold]

thanks it does

idk how I miss that

[frank rune]

my player stopped colliding with the ground

what do i do

[dusty shell]

what happens if an axis positional float of an object over/underflows in unity?

will the game crash? simply reposition the object at the other end of the line?

[austere jewel]

I would presume it just goes to positive/negative Infinity and prints warnings to the console if associated with a renderer. 🌈 Check it yourself 🌈

[fresh salmon]

In a checked environment (outside of an unchecked block), it would simply throw an OverflowException. At least, that's what it does for int values, so it should logically be the same for float

Update, it doesn't, it just goes to (negative) infinity

[frank rune]

closestPoint() is not working for a tilemapcollider2d

what should i use

[fresh salmon]

What do you mean by "not working"?
Reminder that this is #archived-code-advanced, where debugging and more information than this is expected

I could not find any similar issues on the internet, and the Docs for ClosestPoint does not warn about issues on TilemapCollider2D.
The page specifies that it will work with all types deriving from Collider2D

Update, TilemapCollider2D indeed inherits from Collider2D

[frank rune]

When I use the mouse position in the parameter it gives me a position on the corner of the entire tilemap

[fresh salmon]

Check the collider's bounds, debug the positions.
Make sure you're converting mouse position (screen space) to world space correctly using Camera:ScreenToWorldPoint

An alternative would be calling WorldToCell on the tilemap, to get a specific cell from a specific world position

[frank rune]

thanks!

I wasn't converting the mouse position

it works now

[shell grove]

Update to this part: #archived-code-advanced message

https://puu.sh/Iy5PM/4a8d7033da.png

I've got it working to end on a 0 in the wavestream

Where this is the code that I use

yield return new WaitUntil(() => Master.time / ((int) (44000 / notes[btn].Frequency) / 44000f / (int) (44000 / notes[btn].Frequency)) == 0f );

[dusty shell]

@austere jewel @fresh salmon aight thanks alot :D

any suggestion on how to prevent players from overflowing their coords?

in an open space

using death zones?

[fresh salmon]

You'll start getting noticeable float precision issues around 100k, where stuff would start to violently vibrate as you move, so if your map is larger than 200 km long, death zones

[dusty shell]

ive actually noticed this behavior in very old games too ya, where the player model comprised of multiple 3D objects started to vibrate and dislocate

[fresh salmon]

Or, move the whole map and player towards the origin of the world

Your choice

[dusty shell]

so this is due to float rounding errors?

[fresh salmon]

Yes

[dusty shell]

kk

[shell grove]

That is the playback time in seconds of the currently playing Audioclip.

Master.time

This is the amount of samples in any given Audioclip.

(int) (44000 / notes[btn].Frequency)

This is the frequency of the samples.

 44000f

As you can see, I'm calculating the time in seconds it takes to play 1 sample of the AudioClip. So in theory this should stop as soon as 1 sample has been played.

The problem is that

yield return new WaitUntil()

is apparently too slow to accurately check that, since it's frame depended. My Unity does not run at 44000 fps, though I wish it would 😀

Any suggestions how to solve this problem?

[fresh salmon]

You can try it for yourself,
Debug.Log(1000000f + 0.1f)
You'll get 1000000 in the console because it can't handle it

[dusty shell]

that sounds viable too

but would have to work with relative coords then right where everything is dependent on the center

[dusty shell]

ah

yea ive heared about that issue with floats

so at 1m it cant actually calculate floats anymore

1b*

due to the 32bit limit

[shell grove]

Isn't there an equivalent to float as there is BigInt?

[dusty shell]

there is double

but its only double accuracy

not double size

[shell grove]

halfdouble

[dusty shell]

wait it is

64bit

[fresh salmon]

That's where the name comes from.
At larger values it doesn't have enough bits to store the integer part anymore, so it moves the decimal point to the right to allocate more bits to the integer part, in term reducing the bit for the decimal part -> precision issues ensues.
Float = floating point, the decimal point moves left or right according to the needed precision

[shell grove]

Double is more precise than float and can store 64 bits First google result when searching Double or Float

[fresh salmon]

And yes, float is 32 bits, double 64

[dusty shell]

whats a big int

[fresh salmon]

In C# it's long, a 64 bit integer

[dusty shell]

ah long

[shell grove]

Yeah

[shadow seal]

There is a BigInteger in System.Numerics

[dusty shell]

also i think the biggest primitive C supports is 256bit

[shell grove]

I like the naming shenanigans there, "You want a bigger int? BigInt it is. You want a 16-bit int? SmallInt it is"

[fresh salmon]

Oh yeah C is weird, you can do long long as a data type I think

[dusty shell]

unsigned long long int

[shell grove]

Oh long johnson

[dusty shell]

tho idk the c# equivalent for that

[fresh salmon]

There isn't one

[dusty shell]

🤷‍♀️

[fresh salmon]

Usually long is sufficient enough

[dusty shell]

true

[shell grove]

Anyway, does anyone know a frame independent alternative to WaitUntil() ?

[dusty shell]

yes

its called delta timing

if its just time based

what you mean

[shell grove]

see here #archived-code-advanced message

[fresh salmon]

I think you're limited by Unity here, as everything is synced to the framerate, or the fixed time step, nowhere near fast enough to achieve what you want

[frozen imp]

You can't execute something faster than your program can run

[dusty shell]

im not really sure what you want to achieve

do you want to measure the elapsed time?

[fresh salmon]

And 1 / 44000 is quite fast

[shell grove]

The problem is that I want to stop the playback of an AudioClip at any given time. But stopping an AudioClip when it's not on a 0 sample will result in an audible "pop" sound

[frozen imp]

Which format are you using?

[shell grove]

AudioClips generated at runtime

so uncompressed wave?

I actually don't know what format they are

        //Generate Sine
        for (int j = 0; j < notes.Length; j++)
        {
            float[] samples = new float[(int) (44000 / notes[j].Frequency)];
            for (int i = 0; i < samples.Length; i++)
                samples[i] = Mathf.Sin(Mathf.PI * 2 * i * notes[j].Frequency / 44000);
            AudioClip ac = AudioClip.Create(string.Format("Sine {0}", notes[j].Name), samples.Length, 1, 44000, false);
            ac.SetData(samples, 0);
            sineClips.Add(ac);
        }

[frozen imp]

You can sample the timing and stop before that happens earlier in execution order probably instead.

[shell grove]

Could you elaborate?

[flint sage]

OS based timers do not have the precision to stop at 1/44000th intervals

[dusty shell]

basically you'll have to work with your clock speed

you cant realistically measure samples/time faster than you cycle trough your lines

[frozen imp]

I mean test it before it plays.

[dusty shell]

only to use better hardware/write better code is what you can do

[flint sage]

Tbh, a quick fade out would probably work just fine

[dusty shell]

but you dont need to either

[frozen imp]

Use any of the entry points before it happens https://docs.unity3d.com/Manual/ExecutionOrder.html

[dusty shell]

human reaction time is about 15-30ms

[fresh salmon]

Or when you pause, wait for the next "0 sample" as you call them, and stop there

[shell grove]

Hrmm. General question to understand the timing between operations.

var test = 1;
test++;
Debug.Log(test);

How long does it take to execute this code?

[dusty shell]

ask ur hardware

compiler

OS

it varies

[fresh salmon]

Three instructions in IL, with Debug.Log the most expensive one

[shell grove]

So is it tied to the clock of the CPU

[fresh salmon]

Yes

[dusty shell]

for example u have 1 core

[fresh salmon]

The exact timing cannot be precisely determined, as it depends on your CPU

[shell grove]

But something like

yield return null;

Is tied to the framerate?

[dusty shell]

this core now executes the OS, discord, steam, 300gb of internet services and your program

compared to only OS and ur program

the second runs faster

[fresh salmon]

I'm not sure how all of that is related tbh

[dusty shell]

he asked how to determine the speed at which instructions are executed

[fresh salmon]

Plus, CPU speed vary now, it can throttle down if it has nothing consequent to do

[dusty shell]

im explaining

that there are thousands of factors

[shell grove]

I see

[dusty shell]

also if the cpu overheats itll turn its clock down

so there is neither a way to determine how fast your program runs

[shell grove]

So even when the timing may be accurate on my PC, it might not be accurate on other peoples PC.

[dusty shell]

nor to pinlock it at an exact speed

exactly

[frozen imp]

@shell grove All you need to know is everything happens in one frame, and what you can manage is when in the frame it will happen.

[fresh salmon]

The main thing is, you'll never get a program to run as fast as your CPU clock, as most C# calls result in 1+ instructions

[shell grove]

So it would be better to make every timing frame-dependent to ensure accurate timing regardless of hardware limitations?

[fresh salmon]

So 1+ clock cycles

[dusty shell]

what you can do to improve performance for time critical tasks is to start a new thread on a new physical core and dedicate this thread to that very one task

make it async

and communicate with the main thread via an interface

[fresh salmon]

Yes

[frozen imp]

frame independent that is

[dusty shell]

that or start a new thread in a while true

[shell grove]

But I thought frame independent meant that it's up to the hardware

[dusty shell]

but keep in mind that incorrect thread handling might actually turn your processing speed down

unity, c# and c++ based dlls all can offer multithreading solutions if youd want that

[frozen imp]

You can't know at which frame rate different hardware will run the program. You can force it though to lower rate.

But it's not recommended

Only universal factor is timing

[shell grove]

I think Timinator said that deltaTime could work

[dusty shell]

you basically just store the time

add the elapsed time to that each frame or tick

and compare it to your wanted threshold

if it exeeds you execute your whatever and reset the timer

[frozen imp]

What exactly do you want to do?

[shell grove]

var i = time.deltaTime;
yield return null;
Debug.Log(time.deltaTime - i);

I guess this would give me the time it took for that 1 frame to pass with yield return null;

[dusty shell]

that just measures the difference in time between 2 relative time points, in this scenario between frames

[fresh salmon]

That's what deltaTime is, here.
yield return null waits for the next frame, so you already have the time in deltaTime. No need for those calculations

[dusty shell]

just speaking generally btw, unity caps the fps and thus game ticks to the monitors refresh rate right?

[fresh salmon]

Only if you tell it to, or have vsync enabled

[dusty shell]

yea

so

basically thats to be taken into consideration too

[shell grove]

I thought about an Idea, tho it may be a stupid one
Instead of

yield return new WaitUntil(() => Master.time / ((int) (44000 / notes[btn].Frequency) / 44000f / (int) (44000 / notes[btn].Frequency)) == 0f );

I could do

while(Master.time / ((int) (44000 / notes[btn].Frequency) / 44000f / (int) (44000 / notes[btn].Frequency)) != 0f)
{}

[dusty shell]

i spent a great deal of time

asking myself

how do i prevent cheaters from ramping up their clock speed to cheat speed walking

the simple answer is delta timing

[shell grove]

Tho I think that would freeze the program

[dusty shell]

but this applies universally

[dusty shell]

depends, cheat engine can do that

[fresh salmon]

Huh?

[dusty shell]

super old games suffer from this vulnerability

[shell grove]

No I meant the while loop xD

Sorry

[dusty shell]

it seems incredibly accurate only losing maybe 1-2ms

[shell grove]

Sadly not, most of the time the samples take 2-3 seconds to stop

[fresh salmon]

Much more, like 16ms at 60fps

Because WaitUntil is tied to framerate, and is of course, variable

[dusty shell]

also id refrain from while loops in the main thread of unity

it freezes your entire game

[shell grove]

Basically every 16ms we got a shot that maybe we are on a 0 sample, if not we wait again

[dusty shell]

so

we talk about aprox 16ms inaccuracy?

that right?

[shell grove]

16ms in 44000hz is quite the amount of samples

[dusty shell]

then multithread it

and let the thread run async

only way i see in the unity env to get it run faster

[shell grove]

I never multithreaded

[dusty shell]

🤷‍♀️

its actually not that hard

[fresh salmon]

And I don't think it would be appropriate here

[dusty shell]

and both unity and .NET offer solutions

[frozen imp]

You can sample play speed if it's not constant, you cant get current clip time with https://docs.unity3d.com/ScriptReference/AudioSource-time.html And stop it if it's less than sample per frame speed left, or several frames before if you need to.
What's actually a problem

[dusty shell]

then the approach of measuring time seems incorrect

cuz i at least cant think of a way to measure the time faster(at higher frequency)

[fresh salmon]

Even if they multithreaded, it would be far too fast compared to the 44kHz rate the sound imposes

[dusty shell]

well and u have to consider weaker hardware too

welp i havent done much with audio playing yet so im mostly out then

[unreal badger]

yo guys, i wanna ask something, so i make a basketball game then the player can throws the ball when pressed left mouse, but the problem is when i click pause button the player can still throws the ball, so how can i fixed that?

[fresh salmon]

Is that question really suited for that channel, or did you post here because it's active?

Prioritize relevance over activity, folks

[shell grove]

if(paused)
return;

Add that in your ThrowBall function, or any equivalence to that.

I think I will try the FadeOut Solution since getting really accurate timings when dealing with this kind of frequency seems near impossible

Or, big brain moment

I could add a few 0 samples between each cycle of the sine. That way the WaitUntil could maybe catch it better?

Anyway big thanks for the help Timinator, SPR2 and Fogsight 👍

[thick thorn]

Is it possible to write a property drawer for a custom list class (which would basically just be a wrapper for list<T>) that draws the serialized fields for the most derived type of T? Perhaps through reflection?

[shell grove]

I sadly have no clue what a property drawer is Hambones. But I think some will know 👍 Good luck on your project

[unreal badger]

thanks it works, this logic is pretty simple.

[thick thorn]

Basically I want a polymorphic container that allows you to edit derived types of T in the inspector. I think this is possible at least through reflection, possibly even not, but I’m not sure.

I think that in order to actually allow selection of a subtype in the inspector, reflection would be needed…

[dusty shell]

ive seen many talk about the old and new input system of unity

whats the actual technical difference?

[thick thorn]

Basically the new system provides the concept of actions, which lets you separate obtaining inputs from employing them.

[shell grove]

oh btw Timinator, I'm limited to C#4 and Unity 2017

Because I'm modding for a game that runs on that version. Dunno if Multithreading would even work there

[thick thorn]

Multi threading is very old

[normal cedar]

Hey anyone familiar with an issue where i cant make any kind of webrequest without a WEBGL build crashing on me

Maximum call stack size exceeded

using LitJson;
using UnityEngine;
using UnityEngine.Networking;
using System.Collections;
public class SaveFile : MonoBehaviour
{
    [Tooltip("Details from where to get the personalID")]
    public LoginDetails Login;

    public DataTable Data;
    [SerializeField] private string JSONLink;
    public JsonData jsonvale;

    public void Start()
    {
        Init();
    }

    public void Init()
    {
        StartCoroutine(Getlink());
        GetAchievements();
    }

    public IEnumerator Getlink()
    {
        UnityWebRequest client = UnityWebRequest.Get(Login.URL);
        client.SetRequestHeader("AUTHORIZATION", Login.Authenticate());
        yield return client.SendWebRequest();
        if (client.result == UnityWebRequest.Result.Success)
        {
            JSONLink = client.downloadHandler.text;
            Debug.Log(JSONLink);
        }
        else
        {
            Debug.Log("Could not make a connection");
            yield return null;
        }
            if (JSONLink.Length > 0)
            {
                jsonvale = JsonMapper.ToObject(JSONLink);
                Data.isLoggedIn = (bool)jsonvale["loggedIn"];
            }
    }

    public void GetAchievements()
    {
        if (Data.isLoggedIn)
        {
            Debug.Log("Nice");
        }
    }
}

[System.Serializable]
public class DataTable
{
    public bool isLoggedIn;
}

[System.Serializable]
public class LoginDetails
{
    public string URL;
    public string UserName;
    public string PassWord;

    public string Authenticate()
    {
        string auth = UserName + ":" + PassWord;
        auth = System.Convert.ToBase64String(System.Text.Encoding.GetEncoding("ISO-8859-1").GetBytes(auth));
        auth = "Basic " + auth;
        return auth;
    }
}

[shell grove]

If you reach the maximum call stack you have an infinite loop somewhere

[normal cedar]

this is the only script active

and it doesnt have a loop

wait. i just uploaded an empty build and it still does it. let me check

[unkempt nova]

I'm trying to figure out the cause of an android build crash, so I installed Unity's Android Logcat package. I think I found the cause, but have no idea what I'm looking at. Anyone able to give any insight?

[shell grove]

@normal cedar
Maximum call stack size exceeded means that somewhere in your code, you are calling a function which in turn calls another function and so forth, until you hit the call stack limit.

[unkempt nova]

The server backend is written in Java, and this mentions a null pointer dereference, and seems to say it's a Java error, but that's all I've got. If I can just confirm that's the cause, that'd be good, but will need more info on which bit of code is causing it if possible

[shell grove]

nullpointer reference, something tries to access something that isn't there

[normal cedar]

chrome doesnt give me the error so it is opera being annoying now

[unkempt nova]

Yeah, I figured that bit out, is it Java you think? I don't think this sort of thing is Unity-based

[normal cedar]

webgl has been nothing but a pain in the ass for me

[shell grove]

Well the backtrace shows it's mostly Unity

[shell grove]

Try commenting out various calls you make and see which one causes the stack overflow

[normal cedar]

it doesnt even get at that point

cant even start the game

[shell grove]

like GetAchievements(); or StartCoroutine(GetLink());

[normal cedar]

ill try that.

too bad it takes 10 minutes every single time to build

[shell grove]

Can't you run it in Unity?

[normal cedar]

in unity it runs fine

it only breaks on build

and when i put it on a server

and i need to it to load in a personalID to load in all their data

so it is too important

it works now

problem was Opera

[shell grove]

Weird

[fresh salmon]

I'm late, but yes, that's a Java error. It's a bit like a C# NRE, but it's not the same name as the common java.lang.NullPointerException so idk. Names changed maybe?

[unkempt nova]

Hrm, ok. Thanks

[thick thorn]

well... just learned something new... [SerializeReference] will polymophically serialize objects.

Now I just need to figure out how to write the custom inspector to allow selection of a desired derived type when adding an element. That will probably require reflection.

Don’t mind me, im just talking to myself I guess…

[fresh salmon]

If you need a confirmation, Reflection can do that. Look for the Type:IsAssignableFrom(Type) method

Works with interfaces as well.
Classes only, Type:IsSubclassOf(Type) will do

[thick thorn]

Thanks. I think I probably also need to figure out how to create a aerialized object in a custom inspector, so that upon receiving a selection of a type, the aetialized object of the appropriate type is created.

[fresh salmon]

As long as the type can be instantiated (not abstract nor static nor interface), you can use Activator.CreateInstance(Type) to create an instance of an object at runtime

I think the object needs to declare a paramless constructor, unless there is an overload where you can pass arguments

[sage radish]

There are a couple of repos on GitHub that do exactly this, like this one:
https://github.com/mackysoft/Unity-SerializeReferenceExtensions

https://user-images.githubusercontent.com/13536348/118233552-03cd1780-b4cd-11eb-9e5b-4824e8f01f1d.gif

[thick thorn]

sweet, thanks

i keep looking to the asset store for neat tools like this but usually can't find anything. looks like a lot of these types of things are just on github or whatever

[white zephyr]

Not exactly sure if this is the right channel: Does anyone have experience with debugging Unity crashes on Linux?

I have a coredump, but I don't know how to get the right stacktrace out of it, as GDB is unable to deal with the managed symbols from mono

[thick thorn]

oh wow you can make a property drawer apply to a field using an attribute... how did i miss that one... i feel like no matter how much unity i learn, i always feel like i know maybe 5% of it. the more i find out, the more i realize i don't know

[unkempt nova]

For my crash issue, I found the "tombstone" logcat was refering to. Any ideas what I can check for this? Looks like URP stuff, but this is basically gibbrish to me managed backtrace: #00 (wrapper managed-to-native) UnityEngine.Rendering.ScriptableRenderContext:Submit_Internal_Injected (UnityEngine.Rendering.ScriptableRenderContext&) #01 UnityEngine.Rendering.ScriptableRenderContext:Submit_Internal () <0x13> #02 UnityEngine.Rendering.ScriptableRenderContext:Submit () <0x13> #03 UnityEngine.Rendering.Universal.UniversalRenderPipeline:RenderSingleCamera (UnityEngine.Rendering.ScriptableRenderContext,UnityEngine.Rendering.Universal.CameraData,bool) <0x80b> #04 UnityEngine.Rendering.Universal.UniversalRenderPipeline:RenderCameraStack (UnityEngine.Rendering.ScriptableRenderContext,UnityEngine.Camera) <0xbbf> #05 UnityEngine.Rendering.Universal.UniversalRenderPipeline:Render (UnityEngine.Rendering.ScriptableRenderContext,UnityEngine.Camera[]) <0x157> #06 UnityEngine.Rendering.RenderPipeline:InternalRender (UnityEngine.Rendering.ScriptableRenderContext,UnityEngine.Camera[]) <0x3f> #07 UnityEngine.Rendering.RenderPipelineManager:DoRenderLoop_Internal (UnityEngine.Rendering.RenderPipelineAsset,intptr,System.Collections.Generic.List`1<UnityEngine.Camera/RenderRequest>) <0xcb> #08 (wrapper runtime-invoke) <Module>:runtime_invoke_void_object_intptr_object (object,intptr,intptr,intptr)

Looks like the same info as was in my screenshot. Also, I was trying to track down some mobile performance issues with profiler and I'm pretty sure these calls were at the top. I'm thinking they're related, what do you think?

Unity 2021.3.14f1, URP version 10.5.1 btw

Letting the game start in portrait mode, it seems ok (sometimes...), but soon as I turn the phone so it switches to landscape, it seems to crash. Not sure if it's just randomly crashing and it just seems like the orientation is related though Yeah I think it's just random

[stuck onyx]

             transform.position = Vector3.Lerp(startPosition, target, t);```
I'm using this inside of a coroutine but sometimes the timeToReachTarget varies significantly (0.2 -0.3 sec).... I suspected it was because the change on the frame rate and changed Time.deltaTime for Time.fixedUnscaledTime....  now just varies 0.01 - 0.02.   Does it make sense?

[sly grove]

timeToReachTarget is something you would be setting outside of the loop. Like... if you want the lerp to take 1 second, it should be 1.

[stuck onyx]

@sly grove yes thats how is done

its set to 1

[sly grove]

Ok so what are you asking

you said "timeToReachTarget varies significantly"

[stuck onyx]

but when i was using normal deltaTime

i measured that time

and was not exactly 1 dependinng when did i launcched the method

[sly grove]

Can you show a little more context? How is your loop structured and what kind of yields are you using etc

[humble onyx]

can you show the whole coroutine?
what yields do you use int the loop

[stuck onyx]

sure , 1 second

[sly grove]

It will never be exactly 1 second btw, there will always be an error of up to however long a single frame takes for your game.

[stuck onyx]

   private IEnumerator MyCoroutine()
    {
        float duration = 1;
        float lerp = 0;
     
        while(duration>0 && lerp<1)
        {
            lerp += Time.fixedUnscaledDeltaTime/duration;
            sceneFader.alpha = Mathf.Lerp(1,0,lerp);
 
            yield return null;
        }

    }

yeah but the difference sometimes was more than 0.2 seconds

i suspected it was beccause sometimes i call it when there is a framedrp

framedrop*

thats why i changed

Time.deltaTime TO Time.fixedUnscaledDeltaTime

now seems to be always same time...

I'm asking if what I did makes sense

[sly grove]

why are you using fixedUnscaledDeltaTime

it should be unscaledDeltaTime

[stuck onyx]

because i've heard deltaTime is frameDependant and i thought that could solve it

aha

so with unscaledDeltaTime i would have the same result than now

[sly grove]

no... deltaTime is there to fix framerate dependency

since your coroutine is looping every 1 frame, you need deltaTime to adjust for that

[stuck onyx]

but if I use deltaTime the times varies between 0.2 or even 0.3 seconds , i dont understand why

i used StopWatch class to measure it

[sly grove]

print out the current time and the value of lerp every loop.

therre's no magic here

it will stop when lerp reaches 1

[humble onyx]

deltaTime is the time since the last frame
if your frame time / FPS varys so does deltaTime ands thats wht its for

[stuck onyx]

and thats why using deltaTime does this coroutine take more or less time depending on the FPS

corrrect?

[sly grove]

if your game is chugging along such that a frame takes 0.2 seconds to render, there can be an error of up to that much time

Are you having framerate issues?

[stuck onyx]

yes, actually i do some sorrt of render heavy stuff in one ocasion

no, just on that particular moment

ill put some context

[sly grove]

If you have a particularly slow frame, then yes, this can have an error of up to the length of time it takes to render that one frame

[stuck onyx]

aha, so the correct one to avoid this would be the one you suggested?

unscaledTime?

[sly grove]

No, there's no avoiding it

[stuck onyx]

or the one im using (which seems to work)

[sly grove]

The answer is to fix your framerate problem

[thick thorn]

sweet, it works. thanks for the info

[stuck onyx]

yes, the one im using makes it work

@sly grove ill put more context

im doing this:
-1 I fade the screen white with the coroutine (no prob)
-2 I ennable a camera and take a render snap to create a texture (this takes some time depending on the device)
-3 whenn the snap is done... i fade to transparent again

1st fade takes 1.000 second, 3rd fade takes 0.6 - 0.8 seconds

I think its because, even if i do it AFTER the render snap is done there still might be some work going on

this happened using Time.deltaTime

then i changed to Time.fixedUnscaledDeltaTime and 1st and 3rd fades take the same amount of time... just with 0.0X difference

and I'm asking.... does this make sense?

😄

[sly grove]

I think you're just seeing the cumulated effects of error between all three coroutines

Your first one might take like 1.05 seconds
The second one 1.07 seconds
the third one 1.04 seconds

So the whole process takes 3.16 seconds

just as an example

[stuck onyx]

uhmm the first one and third one are the same code... and i do

    sw.Start();
 while(duration>0 && lerp<1)
        {
            lerp += Time.fixedUnscaledDeltaTime/duration;
            sceneFader.alpha = Mathf.Lerp(start,end,lerp);
 
            yield return null;
        }
Debug.Log("SatviewExit Time " +enable+" :"   +sw.ElapsedMilliseconds);
        // sw.Stop();

so i can see pretty clear the times of the first and third coroutine

take into account the second doesnt startt until first is done

and 3rd doesnt start until second is done

[sly grove]

Note that Unity's time API deals with the time as it was at the beginning of the frame always.

[stuck onyx]

im measurinng the times of 1st and 3rd by separated

[sly grove]

If Time.time is 492.3 for example, it will be 492.3 for the entire frame. Unlike System time apis which will keep ticking even during a frame

[stuck onyx]

and what does that mean? sorry i didnt get it 😅

[sly grove]

It means that there will be some discrpancies between what you see with things like Stopwatch and what Unity's time APIs are reporting

[stuck onyx]

oh

so deltaTime should take the same amount of time then.

gonna try again.

[sly grove]

Could you show all your code btw and what is being printed?

[stuck onyx]

sure

how was the color paste thinng?

public void TriggerSatViewExitFader(bool enable, Action onCompleted)
    {
        StartCoroutine(SatviewExitFaderCoroutine(enable,onCompleted));
    }
    private IEnumerator SatviewExitFaderCoroutine(bool enable,Action onCompleted)
    {
        float duration = enable
            ? configService.varsCfg.satViewExitFadeInTime
            : configService.varsCfg.satViewExitFadeOutTime;
        float start = enable ? 0 : 1;
        float end = enable ? 1 : 0;
        sceneFader.gameObject.SetActive(true);
        float lerp = 0;
        Stopwatch sw = new Stopwatch();
        sw.Start();
        while(duration>0 && lerp<1)
        {
            lerp += Time.fixedUnscaledDeltaTime/duration;
            sceneFader.alpha = Mathf.Lerp(start,end,lerp);
 
            yield return null;
        }
        Debug.Log("SatviewExit Time " +enable+" :"   +sw.ElapsedMilliseconds);
        sw.Stop();
        sceneFader.alpha = end;
        sceneFader.gameObject.SetActive(enable);
        onCompleted?.Invoke();
    }```

there

if i pass enable=true fadesInn if i pass enable=false fadesOut

[sly grove]

ok makes sesne

(other than fixedUnscaledDeltaTime)

and what does it print

[stuck onyx]

first i noticed visually it was not totally exact (using deltaTime)

then i placed the StopWatch and confirmed

ill runn, 1 sec

@sly grove this is with fixedUnscaledDeltaTime:
[Log] SatviewEnter Time True :862
[Log] SatviewEnter Time False :969

first linne fadeIn second fadeOut

acceptable

now i try deltaTime

[sly grove]

SatviewEnter Time doesn't the code say SatviewExit Time?

[stuck onyx]

yeah i just took the method below but its the same code actually, didnt feel like merging both

its the same code but taking different fadeTimes

[sly grove]

Btw does your game mess with Time.timeScale at all

[stuck onyx]

but all are 1 currenntly

no,

never

[sly grove]

ok what do you get when you do deltaTime

[stuck onyx]

runnin

[Log] SatviewEnter Time True :1015
[Log] SatviewEnter Time False :696

[sly grove]

🤔

[stuck onyx]

0.3

[sly grove]

the first one makes sense

[stuck onyx]

for a coder is nothing but for my game designer its a lot 😅

[sly grove]

the second one... doesn't

you 100% sure the code is the same?

[stuck onyx]

well, for a coder, in my context

yes, 100%, i paste againn

i paste the correct

private IEnumerator SatviewEnterFaderCoroutine(bool enable,Action onCompleted)
    {
        float duration = enable
            ? configService.varsCfg.satViewEnterFadeInTime
            : configService.varsCfg.satViewEnterFadeOutTime;
        float start = enable ? 0 : 1;
        float end = enable ? 1 : 0;
        sceneFader.gameObject.SetActive(true);
        float lerp = 0;
        Stopwatch sw = new Stopwatch();
        sw.Start();
        while(duration>0 && lerp<1)
        {
            lerp += Time.deltaTime/duration;
            sceneFader.alpha = Mathf.Lerp(start,end,lerp);
 
            yield return null;
        }
        Debug.Log("SatviewEnter Time " +enable+" :"   +sw.ElapsedMilliseconds);
        sw.Stop();
        sceneFader.alpha = end;
        sceneFader.gameObject.SetActive(enable);
        onCompleted?.Invoke();
    }

[sly grove]

Can you print out what the duration is?

        float duration = enable
            ? configService.varsCfg.satViewEnterFadeInTime
            : configService.varsCfg.satViewEnterFadeOutTime;```
I See you're getting two different durations

[stuck onyx]

duration is always 1 , 100%

[sly grove]

Print it

[stuck onyx]

but i can print

[sly grove]

just to make sure

[upper scarab]

I have an Assembly Definition Question!

My folder structure looks like:

Assets

• Game1
• Game2
• Plugins
• Common

So, each of the folders above has an asmdef. All of the asmdefs rely on "Plugins".

I have TextMeshPro as a reference in Plugins. What's curious is that even though Game1, Game2, and Hub rely on plugins, they can't seem to find the reference for TMP. Is there a simple setting I'm missing? It feels overkill to have to add TMP as a hard reference in every assembly.

[stuck onyx]

running...

Duration: 1
SatviewEnter Time True :1002

Duration: 1
SatviewEnter Time False :741

[sly grove]

I don't understand how lerp could be getting > 1 then

[stuck onyx]

what is taking > 1 is the time but not the lerp result

you mean the alpha?

[sly grove]

I mean the lerp variable

that loop continues as long as lerp is < 1

If you're adding deltaTime each frame (divided by 1 which doesnt matter), that should take 1 second (or slightly longer)

as it does for the first run

Unless timeScale is getting affected

[stuck onyx]

oh!

[sly grove]

Can you search your project for Time.timeScale?

[stuck onyx]

yeah, notim eht whole proyect, just comparsions

not in the whole project

[sly grove]

What happens if you use Time.unscaledDeltaTime

[stuck onyx]

let me try

[upper scarab]

did you use control+f or control+shift+f to search the project?

[stuck onyx]

Time.timeScale= 1

but thats the default so its alright

[stuck onyx]

command F

but i use Rider

[sly grove]

That only searches one file, no?

[stuck onyx]

sorry, command + Shift + F

[sly grove]

Regardless, if unscaledDeltaTime works then we know it's a time scale issue

[upper scarab]

cool, just wanted to make sure 🙂

[stuck onyx]

Duration: 1
SatviewEnter Time True :1022
Duration: 1
SatviewEnter Time False :152

this was even worse 😮

[sly grove]

wtf

[stuck onyx]

yeah, did it again

same

144

[sly grove]

that makes... so little sense lol

[stuck onyx]

ill try fixedDeltaTime

[Log] Duration: 1
[Log] SatviewEnter Time True :872

[Log] Duration: 1
[Log] SatviewEnter Time False :1068

hmmmmmmm

nnot as good as ffixedUnscaledDeltaTime

why the heck is happenning this

with fixedUnscaledDeltaTime i get it almost the same

it has to be something related with the camera.render() im calling

[sly grove]

What happens if you print out Time.time before and after the loop

instead of using StopWatch

[stuck onyx]

ok 1 second

look

if i use the same method

but without the camera.Render() between them

Time.deltaTime works fine

[sly grove]

oh there's a Camera.Render happening?

[stuck onyx]

of course its the second step i told you

[sly grove]

how come

[stuck onyx]

thats why i said the framedrop thing

[sly grove]

Ohh

[stuck onyx]

-1 I fade the screen white with the coroutine (no prob)
-2 I ennable a camera and take a render snap to create a texture (this takes some time depending on the device)
-3 whenn the snap is done... i fade to transparent again

[sly grove]

ok - I wonder if there's a bug with Camera.render where unity doesn't account for the time that takes appropraitely

[stuck onyx]

if step 2 is out of the equation then works fine

[sly grove]

can you see if you wait like 1 extra frame between step 2 and 3

[stuck onyx]

sure

[sly grove]

like add a yield return null at the beginning of the coroutine or something to do it quickly

[stuck onyx]

but then the time will be affected

or not

ah okk ok

SatviewEnter Duration: 1
SatviewEnter Time True :1021
SatviewEnter Duration: 1
SatviewEnter Time False :700

same

[sly grove]

I don't ujderstand T_T

can you show your camera render code

[stuck onyx]

sure

  void Awake()
    {
        if (snapCam.targetTexture == null)
        {
            snapCam.targetTexture = new RenderTexture(resWidth, resHeight, 24);
        }
        else
        {
            resWidth = snapCam.targetTexture.width;
            resHeight = snapCam.targetTexture.height;
        }
    }
    
    public void SetWaterObject(GameObject waterGO)
    {
        waterObject = waterGO;
    }

    public void TakeSnapshot(Action onCompleted)
    {
        waterObject.SetActive(false);
        waterSprite.enabled = true;
        snapCam.enabled = true;
        snapCam.Render();
        snapCam.enabled = false;
        waterObject.SetActive(true);
        waterSprite.enabled = false;
        onCompleted.Invoke();
    }

gotta go now @sly grove thanks a lot for your help, if you come up with something please let me know

[undone coral]

what's your objective? you're trying to create a reflection?

like a reflection effect for sprites?

i think there's a lot about this online

usually you have to make sure the camera renders at the "right time", and then the most performant way to use the render texture is through a shader property that isn't exposed in the shader graph

you can make 2d sprite shader graphs in URP

https://www.youtube.com/watch?v=O1lRGKfCi9o

@stuck onyx

[undone coral]

you have to add text mesh pro to every assembly. but you only have 4

[undone coral]

try changing from forward to deferred rendering

[upper scarab]

we have 27 games sadly. It just feels weird to have all of these hard refs, I feel like dependency management shouldn't work this way 🤔

[undone coral]

you don't need to reference tmp in assemblies that don't use them

what is the underlying issue?

you're saying 27 folders with asmdefs in them, each with 27 scenes, that you build to 27 distinct exes?

you can just reference it once in common

[upper scarab]

The underlying issue is that we licensed several games to our platform. They each used their own version of TMP when they were made. We wanted to unify them all to use the Package Managed version of TMP

[undone coral]

you should be able to reference it once in common

[upper scarab]

So "plugins" is our "common"... and all of the games reference plugins. You're saying that if Plugins references TMP, then the rest should have access? That was my thought as well, but that doesn't seem to be the case

[undone coral]

what kind of error do you see?

[unkempt nova]

I think I'm missing something. The docs say I should have the option to change it here, but I don't have Lighting at all https://docs.unity3d.com/Packages/com.unity.render-pipelines.universal@12.0/manual/rendering/deferred-rendering-path.html

[undone coral]

hmm

[upper scarab]

So in this case, the scripts are in Assets/Hub

the Hub.asmdef, relies on plugins.asmdef. plugins.asmdef references TMP. I was hoping that the scripts in Hub could use TMP, without adding a reference to TMP in the hub.asmdef

[static marsh]

I just confirmed this; the dependencies do not seem to carry over the way you would expect. You have to manually reference the assembly, even if it’s already referenced by an assembly in your references list

[upper scarab]

Yeah i created a test project to test this and ran into that. I was just curious if there was a setting or set-up that i was missing. Thanks for the confirmation! ❤️

[static marsh]

Glad I could help! There could still be a setting, but I don’t know where. I just ran through a test with one of my projects and everything is manually referenced.

[upper scarab]

that feels weird, doesn't it? I suppose I'm just used to something like Nuget that makes it easier lol

[crystal ridge]

If I have an abstract base class called Food and it implements an interface called IFood, does that mean when I create a new script and inherit from Food it also inherits the interface (IFood) as well?

the reason is, because I have a trigger looking for the interface

IFood food = other.transform.getcomponent<IFood>();

But i'm wondering if I even NEED that, or I can just look for the base class instead from the derived script

Food food = other.transform.getcomponent<Food>();

[sly grove]

yes

If you have Food : IFood and HotDog : Food then implicitly it is true that HotDog : IFood

[crystal ridge]

Thanks @sly grove! I think you're like 5/5 of my questions answered lol, I need to buy you a coffee sometime

[polar ermine]

i have a scriptable object class Statuses that acts mostly as a dictionary to hold objects of a serializable class Status in an array. I'm able to add and remove statuses in the inspector very easily. Now I'd like to be able to reference these statuses elsewhere in the inspectors of other gameobjects, somehow.

example being i have an enemy type as their own scriptable object. I'd like to have a list of "statuses" in that script, and I'd like to simply pick and choose which status that enemy will have in the inspector from said list that i filled out in the original Statuses SO.

public class Statuses : ScriptableObject
{
    public List<Status> statuses;
}

this is the whole class; i basically want to use whatever's in the statuses list as a dropdown to select from. it's probably simpler than i think but im not familiar enough with unity 😔

[normal cedar]

@polar ermine It would just be better to call the status script in your enemy as a variable

if you have to select it it shows you the statusses anyway

public class Enemy : MonoBehaviour
{
  public Status status;
}

but if you wanna go real fancy it is possible but the solution i showed above is just way cleaner

[polar ermine]

The only problem with that is i need to input the values for that particular Status object all over again

i write it once in the dictionary and need to write it again it kinda defeats the purpose of a dictionary

[normal cedar]

wait wut?

you know scriptableobjects can not save contents that have been changed during runtime right?

you have to pre-assign them.

[polar ermine]

im pretty sure i've accidentally permanently mutated values in a SO from in game but that's not what im trynna say, i think

[normal cedar]

that would be odd but possible if you have a save loading system.

But what would be better is to do this:

[CreateAssetMenu("idk what goes here anymore")]
public class Status : ScriptableObject
{
  Data
}

and then have a seperate class for your enemy where you just call status

public class Enemy : MonoBehaviour
{
  public Status status;
}

[sly grove]

Yeah the worst part about that though is that it only works that way in editor. In a build it won't.

[normal cedar]

that was it yeah!

[polar ermine]

i see, good to know

[normal cedar]

if you have any question about how it goes in working i could show you

[polar ermine]

im just trynna think if my current set up could work

[Serializable]
public class Status
{
    public string name;
    public string description;
    public Sprite sprite;
}

this is my current status class, and i showed the statuses dictionary SO class earlier

[normal cedar]

oooh yeah it is very possible

you can just make it into a CreateAssetMenu

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

[polar ermine]

right. in my enemy class i started by doing Status[] statuses, but in the inspector i have to fill all that information again, i can't pick from the list of statuses i already made.

i guess making each Status it's own thing would be good too, though

and assign each status to a list object in the dictionary so?

[normal cedar]

yes

very possible

[polar ermine]

makes it a little bit less modular than I'd like, but I'll give it a shot

[normal cedar]

ofcourse what you could do it make it like you want, then have an enum with all the statusses name and then do a for loop

[polar ermine]

right. and link each status with the associated enum at runtime and just use that. i started doing that but started to wonder if i could just simply use a list of statuses in place of an enum

because i'd need to be sure i had a good system of "linking" that would involve a standardized naming scheme or something, which is good practice but prone to error because im dumb

that said, I did get something to work to some effect of how i wanted it to.
i just worry about scalability and how much work I'll have to do in the long run using one solution over the other.

might be minor, but needing to make a scriptable object for each status, and putting that object in the statuses dictionary once and then also adding them to cards that use that status instead of being able to directly link to the dictionary feels inefficient to me.

[viral flame]

Anyone mind looking at https://gist.github.com/vantreeseba/7f420eee29a6fd3963eacc901a198e83 and helping me figure out why it's so slow? (to be fair, I'm testing it on like 100 ai dudes running around, but still it's creating like 10/20KB of GC per frame, and taking like 8ms total).

I know I can timeslice it, but it still feels wonky that simply sorting some lists consisting of not many things is so slow.

[thick thorn]

well are you running this on each ai dude?

if so, maybe it has to do with how many times you're doing the for loop on lines 53-67

[viral flame]

Yeah, I mean 80 * 16 = 1280 but that doesn't seem like thaat many raycast calls, and if I remove the sorting, it takes ~1ms (total time for all 1280) to do that.

[thick thorn]

a lot less gc without sorting?

[viral flame]

That makes total sense yeah, but no idea why the sort is exploding.

If I have the sort in, the profiler is telling me between 8ms/12ms per frame for this func (and 10/20KB of GC), without it, no GC and ~1ms. 😦

So it's 100% the sorting, but I feel like there must be something obvious I'm missing or etc, it shouldn't take that long to sort a max of 16 things.

[thick thorn]

https://stackoverflow.com/questions/25730010/is-there-a-way-to-sort-arrays-without-allocating-any-memory

looks like using an IComparer causes per-frame allocations

maybe convert your class to just a static func, as stated in here?

dunno

or i guess implement your own sort algorithm. shrug.

not the nicest answer i guess

your compare looks like a pure function, so should be easy to convert to a static func

or you could even use an anonymous method i guess

(is that the same as a static func?)

[viral flame]

An anonymous method I believe will cause some GC since it must capture surrounding context for the lambda

[thick thorn]

ahh

[viral flame]

Yeah I looked at the non gc list thing you linked earlier, and tried many diff methods, and they all still cause GC, I dunno lol.

[thick thorn]

shrug.

sry

[viral flame]

It's good, thanks for the help. I did manage to get it a bit faster by hinting compiler to inline the comparer, i.e.

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public int Compare(Transform x, Transform y) {
      var xdist = (x.position - _me.position).sqrMagnitude;
      var ydist = (y.position - _me.position).sqrMagnitude;

      return xdist.CompareTo(ydist);
    }

So that's cool, I guess, now it's like ~5ms on avg.

[thick thorn]

nice

[viral flame]

I ended up removing the sort, and just searching for the closest in the list when a function is called, ended up being much faster, and will refactor later if required. 😛

[undone coral]

20kb of gc isn't going to matter

it looks like you figured it out

what were you trying to do?

this kind of spatial querying is expensive

you might want to use a dedicated thing

[undone coral]

no matter what you'll have a giant switch statement

[undone coral]

this is fine. but don't make a scriptable object for every status. it's painful. you should have a single object for all of your game logical assets.

public class CardGame : ScriptableObject {
 [SerializeField] private Status[] m_Statuses;
 [SerializeField] private Card[] m_Cards;
 ...
 public Status[] statuses => m_Statuses;
 // or??
 public Status GetStatus(StatusEnum status) { ... }
}

@polar ermine

you only need one

do you see why this is less painful?

still works with the inspector

"but what if i have lots"

you don't right now

and even when you have more, it'll be less painful than having a ton of individual assets

that you will have to gather into an array anyway

[polar ermine]

yeah, i'd prefer not to have a SO for each status. it doesn't scale well i dont think
how does it work in the inspector exactly?

[undone coral]

you click on it

and then you hvae an array

etc. etc.

no custom editor necessary

anyway i think you'll figure it out

[undone coral]

okay, that's not what the error message says

it's saying it can't find TMPro

[upper scarab]

it can't find it because it isn't directly referenced in the assembly

[undone coral]

i think just go ahead and reference the assembly

if that works for you

[upper scarab]

yep that seems like the only solution!

[undone coral]

27 clicks... it'll take you 2 minutes

[upper scarab]

Im not worried about the time, moreover the fact that if we end up with other dependencies it makes connecting all of the wires more complicated 😅

It feels unclean is all lol

[viral flame]

Just create a list of nearby entities within the FOV not obstructed by obstacles.
It was 20KB per frame, which would end up being quite a bit over time IMO. (1.2MB / second, so after 10 mins, you're gonna have nearly 1GB of memory leak)

[undone coral]

yeah

[viral flame]

With the switch to not sorting, and only searching on demand lowered the per frame time (for ~100 of them), from 5ms to like 0.2ms, so it's basically nothing now lol.

[plucky laurel]

what does the profiler say

from my experience overlap sphere even non alloc is slower than any conventional spatial structure

i had scenarios where 1000 units were querying in radius, at first i naively used overlap, then a simplest grid possible, and grid ran at 200 fps vs 20

now i use mix of grid and spatial hash, and i can go to 100k at 60 fps on a single thread, in a pure scenario without transforms

transforms are quite expensive

[viral flame]

Well, the overlap sphere itself wasn't the issue, it turned out to be the sorting which is weird, and at the radiuses I'm expecting to use, I'm not too worried about it (for now), though I may end up doing something smarter if required i.e. space binning or some quadtree, though I would assume they do that in the physics engine internally, but who knows

The raycast was the other expensive part, but I fixed that (mostly) by limiting it to the single hit, since ANY hit is "obstruction"

I should have SS the profiler, but again, something like 98% of the time was in the sort, which is wild, but ... .net strangeness I suppose

[plucky laurel]

you can cache the positions and ditch transforms

write your own sqrmag and other simple api calls (angle) with agressive inlining

should batch the whole thing tbh, at end of frame or some sync point, cache positions, update the buffers, next frame they have the info

[viral flame]

Yeah, it's fine for now with slight time slicing, can render 400+ npcs at 60fps, which is more than I'll ever need for the type of thing I'm doing. thanks for the advice though!

[stuck onyx]

@undone coral what i'm doing is to create a texture from all my city in 3D taking a capture, then i set that as texture to create a big map of all my city. I do that to create a satellite view, kid of a google maps if you like....

but the problem here is we dont know why the times of execution of a coroutine are so different using deltatime, unscaledDeltatime, and unscaledFixedDeltaTime

private IEnumerator SatviewEnterFaderCoroutine(bool enable,Action onCompleted)
    {
        float duration = enable
            ? configService.varsCfg.satViewEnterFadeInTime
            : configService.varsCfg.satViewEnterFadeOutTime;
        float start = enable ? 0 : 1;
        float end = enable ? 1 : 0;
        sceneFader.gameObject.SetActive(true);
        float lerp = 0;
        Stopwatch sw = new Stopwatch();
        sw.Start();
        while(duration>0 && lerp<1)
        {
            lerp += Time.deltaTime/duration;
            sceneFader.alpha = Mathf.Lerp(start,end,lerp);
 
            yield return null;
        }
        Debug.Log("SatviewEnter Time " +enable+" :"   +sw.ElapsedMilliseconds);
        sw.Stop();
        sceneFader.alpha = end;
        sceneFader.gameObject.SetActive(enable);
        onCompleted?.Invoke();
    }```

this is the coroutine....
the steps are

1. Execute coroutine with enable=true (fades to white)
2. Once is done i do the Camera.Render to create the texture
3. Once is done i execute the coroutine step 1 with enable false (fades to trasparent)

step 1 and step 3 take different times if we use Time.deltaTime

a 0.3 sec difference approx

if we take step 2 out of the equation coroutine always take same amount of time of course

@sly grove was taking a look to this yesterday too and didnt find too much sese

sense

[versed pewter]

Hello, I am trying to do a HTTP post from C# into PHP.
I have used this example: https://www.php.net/manual/en/features.file-upload.post-method.php
And have successfully uploaded files on my website to my server.

If I try to send a json file with the C# WebClient, the response from the server is; "Possible file attack" or "Failed to open streams: permission denied".

I have tried to change the folder permissions trough a FTP client and trough Chmod but this still does not work.

Anyone here familiar with anything like what I am implementing?

[cosmic basalt]

dont use webclient it's old as hell

use httpclient

who knows what that thing's doing

[covert aurora]

Hey

👋

[versed pewter]

Okay, will try with httpclient.

[plush lion]

Hey, I'm trying to use the FreeImage external library in an Android build. It works fine on the editor, but when I build the dll goes missing and I get a dllNotFound exception through logcat.

I tried to force Unity to include it with a linker, the resources folder, and even tried to include it manually in the apk after the build (lmao), nothing worked. Anybody got an idea?

[tough tulip]

Camera.render might take a very significant amount of time. And Unity works on time slicing. There's no way you can get a consistent timing. Delta Time is accumulated throughout the frame from start to end and is applied on the next frame at beginning.

From what I gather, something like this is happening at your end. It's just an example and are not at all representing the actual timings from your case.

Second last frame before loop ends:
value of lerp = 0.9 or something else behind 1. Throughout this frame, your Time.time remains same.

Last frame:
Suddenly your deltaTime has a large spike and crosses way above 1f. Say 1.2f which you see as the error in your measurement. It's totally fine and you can't avoid it.

Though I'm curious about the practical issues you are facing because of the unexpected error

[long abyss]

Hello, i've created a shader graph and I want to show it "fullscreen", how do i do that?

[fresh salmon]

I don't see the relation between this question and the channel you posted in (programming). If you're asking how to make a window full-screen, ask in #💻┃unity-talk. If that's specific to the shader graph, #archived-shaders.

[long abyss]

Oh ok sorry, I'm not really familiar with this ds server so i didn't know where to ask, thx

[thick thorn]

is anybody familiar with the performance implications of [SerializeReference]? I have a List<BaseClass> that I've annotated with [SerializeReference] and i notice that there is a significant slowdown in play mode when i have that list expanded, but the slowdown goes away when i unexpand the list.

when i had the underlying class viewed in the inspector without [SerializeReference], I didn't get the slowdowns, so it does seem to be that annotation

and the profiler shows that the massive slowdowns are definitely in the editor loop.

[thick thorn]

maybe this is more or an editor question...

[undone coral]

i feel like you have greatly overcomplicated this effect. also, camera.Render() is synchronous

[undone coral]

you should really not use SerializeReference

[thick thorn]

for performance reasons?

[undone coral]

try to solve your problem without it

no, just in general

[thick thorn]

why?

[undone coral]

it doesn't do what it advertises it does

[thick thorn]

can you explain further?

[undone coral]

it looks like it can solve problems like "lists of differently typed objects" but it can't

it does solve that problem, but very poorly

how many different subclasses of baseclass do you have?

[thick thorn]

right now just one

i'm just testing this out

[undone coral]

and how many do you think you will have?

[thick thorn]

maybe 4 or 5

[undone coral]

then the best thing to do is

[Serializable] class T {
 ...
}

class Container<T1, T2, T3>
  where T1 : T
  where T2 : T
  where T3 : T {
  [SerializeField] private List<T1> t1s;
  private List<T2> t2s;
  private List<T3> t3s;
  ...
  public IEnumerable<T> allTs() {
   return t1s.Concat(t2s).Concat(t3s);
  }
  ...
}

or, use odin

because the inspector just doesn't interact well with this sort of model

usually when you have a list of things that can be different types, you're doing something wrong anyway

so what if the types all have the same base class? it isn't as useful as it seems

because you tend to have very few subclasses

it makes sense when it's List<UIElements> and there may be hundreds of different UI elements

it makes less sense for gameplay stuff you're writing yourself, when invariably the number of subclasses is like, 1-5

i would recommend just using odin because i bet you are messing with serialize reference

for the purposes of getting the inspector to work and nothing else

[thick thorn]

yes that's correct

[undone coral]

SerializeReference is for something else entirely

so just use odin

[thick thorn]

what's it for?

[undone coral]

making it easier to get useful inspectors

without writing custom inspectors

[thick thorn]

isn't that what you said above?

[undone coral]

i'm saying odin gives you exactly what you want

[thick thorn]

oh i mean...

[undone coral]

an inspector for a list of subclass instances without having to write a custom editor

[thick thorn]

what's serializereference for?

[viral schooner]

I have a question. I am trying to destroy an object when it is hit by light, colliders are not working, and I do not want to raytrace every instance in the game, as these are enemies that multiply at respawn once destroyed.. any ideas?

[undone coral]

it's for serializing objects when you don't know the type ahead of time. for example, if you're creating a Localization package and you have a class Localizable<T>, and you'd like to serialize an instance of something like X : Localizable<string> but at the time your code is written, you don't know that

[thick thorn]

ahh

well anyway the easiest solution here is probably make the objects derive from monobehaviour and just add each as a component to a gameobject

[undone coral]

people use it for "I would like to serialize an IBlah"

yes, that definitely works

but Odin is what you really want

which is a better inspector

[thick thorn]

i'll look into it. good deal now i guess

i've sort of known about it for a long time but never used it

mostly because i couldn't find a list of its capabilities

[undone coral]

carefully look at what the inspector says when you have the .dll and .so files selected for android

is this a purpose built unity library?

[thick thorn]

i don't really understand your assertion that a list of base classes isn't useful. for example, i use it to polymorphically execute actions that other entities add into a queue. not sure how i'd do that without list<UnitAction>

[undone coral]

in unity, you should use unitywebrequest

[sage radish]

What Unity version are you using? I saw the same thing, I think in Unity 2020.3, but it was fixed in Unity 2021.2 or something like that.

[thick thorn]

2019 something. Guess I should upgrade…

Actually lemme check…

oh yeah it's 2020.3f.1f1

so i guess i'm upgrading

[sage radish]

I wish there was a place I could search through all Unity version changelogs. I'm trying to find mentions of SerializeReference in recent changelogs.

[plush lion]

It's a library completely independent from unity. I just have a .dll, not a .so. But it is set to load on all platforms.

[undone coral]

isn't freetype a c++ library? you should have an .so file

otherwise this will just fail

[plush lion]

It's FreeImage, and I checked on their site, no way to get a .so with my .dll

Is this the reason it fails ? ^^"

[undone coral]

the thing you downloaded is only for windows

better question: why do you need freeimage?

what 1 image format are you trying to read

you would need to build freeimage specifically for android

which is straightforward but arcane

[versed pewter]

Will definitely look into that, thanks

[thick thorn]

sorry, can you just explain this a little more? i use this pattern relatively frequently for polymorphism and i just want to understand if i'm doing something wrong.

[undone coral]

got it

you're doing everything fine

what i really mean is

[regal olive]

I do open source collaborative game dev. Paid assets are kind of a drag for the community.

[undone coral]

the super brief version is, almost always you're doing a list of data instead of objects

[regal olive]

It’s good for proprietary, closed team though.

[undone coral]

unity's scene is data, it's not like, a piece of runnable c# code

it's not an instance of a class

they pretend the stuff you drag and drop into the inspector is instances of a class

it is clunky to turn polymorphic c# objects into data. you always wind up with a "type" field somewhere

SerializeReference ships that type field for you, but in a way that has surprising consequences

polymorphism makes the most sense when you're designing something that's meant to be extensible by other people, or has behavior that is unknowable

it's less useful when you're building a game, because you sort of know 100% what everything will do. the polymorphism turns into an organizational feature, not something essential to getting the code to work

concretely, what do you do with this advice?

usually when i want something to appear in the inspector without causing a huge headache, i make a struct and stuff all the aspects of it into there

[regal olive]

If T1, T2, and T3 are asserted as being of the lineage of T, it seems more simple to have the container be a container of type T.

[undone coral]

it can for sure. i think the thing i'm writing already exists in some library

for small numbers of types known ahead of time

so for example

like tell me what you are trying to store?

like enemies?

you want to describe enemies?

or something

[thick thorn]

it's basically a terrain tile type. or a terrain tile feature type

actually it's a controller for those things

[thick thorn]

I think he's just saying for the instance where you want the things to appear into the inspector and you have a small number of types, it's easiest to have them appear in the inspector by just having a list of each type

it's kind of unfortunate that the unity inspector fights so hard with the basic nature of c#

[plush lion]

I'm loading a shit ton of photos at once, and to not freeze the application I use some code to load images async but it's dependent on FreeImage

[thick thorn]

like c# has all these elegant solutiosn for problems but the inspector just says no

[regal olive]

The code referenced above looks a bit off. I probably don’t have a ton of context to say for sure, but you do! Just surfacing this as an opportunity to reevaluate the desired traits of the code you’re producing. If you kick the tires and find everything okay, that’s awesome!

For me, I try to look out for overly generic code as it’s a sign I’m conplexifying code, or trying to buy abstractions prematurely? Not trying to be a jerk or anything, just trying to help

[regal olive]

Icic. It’s to buy the ability to see it in the inspector. That makes a little more sense!

[undone coral]

yeah this async texture loader package?

did you try just using unitywebrequest to load images

@plush lion https://stackoverflow.com/questions/52357990/read-an-image-file-asynchronously

Texture texture = ((DownloadHandlerTexture)uwr.downloadHandler).texture;

[undone coral]

usually you want to do something of the form

class Map {
  public Tile[] tiles;
}
struct Tile {
  int type;
  // stuff that is common to
  // all tiles
  int elevation;
  // thing that is interpreted
  // differently by the tile
  // types. for example, maybe
  // value means the height of
  // a tree tile, or the depth
  // of an ocean tile, or the
  // number of berries on a 
  // berry tile
  int value;
}

you can still have

class GameTile {
  Tile tile;
}
class BerryTile : GameTile {
  int startingBerries => tile.value;
  int currentBerries {get; set;}
}

[tough tulip]

The problem with a texture loading is that even if you create a texture on a different application you need to marshal that texture data from Free image library to Unity in the end.
UnityWebRequest handles texture very well in that aspect. I was able to load more than 200jpgs without any issue on mobile devices.

The main challenge comes when the texture is being uploaded to GPU. That has always been a bottleneck in engines because that upload process is synchronous and will stall your main thread. There are techniques to avoid it for ex texture streaming. It loads a texture in lower mips and then gradually raises up the mips. It's quite evident in most cases as you can see the texture popping up. There's also SVT in hdrp. This will although work with only asset bundles or assets built with project (excluding Resources)

in your case you should be able to load images without much stalling. Male sure you correctly manage the image loading and uploading part. don't try to load all textures at the same time. load them one by one using UWR

[hollow wing]

(solved)
hi having trouble with HashSet<..> with a custom MonoBehaviour

it seems the Contains method does not work as expected

in my case when there are 2 components in the HashSet, and I call Contains with one it sometimes does not return true

I checked GetHashCode that the custom MonoBehaviour is definitely in there

Is there something I'm missing here with HashSets and Components/custom MonoBehaviours?

oh well I think I made a mistake thanks 😄

[silent rivet]

Does unity have a way to view logs sent to Console.Error?

[merry robin]

@silent rivet https://docs.unity3d.com/ScriptReference/Application-logMessageReceived.html

if you're trying to read Console.-anything you'll need to use the Console callback methods

[plush lion]

Alright thanks I'll look at it

[proud iris]

An external C# DLL library in my Unity project does not seem to be protected/exempted from the Low Managed Stripping Level from WebGL. It functions correctly when compiled for Windows (Mono) but not for when compiled for WebGL.

I get this error.

I have a link.xml file in the project root assets folder. Though, it doesn't seem to do much.

<linker>
    <assembly fullname="FridayNightFunkin">
        <type fullname="FridayNightFunkin.Json" preserve="all"/>
    </assembly>
</linker>

[cosmic basalt]

@proud iris are you using il2cpp rather than mono?

[proud iris]

For Windows, I use Mono. For WebGL, it seems to forcefully use IL2CPP, as the Managed Stripping Level can't be lowered to None/Disabled.

[cosmic basalt]

try changing it to mono in player settings -> configuration -> scripting backend

it might be because of the aot compiler not playing nice with external libraries

[proud iris]

WebGL does not support Mono.

[cosmic basalt]

oh huh i thought you could do either on it

[thick thorn]

you were right -- upgrading unity got rid of the slowdowns.

Well.. now it seems to pack all the work into one frame… not sure what’s going on… but seems to be better most of the time at least lol.

[tough tulip]

If you know the type you're constructing before compilation you can make some dummy object which implements that type. Not sure if that would solve your issue though

[wet sail]

hello, setting AnimationClip.legacy does not work in Build (has no effect at all), this has been a problem (at least for me) since unity2020, Can someone check if it works for you? if not i'd file a bug ?

[sly grove]

Report it as a bug.

[wet sail]

is it same for u?

[sly grove]

I haven't tried it.

[wet sail]

Can you do that?

[sly grove]

But if it's as clear-cut as you say, should be easy to make a simple repro project

I don't have time.

[wet sail]

alright

[dapper girder]

Hey, I'm trying to bake a navmesh from a custom mesh. That mesh does not belong on any game object, which means I manually have to create a NavMeshBuildSource at the very least. Does anyone have any idea of how to do that? The documentation isnt really helping me

[dapper girder]

despite figuring out these inner workings of unity being like landing a plane blindfolded, I figured it out if anyone is ever interested

[regal olive]

What’s the solution?

[dapper girder]

Working with the navmesh surface package on git; if the surface navmeshdata is null you generate it via NavMeshBuilder.BuildNavMeshData and set it; if you're updating an already existing data, you update it via NavMeshBuilder.UpdateNavMeshDataAsync set it

If anyone needs concrete details just ask here

[regal olive]

That makes sense. Your initial post hinted at the navmesh not being baked at edit time.

If I understand the solution, it sounds like you figured out how to bake the navmesh at run time.

[dapper girder]

Yeah, not only that, but using any random meshes you might want to and for any number of different surfaces. Since the regular system won't work for ECS

[regal olive]

ECS? You mean the Entity Vaporware System? Sorry, sorry had to >.> Mostly because of how huge Unity had convinced everyone it would be. And yet, 3-4 years later 😆

[lapis arch]

Hey guys, I have a small "problem" with visual studio code

When I'm defining a field it suggests it to have pascal case naming

for example I type:

private Sprite

And it suggests this name:

private Sprite Sprite;

[shadow seal]

Does it think you're beginning to create a method

I only suggest that because the icon matches the one above Awake

[lapis arch]

Yes

I think it does... Can I change that somehow?

[shadow seal]

Why? Then whenever you try and make a method it won't suggest names either

Plus in this instance, PascalCase is correct, it's a public field

[lapis arch]

well for me pascal goes only to non-fields

doesn't matter the access modifier

I just want to delete name suggestion for methods only, as it doesn't make sense to propose method names anyway

The editor suggests names based on the return type which is useless for me

[livid kraken]

Just like write your code and ignore the suggestion then

You dont have to click it

[shell grove]

Question: An IEnumerator is resettable by calling IEnumerator().Reset().
I store the IEnumerator in a Coroutine variable by calling var Couroutine = StartCouroutine(IEnumerator());
Now, Couroutine.Reset() does not exist. Is there a reason for that?

[fresh salmon]

Because the coroutine (iterator method) with all the neat methods like MoveNext and Reset is wrapped into the Unity Coroutine type that handles all the logic

[devout hare]

StartCoroutine doesn't return an IEnumerator

[fresh salmon]

It returns that wrapper

[shell grove]

Interesting. How can I start the IEnumerators then so that I can later access them again? Since StartCoroutine doesn't work?

[devout hare]

IEnumerator coroutine = MyCoroutine();
StartCoroutine(coroutine);
coroutine.Reset();

Untested, may or may not work
Intuition says it won't

If you want to reset a coroutine properly then stop it and start it again

[shell grove]

A coroutine stops at it's next yield instruction correct?

What I mean is that when I call StopCoroutine("Coroutine") it continues it's execution until it reaches the next yield. Atleast if I remember correctly.

[devout hare]

Functions always run to completion, which is either when the function ends or when there's a return statement. When you stop a coroutine, it stops immediately. But it may have done something before you stop it.

[tough tulip]

this will not work. last i remember this method was not implemented in unity.
the only way to restart a coroutine is to

private Coroutine coroutine;

void Restart()
{
  if(coroutine != null) StopCoroutine(coroutine);
  coroutine = StartCoroutine();
}```

its more convenient to use a wrapper class for this logic with MonoBehaviour and IEnumerator reference if this logic is being used at too many places

[flint wind]

StartCoroutine(coroutine);
Otherwise, the above is what I do too.

[sly grove]

Yes, correct

[shell grove]

Thanks 👍

[sly grove]

I recommend not using the string version of StartCoroutine though

[shell grove]

true

[normal cedar]

A simple bool in while loop should also suffice

[wicked bear]

If I could weigh in on the coroutine situation, IEnumerators can be manually progressed by using .MoveNext() on an IEnumerator, allowing you to progress them via your own triggers rather than a coroutine. So you call your IEnumerator method once to create an instance* of it, then call .MoveNext() on that instance at your own whim.

*instance is a simplification for education purposes only

An old screenshot from how I'm using it, where TeleOrb() essentially acts as a logical wrapper for the TriangleDash IEnumerator.

[sly grove]

Iterator methods cannot be reset, even though Reset() exists on IEnumerator

It's not anything to do with coroutines (which are a Unity concept)

it's due to the fact that C# iterators are not resettable

[shell grove]

huh, that baffles me a bit

so IEnumerators cannot be Reset by calling Reset()?

[sly grove]

Depends on the particular IEnumerator you are talking about

an IEnumerator produced by an iterator method I.e. one that returns IEnumerator and using yield statements, cannot be Reset

[shell grove]

private IEnumerator ColorFade(int btn)
{
    release = true;
    RoundButtons[btn].GetComponent<MeshRenderer>().material.color = colors[btn];
    RoundButtons[btn].GetComponent<Light>().color = colors[btn];
    RoundButtons[btn].GetComponent<Light>().enabled = true;
    var duration = 1f;
    var elapsed = 0f;
    while (duration > elapsed)
    {
   yield return null;
   elapsed += Time.deltaTime;
   RoundButtons[btn].GetComponent<MeshRenderer>().material.color = Color32.Lerp(colors[btn], new Color32(255, 255, 255, 150), elapsed / duration);
   RoundButtons[btn].GetComponent<Light>().intensity = Mathf.Lerp(20f, 0f, elapsed / duration);
    }
    RoundButtons[btn].GetComponent<Light>().enabled = false;
    RoundButtons[btn].GetComponent<Light>().intensity = 20f;
    release = false;
    colorFade = null;
}

for example

[sly grove]

Can you maybe explain what your end goal here is? Resetting a coroutine is not possible.

[shell grove]

This one here

[sly grove]

Is there a reason you can't just call StartCoroutine(ColorFade(n)) again?

[shell grove]

    private IEnumerator FadeOut(int btn, AudioSource master)
    {
        var duration = 1f;
        var elapsed = 0f;
        var btnColor = colors[waveforms.IndexOf((i) => i == currentWaveform)];
        var wfv = wFV[waveforms.IndexOf((i) => i == currentWaveform)];

        if (currentMode == "Sustain")
            while (!release && !btnRelease[btn])
                yield return null;
        if (btnRelease[btn])
            btnRelease[btn] = false;
        while (duration > elapsed)
        {
            yield return null;
            elapsed += Time.deltaTime;
            master.volume = Mathf.Lerp(wfv, 0f, elapsed / duration);
            SquareButtons[btn].GetComponent<MeshRenderer>().material.color = Color32.Lerp(btnColor, new Color32(255, 255, 255, 150), elapsed / duration);
            SquareButtons[btn].GetComponent<Light>().intensity = Mathf.Lerp(20f, 0f, elapsed / duration);
            if (masterReset[btn])
            {
                elapsed = 0f;
                master.volume = Mathf.Lerp(wfv, 0f, elapsed / duration);
                SquareButtons[btn].GetComponent<MeshRenderer>().material.color = Color32.Lerp(btnColor, new Color32(255, 255, 255, 150), elapsed / duration);
                SquareButtons[btn].GetComponent<Light>().intensity = Mathf.Lerp(20f, 0f, elapsed / duration);
                yield return new WaitUntil(() => !masterReset[btn]);
            }
        }
        master.Stop();
        SquareButtons[btn].GetComponent<Light>().enabled = false;
        SquareButtons[btn].GetComponent<Light>().intensity = 20f;
        Destroy(masters[btn]);
        masters[btn] = null;
        fadeOuts[btn] = null;
    }

Currently I'm resetting the internal values back to 0 and halting the progress of the IEnumerator when masterReset[btn] is true

[sly grove]

You can stop a coroutine

Coroutine inProgress;
...
inProgress = StartCoroutine(FadeOut(n, a));
// then later:
StopCoroutine(inProgress);
// And start it again:
inProgress = StartCoroutine(FadeOut(n, a));```

[tender quarry]

When I try to extract textures from a .fbx file from blender, I get the error ```
EndLayoutGroup: BeginLayoutGroup must be called first.
UnityEngine.GUIUtility:ProcessEvent (int,intptr,bool&)

Anywone knows what I can d about it?
I'm using Unity version 2020.3.25f1

[regal olive]

Ha! This is good information. I can see how I could use this in some automated tests.

[modest lintel]

Ignore the error. (It shouldn't affect you)

[wicked bear]

I'm glad it helps out! I spent ages trying to find out how to do it :P

[tender quarry]

But it doesnt extract the textures 😂

[regal olive]

Yea! I’m kinda stoked. I usually test coroutines functions in what I dub “integration” tests indirectly in a play mode test. You pointing this out let’s me think about writing edit mode “unit” tests, I think.

[wicked bear]

im demanding a mention if you post anything about this. I've not done much like this yet.

[flint wind]

I've got events in my Character class for things like onNameChanged and onCharacterKilled and so on. They work just fine when another class has a 1-to-1 relationship with a specific character, like a UI Gauge that's just tracking a specific character's health stat.

The problem I'm running into is that I now have a class that needs to track the onCharacterKilled for multiple characters in combat, so that it can remove them from the turn order if they die. Problem is that onCharacterKilled doesn't send any parameters, so there's no way for the listener to know which character died.
I could update all my events to include this information, but it would be overkill for the listeners that only deal with one instance and already know which one it is.

Should I add the parameter of who is invoking the event to all my events, across the board, to resolve this? Or is there another way?
This will probably affect all the other classes I have events in, not just Character, eventually

[modest lintel]

While subscribing to said events, you could construct an identity of the Character who's event you're subbing to, and pass that into a function which can deal with the actual logic of the listener with the benefit of having the information on which character it is

This is a pretty bad example

But should get the point across

[flint wind]

I was struggling a bit on the implementation of that. But that {} notation is probably what I need.

I really should look into that notation more.
Thank you!

Oh, is there a way to then reference that delegate later and unsubscribe when the character dies?

[modest lintel]

Err, you have to store a reference to that delegate, and then use that to sub and desub

Not the cleanest way really

But, personally I'd just put the character info in the event itself if you are managing multiple characters

or have two sets of events

[flint wind]

I might have to in the end.
Also for things like onStatIncreased, I'd need to send back the Stat itself, and the Character that owns that stat, along with the amount

Since Stat doesn't have a reference to its owner, in order to avoid a lot of back and forth linking

But if I did do it this way, would you recommend:

public delegate void CharacterDelegate(Character character);
private Dictionary<Character, CharacterDelegate> characterKilledDelegateDictionary = new Dictionary<Character, CharacterDelegate>();

And later on...

for (int index = 0; index < activeActors.Count; index++)
        {
            CharacterDelegate characterDelegate = (Character character) =>
            {
                CharacterKilled(character);
            };
            characterKilledDelegateDictionary.Add(activeActors[index].character, characterDelegate);
        }

And then I'd just += and -= off the specified dictionary entry?

Actually, it's running into issues on -=
character.onCharacterKilled -= characterKilledDelegateDictionary[character];
Since CharacterDelegate has a parameter, on onCharacterKilled doesn't.

And +=, actually. I think I'm setting up the delegate wrong

So even though my delegate had the same signature in both classes, it still said it couldn't convert them. So I manually referenced the delegate in Character, and that seems to have got it to work

[regal olive]

A dictionary for the delegate registration/deregistration? What’s going on?

Sorry, slightly distracted. I think I may have just typed out my inner head thoughts out loud. I’m not saying anything is wrong but I did not expect to see a dictionary involved in a delegate assignment

…interesting code now that I’m reading it. Haha okay I’ll keep my thoughts to myself

[modest lintel]

They want to sub/desub anonymous functions. They need said anonymous functions because the parameters in their events don't have enough information

[regal olive]

@flint wind and/or @modest lintel what do you think about inverting the relationship between the two scripts? What if the character game object tell this character, erm, container(?) that it spawned and then when it dies?

I haven’t read enough for heaps of context, but I subscribe to the concept of code “telling other code what it wants to happen.” I’m still reading, so if this suggestion doesn’t make sense that’s cool. I figured I would surface the idea of flipping the relationship as it looks right now (on brief read).

[modest lintel]

That is what is happening right now. Their Character class is telling other classes that it died, spawned, gained health etc. But its not telling the other classes "which" character actually spawned/gained health

[regal olive]

But that interaction is happening based on delegates. It seems like the ‘container’ shouldn’t really care about that?

[regal olive]

🙂

[flint wind]

Some of my code will happily use the event on a specific item, like a Stat, and work just fine, because it's a 1-to-1 relationship. Like a UI health gauge tied to a health stat. It doesn't need to know what stat is firing the event, since that's implied.

But when I scale up to a larger scale and have multiple items to track, a single listener can't make heads or tails of which item is firing the event. Like in the case of my combat class needing to track dead characters, so they get removed from the turn order sequence. Also if a Character needs to track which of its many stats got modified.

I haven't made a call yet on which of the two methods to do:

1. Have the delegates for all the events return the class that is firing it. So an event on a Stat for handling a onStatAmountChanged event would return the Stat and an int.
2. Keep the events as they are, but any class that needs to track multiple instances will have a dictionary of delegates that it subs/unsubs and uses
3. Something else that hasn't been mentioned yet?

[sly grove]

I usually make custom delegate types that pass the event source. e.g.:

public delegate void OnHealthChanged(Character source, int oldHealth, int newHealth);```

The official Microsoft way is something like:

public delegate void OnHealthChanged(object sender, System.EventArgs args);``` But I don't see it as appropriate for games due to extra GC overhead on the EventArgs instances

And some people of course swear by using Action<X, Y, Z> instead of custom delegate types but that's neither here nor there

[flint wind]

I was wondering about that too. My gut was telling me that an EventArgs class being made for every event call was going to generate a ton of garbage. Though i'm no expert on delegates... I was only guessing they don't generate garbage on calls?

[sly grove]

I don't think they generate garbage on calls, but I think the delegates themselves do live on the heap and will be GCed when discarded

[flint wind]

Action<> didn't seem very nice to use. At least with a delegate you can see the parameter names. With Action<> it's just arg arg arg arg arg...

[sly grove]

I agree but it's also not super relevant to the real question here

[flint wind]

True

[sly grove]

e.g. Action<Character, int, int> would be equivalent to my suggestion

[flint wind]

Yeah, I've also read that the compiler treats both Action and Delegate as the same for something like this

[sly grove]

I believe so, yes

it's purely a matter of readability/convenience tradeoffs

[flint wind]

In the case of Stat, it would need a reference to its owner Character, creating a circular reference structure. Just returning Stat in the event would not be informative enough to determine who owns it.

Alternatively I could have the container class for all of the stats on a character handle any outside event requests.
So a Stat returns a reference to itself and the amount with its events, and if something outside of the container class needs to know if a certain stat was changed, it subscribes to events on the container class, not the stat itself.

So event -> intermediate listener -> event -> final listener

[regal olive]

I like how you consider this important. I do too, because I think our code needs to tell a story.

Using project specific language

Er, context aware language?

[flint wind]

I mostly don't want to pick up this project after a hiatus and have to decipher what past me was thinking when I wrote an Action signature 😄

[regal olive]

Ha, true! I use automated testing as my way of offloading human readable context off into the code. Revisiting a 6 month old project and reading what tests I wrote offsets the context degradation a bit.

[flint wind]

I think the best approach in this case is to have the event return the originator. Adding it saves me the hassle of managing a dictionary of runtime-created delegates, and the only price is extra parameters in the delegates involved.

[regal olive]

Okay here’s the message I was missing. Good intuition on pushing back a bit — making a dictionary that keeps track of delegates feels off. Great job on using events!! Yess!

Adding details about what character died into the payload of the event is definitely what I would expect. Even if most listeners don’t care who died, having that sort of information is a good idea. An event should provide a minimal yet enough contextual information to be a story. YMMV

[regal olive]

I mean this message

Also events are real hecking cool for tests

[lofty inlet]

So Application.persistentDataPath can vary based on the OS right? Is there any way to instead save files directly to the _Data folder of the build folder or create a folder inside the build folder and save to there?