#archived-code-advanced

1 messages · Page 42 of 1

cursive horizon
#

ok so your problem is really that your cards have some opaque 'Do' logic which you select from a type dropdown on the card SO, and you need to resolve that to the type properly?

kindred remnant
cursive horizon
#

honestly just do a big pattern match

sly grove
#

Big switch or big dictionary

#

That's all

cursive horizon
#

ya

#

if you wanted to generate that at runtime with reflection i don't think that's terrible (as long as you aren't doing the reflection every time)

#

but i would just do a big switch statement

#

also keep in mind that if you refactor your class definitions, that may well beak your type dropdowns (or old savegames if they also rely on the type)

#

be careful directly serializing classnames

kindred remnant
#

Yeah suppose that would work - I would have to add to the dict everytime I make a card but same issue as what you just wrote

cursive horizon
#

i find it can be nice to have that step

kindred remnant
#

Some sort of CardLookup class that the SO references, and the CardLookup just returns an instance of that card's controller?

cursive horizon
#

it's the step where you map the 'design time' configuration to the 'runtime' entities the game core will use

cursive horizon
cursive horizon
#

but running as a server you'll have different considerations

#

for instance if you might be hosting a bunch of games, you don't need the 'game' to store that lookup

#

you could have it cached somewhere that is shared between all games

kindred remnant
#

I'm new to networking and have no idea how I would cache anything to be shared across all servers

#

but yeah, that sounds sensible

cursive horizon
#

i just meant on a single server between multiple instances of the game

#

but if you do have multiple servers then yeah you could do that too

kindred remnant
#

how would the CardLookup reference the correct card? Lets say the SO passes that card's ID - I would have to hardcode a dict that maps the IDs to the correct type? Is there a better way to resolve that without the ID having to be changed in 2 places if they were to change?

cursive horizon
#

ionno, that stuff gets harder if it's some stored procedure and not just data

rugged pollen
#

I've got a single player game with a bunch of state in the world that accumulates. changes to the world, items you've retrieved, doors, etc. etc. it's organized along acts -- act 1, 2, etc. any pointers on how to create a system for devs to load up an arbitrary act and start playing? my current thought is to run some code that manually configures the world state, but that seems very error prone / likely to become stale quickly

cursive horizon
#
  1. you build at design time or when the game starts if your cards are moddable
#
  1. you have to build at runtime because you can't represent the controllers as data beyond some kind of ID
cursive horizon
#

if your state is easy to update in-place you can build a couple of 'test saves' and you're golden

rugged pollen
#

yea that exists, I could set aside saves at the start of acts -- but then I'd have to recreate them everytime sometime changes in an earlier act 🤔 maybe still the way to go though

cursive horizon
#

it depends how far along you are, cause you have to do that anyway once you have users with your game

#

like if you update the game such that old saves will break, forcing yourself to use old saves to test will also force you to make sure that doesn't happen

#

but then you have to solve that problem which may not be worth it if you're still prototyping

kindred remnant
rugged pollen
#

I'm meaning more, if a dev adds some new puzzle to act 1, then acts 2 & 3 save files will need updated to reflect that the puzzle is solved by that point. I'm not terribly concerned about players in the future. the game will be one-and-done

cursive horizon
#

ah yeah, that's something you need to solve anyway if you plan to update your game ever, but if you don't then maybe it's not worth it

#

if things in are constant flux the easiest may just be to have some 'skip to next act' buttons which you maintain as doing whatever specific things are needed

cursive horizon
cursive horizon
#

like if the game understands that it contains objectives and some objectives are required to reach a given act, you could pretty easily make a 'complete all act 1 requirements' button and have it continue to work

rugged pollen
#

nothing quite as concrete as an "objective" more like, to get from A to B you must watch this cutscene, kill this enemy. from B to C you must get this item from the table and open this door. C to D requires solving this puzzle, etc. conceptually, it's easy enough to write it all out and then update things as needed, but in practice I think people will forget to update the metadata and then the things will be act unpredictably/oddly. that said I'm struggling to imagine a solution that isn't either easy-to-forget metadata or bookmarked save files

cursive horizon
#

well if you make it as concrete as 'objective' then you could

#

if instead your game is built of objectives and objective types with a completion state and some of those types are flagged as required for a given act

#

but it sounds like that would be a significant restructuring, so maybe your time is better spent elsewhere

rugged pollen
#

likely the ship has sailed on that front -- and it seems like that just moves the state tracking down a layer, which might help keep it from going stale I suppose

cursive horizon
#

it codifies the state tracking as a known thing which the program can reason about

#

you could do that other ways too if you wanted

rugged pollen
#

I hear you that I could try to hook into save file upgrading as a possible way to alleviate re-recording stale bookmarked save files 🤔

cursive horizon
#

mark state variables with attributes

#

doesn't sound worth the time to me, but yeah a simple migration system

#

you still have to remember to write the migrations, but you can automate running them for savefiles with an older version

#

maybe you'll be glad you made it when you do ship and then someone immediately finds some small bug which requires you to push a patch which needs to include some state structure changes 😉

past shale
#

Hi guys when i try to export my project to android studio
am missing libil2cpp.so file, and currently getting a crash due to libil2cpp.so not found. Does anyone have any idea how this can be solved?

undone coral
#

...performant or the cleanest - namely, iterating through the card pool and using reflection to grab it's controller
i don't know if it's a performance or a cleanness issue really. you could have tens of thousands of cards, and it will perform fine

#

might be digging deeper into the wrong approach rather than finding a more suitable one
it sounds like you have a server and execute the rules there. if you don't experience pain there, where there would be a lot of bugs, you will be fine in the client. overall the way you are doing things doesn't sound buggy to me, even if it sounds clunky

regal olive
#

Hey, for a multi target system i use Transform[] , anyone has a better variable to do so ?

#

also do you guys think it's optimized enough ?

#

(mods : can we send short videos here ?)

dusk plaza
#

ref ItemForQuestEntry ItemForQuestEntry = ref RecipeForQuestEntry.ElementAt(i); if (ItemForQuestEntry.ItemID==DatabaseEntityReference.DatabaseEntityID) { if (ItemForQuestEntry.Amount <= SplitItem.Amount) { RecipeForQuestEntry.RemoveAt(i); break; } else { ItemForQuestEntry.Amount -= (uint)SplitItem.Amount; } }
Does this just work? I am not sure how local ref variables work

#

The intent is to either remove the entry from the DynamicBuffer or to reduce it's amount
ItemForQuestEntry is a struct and RecipeForQuestEntry.ElementAt returns a ref T

undone coral
#

what are you trying to do

undone coral
regal olive
#

each enemy can be locked multiple times and a projectil can be throwed each time

dusk plaza
# undone coral what are you trying to do?

The intent is to either remove the entry from the DynamicBuffer or to reduce it's amount, I am wondering if the refs make so I don't need to assign it back to the buffer in case it was not removed

undone coral
# 
var x = listOfStructs[0];
x.field = 1;
listOfStructs[0] = x;

is the appropriate pattern for structs. otherwise, declare a listOfClassInstances and use class instead of struct

#

just use class

#

it sounds like you want classes

undone coral
regal olive
#

i got a video to show how it works, but dunno if i can send one here

dusk plaza
undone coral
#

there's a reason dynamic buffers don't support classes

#

this is a DOTS question ight?

#

i think maybe put it in the forum

dusk plaza
#

not really, it's a c# question, but the context is DOTS

undone coral
#

because i don't know

regal olive
#

isn't it overdoing to create a class for DynamicBuffers ?

dusk plaza
#

I am asking how the 'ref' qualifier modifies the behavior of the code. If I just wanted the code working I could just assign it back later

regal olive
#

i think your question is too much specific, you gotta find that out yourself

undone coral
stiff hornet
regal olive
#

i think that's what he really want to know

stiff hornet
# 
        /// <summary>
        /// Gets the reference to the element at the given index.
        /// </summary>
        /// <param name="index">The zero-based index.</param>
        /// <returns>Returns the reference to the element at the index.</returns>
        public ref T ElementAt(int index)
        {
            CheckWriteAccess();
            CheckBounds(index);
            return ref UnsafeUtility.ArrayElementAsRef<T>(BufferHeader.GetElementPointer(m_Buffer), index);
        }
#

It is a ref return

fresh salmon
regal olive
dusk plaza
stiff hornet
fresh salmon
kindred remnant
# undone coral > might be digging deeper into the wrong approach rather than finding a more sui...

thanks for the reply. what makes this design clunky out of curiousity?

separate Q - I went through some of your projects on github the other day and some of your solutions are very well thought out. is this typically a result of careful planning, or are there huge refactors involved once the codebase is developed, hindsight 20/20 and all that? I always get paralysed at the start of a project, wondering if my design choices will give me the modularity I need, without over-engineering everything. I'm trying to get better at balancing that.

echo timber
#

I have a problem with cosine here. It's a simple helicopter lift&thrust calculation, and I need the cosine to check the helicopters angle. The problem is that the cosine is periodic, even though the eulerAngle.x stays at -45 degrees for example. How could I fix this?

image.png
#

Nevermind! I had to use radians instead! Sorry guys!

lethal badge
#

hey, it seems my scriptableObject here isnt working correctly. Unity inspector shows be one value, but when i open the file as text, it shows an older one.

image.png
#

has anyone seen this before?

empty shell
#

I feel like this is simple, but I don't know how to google it. What does this code give as an answer:

int integerAnswer = (intToUse< 5) ? intToUse+ 1 : 0

I have seen this in some code online.

fresh salmon
kindred remnant
fresh salmon
#

Put your condition in the first part of the expression. If it's true, then it returns the value after the ?. Else, the value after the :.

empty shell
#

Ok, that sounds interesting.

fresh salmon
#

So the answer depends on the condition intToUse < 5.

empty shell
#

Thank you very much @fresh salmon and @kindred remnant ! ❤️

kindred remnant
keen cloud
#

whats the best way to have an editable list of objects before i use a datatable lol

brisk pasture
#

why is not just a list on a ScriptableObject or component not fine?

timber flame
#

For this, do you prefer to implement different factory (zenject) class or not?
There are different buildings. They have same components as well as different ones. There is a prefab for each building.
To create a building through prefabs, the input is level (int) and definition of that building (stat/data).
Definitions are SO. For each building type, a definition type exists.

Building1: BaseBuilding
Building2: BaseBuilding
Building3 BaseBuilding

BuildingDefinition1:BaseBuildingDefinition
BuildingDefinition2:BaseBuildingDefinition
BuildingDefinition3:BaseBuildingDefinition

I can have only one factory class for all and passing int level and base definition to Create method. Then, inside each class, cast base definition to the desired definition type and initialize components based on.

// Building1
public void Initialize(int level, BaseDefinition definition){
    var buildingDefinition = definition as BuildingDefinition1;
    //...
}

What is your suggestion?

obsidian flare
#

How do you go about video streaming in Unity?

#

And do you decide to take streaming outside of unity into native app and render unity on a canvas?

regal olive
#

What could be the reason for terrain not being rendered in runtime (in build), and how to debug it? It seems bounds are correct, but im not sure which terrain/terraindata variables should i check to know if everything is ok
more info:

  • bounds are fine
  • I already activated all GOs and monos
  • terrain.drawHeightmap and terrain.drawInstanced set to true
  • terrain.materialTemplate not null
  • terrain collider's collisions are working
    My guess would be that the mesh is there, but the visual data (as in the color of terrain etc) is either missing or failing to render
dusty wigeon
dusty wigeon
# timber flame For this, do you prefer to implement different factory (zenject) class or not? T...

The responsibility for the creation of the building should be in the definition. No need to add an extra layer with a factory. Something like Instantiate that will fill the required dependency/source data, but on the Scriptable Object does the job. Otherwise, you are over engineering your architecture in my opinion.

That being said, I don't really know how Zenject works. I'm not really fan of Injection Dependency in Unity, I prefer to use simpler but as efficient method like ServiceLocator.

craggy spear
regal olive
blazing verge
#

Hey, I need to replicate the main mechanic of this game:
https://play.google.com/store/apps/details?id=com.playstrom.dop2
Basically, I need to erase a part of the sprite on touch. What would be the best approach?
I know that I can use Texture2D.SetPixel to set the pixels of the image, but this results in low performance for big images. Is there another approach?

pale salmon
#

Hey Im having an issue understanding unity authentication

So I use authentication with steam, I got my user Web API key from here https://partner.steamgames.com/doc/webapi_overview/auth as described in the docs, I got the error 401 "unable to validate token", so I tried again asked my mate to create a key of his own, same result.

Then I went to an old project of mine where I still know it worked, copied the key from there, and now it works. Thing is: I have no idea where that key is coming from, can anyone explain me why this happens and maybe tell me how I can get it to work with my own key?

undone coral
#

it will give you a client ID. or sometimes you get a secret.

pale salmon
#

I dont think I understand what you mean

I have a game ID on steam and I have an API user token, thats the only 2 unity asks me for, now the problem is the API token, because as I said it works with the token of an old game

I dont know what you mean by associate my game with the dashboard, all I can do is create a new AppID, but that I did and is not the problem

onyx blade
#

Does anyone know how I can deduce a working build app not starting on mobile? It worked yesterday

#

Is there an easy way of doing this? Its crashing instantly and I can't get any kind of info from the mobile

flint sage
#

Grab the device logs

#

Not just unity logs

undone coral
onyx blade
undone coral
#

or you can paste the token to me in a DM, and i can decode it and look at it for you

#

you can paste the decoded token output here

flint sage
undone coral
#

from payload

#

don't paste the actual token

onyx blade
undone coral
#

you can remove your name from the Payload data if it's in there or whatever

flint sage
# onyx blade installing package and rebuilding

adb logcat -s Unity ActivityManager PackageManager dalvikvm DEBUG is a decent start, if that still doesn't show anything, just do no arguments and output it to a file (adb logcat > test.log)

undone coral
#

use your best judgement

pale salmon
#

I dont know why the key itself would be important tbh I mean I think its safe to assume steam wouldnt give me a faulty token twice

undone coral
#

nothing else should be sensitive

flint sage
#

You don't need the package, in fact the package is bad

#

Use the command line

warped flicker
#

Quick question about Dots. What happens if the entity I get this aspect for doesn't have Grid? Do I get an error message?

image.png
undone coral
#

it's not a key

#

there might be a iss or aud fields that are relevant

#

more likely than not you misconfigured something

#

it will tell you in this token

pale salmon
#

They literally call it a key

undone coral
#

and giving it a try

pale salmon
undone coral
#

hmm

#

that's definitely not what i'm asking for or what unity expects

onyx blade
undone coral
#

so i think maybe start fro the beginning on a tutorial

pale salmon
flint sage
undone coral
#

Then I went to an old project of mine where I still know it worked, copied the key from there, and now it works.
i'm not sure what is going on. like i siad, you have misconfigured something. this is consistent with a misconfiguration

pale salmon
#

Also unity does expect a key

image.png
undone coral
#

this thing is a client id and a secret... but i think you will sort this out

#

i would start from the beginning

pale salmon
#

the site I sent you is what unity expects Im fairly sure

undone coral
#

from a tutorial

onyx blade
#

most likely im dumb

pale salmon
#

its not an ID

flint sage
# onyx blade

Just go to the android sdk build-tools folder intead and run it there

undone coral
#

401 "unable to validate token" is telling you something about the token, not the key

flint sage
# onyx blade

You should add the build-tools folder to your path environment variable so that it works everywhere but 🤷‍♂️

onyx blade
flint sage
#

Stupid companies

onyx blade
#

I agree

pale salmon
#

I get the token from the key with this code:

Callback<GetAuthSessionTicketResponse_t> m_AuthTicketResponseCallback;
HAuthTicket m_AuthTicket;
string m_SessionTicket;

void SignInWithSteam()
{
    // It's not necessary to add event handlers if they are 
    // already hooked up.
    // Callback.Create return value must be assigned to a 
    // member variable to prevent the GC from cleaning it up.
    // Create the callback to receive events when the session ticket
    // is ready to use in the web API.
    // See GetAuthSessionTicket document for details.
    m_AuthTicketResponseCallback = Callback<GetAuthSessionTicketResponse_t>.Create(OnAuthCallback);

    var buffer = new byte[1024];
    m_AuthTicket = SteamUser.GetAuthSessionTicket(buffer, buffer.Length, out var ticketSize);

    Array.Resize(ref buffer, (int)ticketSize);

    // The ticket is not ready yet, wait for OnAuthCallback.
    m_SessionTicket = BitConverter.ToString(buffer).Replace("-", string.Empty);
}

void OnAuthCallback(GetAuthSessionTicketResponse_t callback)
{
    // Call Unity Authentication SDK to sign in or link with Steam.
    Debug.Log("Steam Login success. Session Ticket: " + m_SessionTicket);
}```

And this code is poasted straight from unity docs
flint sage
#

Anyways, you can just execute the command from the build-tools folder instead

undone coral
#

and it will give you an illuminating piece of diangostic information

#

i'm not saying put the key in there

#

which in more technical terms is known as the secret

#

anyway

#

you will figure this out, good luck out there

pale salmon
#

You do realise there is a reason Im here right?

undone coral
#

i'm sorry, i have no insights as to what is misconfigured

#

it sounds like you have an earlier project that works. maybe continue working from there

pale salmon
#

The current project works too

undone coral
pale salmon
#

Just with the key of the old one

#

And I have no idea where that key comes from

undone coral
#

but i'm not sure if there's a group policy for that... pretty nuanced thing lol

onyx blade
#

well, this doesnt work either

#

or int he folder above it

pale salmon
#

Well there gotta be a reason it works with some random key I found and not with the key that has a reason to exist

onyx blade
#

just nope

pale salmon
#

Its literally like I bought a new lock and it opens with my old keys but not my new ones

onyx blade
#

I did have a message about jetifier earlier but I am trying to build for android 8 which doesnt have that

#

ah uh, this is probably relevant...

image.png
undone coral
# kindred remnant thanks for the reply. what makes this design clunky out of curiousity? separat...

separate Q - I went through some of your projects on github the other day and some of your solutions are very well thought out.
i try to improve architecturally in a lot of stuff that i make - like to try one new thing. so many games, zero people play, or will have a meaninglessly small audience, so it's really worth it to learn something.
is this typically a result of careful planning
only very rarely. the monster match game was planned. all of its gameplay is so straightforward, you can contain what the whole game does in your head. for the minigames, same thing - they are so small, experiences fewer than 10 minutes, so the whole thing can fit inside my head. otherwise for something new, better to get something working and fun, and then rewrite it if there's an audience.
or are there huge refactors involved once the codebase is developed, hindsight 20/20 and all that?
this is the case with spellsource. the client, whose source code you cannot see, was rewritten after the first prototype. it found an audience so it was worth doing correctly. otherwise i spend a lot of time researching pre-existing implementations, wherever i can find them.

some games like FPSes, retro platformers, and fighting games have so many traditions that it's not worth changing anything. the architectures are all very tightly coupled to things like user input on purpose, they have a very specific UX players expect, and writing that from scratch means reinventing a ton of details that end users are really sensitive to. i'd probably use Unreal for an FPS because it has all those things.

balancing
imo there's no hard and fast strategy to how you write something in the beginning. unless the game follows a very specific tradition, you should write the fun and innovative thing first before anything else.

undone coral
#

use the android sdk it ships with

onyx blade
#

I did that, but it didnt want to build

#

so I am using the API for 9.0 Android 28

undone coral
onyx blade
#

its been a while, let me just run a build and provide the output

undone coral
#

these aren't really advanced programming questions by the way. a lot of people suffer through misconfigured editors trying to build for android

#

and then you try a bunch of random stuff to fix an error you didn't understand

#

you will have to google the error and do exactly what it says

#

and comprehend whether or not it worked for people to decide if you do it

onyx blade
#

tbf, I fixed this problem 3 weeks ago. then the office decided upgrading to windows 11 was a good idea

# 
Picked up JAVA_TOOL_OPTIONS: -Dfile.encoding=UTF-8

FAILURE: Build failed with an exception.

* What went wrong:
Could not determine the dependencies of task ':unityLibrary:com.voxelbusters.essentialkit.androidlib:generateReleaseRFile'.
> Failed to install the following Android SDK packages as some licences have not been accepted.
     build-tools;30.0.3 Android SDK Build-Tools 30.0.3
  To build this project, accept the SDK license agreements and install the missing components using the Android Studio SDK Manager.
  Alternatively, to transfer the license agreements from one workstation to another, see http://d.android.com/r/studio-ui/export-licenses.html
  
  Using Android SDK: C:\Program Files\Unity\Hub\Editor\2023.1.0a20\Editor\Data\PlaybackEngines\AndroidPlayer\SDK

* Try:
> Run with --stacktrace option to get the stack trace.
> Run with --info or --debug option to get more log output.
> Run with --scan to get full insights.

* Get more help at https://help.gradle.org

BUILD FAILED in 2s

UnityEngine.GUIUtility:ProcessEvent (int,intptr,bool&)
#

Im assuming this is probably my issue

#

To fix that, I had installed the Licensing Library via Android Studio but that doesnt seem to work

#

I just reinstalled it and it doesnt even ask me to accept the license

undone coral
#

but basically something is messed up when you installed the sdk

#

normally you are prompted to accept the license in unity hub at installation time

# 
cd 'C:\Program Files\Unity\Hub\Editor\2023.1.0a20\Editor\Data\PlaybackEngines\AndroidPlayer\SDK\tools\bin'
.\sdkmanager --licenses

this will also work from Windows Powershell

#

but this will not resolve the error completely

#

i believe one version of unity ships with build-tools;30.0.2. i assume you have not modified your build.gradle

#

this is a regression

#

i don't think it has anything to do with windows 11

#

what version of unity are you using

#

oh

#

it's right there

#

why are you you using 2023 alpha?

#

well that pretty much settles it

#

you should be using 2021 LTS

#

clearly you had installed, previously, a more appropriate version of unity

#

do you use source control?

onyx blade
undone coral
#

you can look at ProjectSettings/ProjectVersion.txt, or let hub tell you the appropriate version

#

hmm

#

do you use source control?

onyx blade
#

I ll make a backup and load up a different version in the lts

undone coral
#

like git?

onyx blade
#

I do

undone coral
#

then no need for backups right?

onyx blade
#

technically true

undone coral
#

💯 true

onyx blade
#

I can always roll back

undone coral
#

this is code advanced

#

so you installed the version that hub said huh?

#

and he chose 2023 alpha

#

the project is likely not going to be complicated enough for this to matter

#

do you have any meaningful uncommitted work?

#

if so, you will want to commit it

#

then do git clean -dxf in order to remove the pre-existing Library folder and other dirt that may be in there

#

if you git clean -dxf it will also remove uncommitted files everywhere

#

and if you're like, well i'd rather create a duplicate folder

#

it's your prerogative

#

you're going to have to go into the deep end of the pool some day

#

but yeah, i wouldn't use an alpha version of unity

#

and expect things to work

#

or you have to do this stuff a million billion times faster

#

like trying to accept the license the way i described

onyx blade
#

Reading the project documentation did mention some features that were being used from the 2023 version of unity but those might not even matter

#

regardless, I should probably get to an lts as well as fix the license but that could happen during that migration process

undone coral
#

is this a VR game?

onyx blade
#

AR

undone coral
#

and why does it have to be on android?

#

is that the device you own?

#

or is it for an android based HMD

#

there's a lot going on i apologize

onyx blade
#

this is both android and iOS. with priority and testing on android at this time

undone coral
#

do you know what feature it needs in 2023 concretely?

onyx blade
#

related to anchoring of ui elements in UITK. but not much specifics are mentioned

undone coral
#

that seems so low value

onyx blade
#

I doubt these differences even matter cuz there is very little ui relance on anchoring. most of it is vertical stacking

undone coral
#

anyway the game at some point did build for android right?

#

or as far as you know, never

onyx blade
#

before we updated to win11 it worked fine

undone coral
#

the update to windows 11 is a red herring

onyx blade
#

yes and no

#

pc's got reset

undone coral
#

i agree that there might be some configuration that hasn't been reproduced yet

#

it could very well be this powershell command

#

that is seemingly taking you forever to do lol

#

that you could have done while we were writing here

#

i am just razzing you it is fine

#

try the command and see what happens

smoky kite
#

Hey guys do you know if there is a way to measure and evaluate energy cost in unity?

onyx blade
#

yeah, most of the reason I'm just typing is cuz I'm currently making food atm. I entered the command before and didn't work, assuming I was in the wrong directory

undone coral
#

okay that's understandable

#

i mean this stuff is frustrating

#

while it does say the error message, and it does indeed match something in google i parlayed to you

#

there's going to be other feedback in the terminal about whether or not the command executed successfully

#

it's going to be a lot of little details

#

personally, if i were your people, i wouldn't waste a single iota of time on android users

#

even if it's a branded AR experience

#

and this is all about CPMs

#

even if it's a 3d scan of a wendy's hamburger

#

and presumably android users would eat that garbage

onyx blade
#

Im tempted to agree, we are a b2b and it's not exactly my choice sadly

undone coral
#

lol

#

i mean zero people will use this thing on android so it might as well be buggy as hell

onyx blade
#

client wants AR, and we told him it's not a thing anyone cares

onyx blade
#

it has 50 downloads

undone coral
#

lol

#

50 downloads from bots

onyx blade
#

and they are a company with 1000+ employees

#

which is both funny and sad

undone coral
#

okay well when you're ready try figdgeting around iwht that command and reply here

undone coral
#

you can use the macOS / iOS energy visualizations as a "report" and maybe that would be sufficient for your use case

onyx blade
#

Will do. it will probably fix it and I can move to the next issue. then round up for the day as this is all unpaid overtime hehe

undone coral
#

it has about as much meaning as Lighthouse's Accessibility score

#

which is meaningful to people who don't know anything

#

really depends on what your objectives are

smoky kite
undone coral
smoky kite
compact ingot
smoky kite
kindred remnant
# undone coral > separate Q - I went through some of your projects on github the other day and ...

i spend a lot of time researching pre-existing implementations, wherever i can find them.
...
some games like FPSes, retro platformers, and fighting games have so many traditions that it's not worth changing anything

how do you find the material you need? in my experiences, games are far less open source than other tech, and I find it really hard to know where to start looking if I want to research genre-specific implementations/common architecture.

Thanks for your response - very insightful 🙂

undone coral
#

so like the ultimate fighting engine is a good attempt at fighting games in unity

#

a lot of traditional games people try to do multiplayer, which seems so hard to me

#

and unity is really poorly suited to realtime multiplayer games

#

there might, essentially, be no good architecture for them in unity. like networked FPSes in unity can only really be done to the T of the tradition using DOTS, which is like using a whole different engine

#

some things which don't have a big multiplayer thing going on, like metroidvania platformers, there are assets that are like 90% of the way there

#

like the MoreMountains 2.5D Corgi Engine is as flexible as you can get while still feeling like Mega Man movement

#

for a platformer

#

and it's a very very popular asset

#

the source code is open

#

and the moremountains guys you can talk to directly

#

there's other stuff like graphics solutions that also have a lot of traditions. jason booth is very responsive on his discord

#

if you want to learn about How to Do Terrains

#

if it's not these big realtime traditions there are some gems out there

#

the spellsource backend, which is open source, is a fork of metastone, a hearthstone simulator

#

metastone was inspired by xmage

#

a lot of simulators are inspired by OpenTTD

#

if you wanted to make a civilization-esque game, FreeCol is very well written

#

a really well written and coherent java game

kindred remnant
#

Wow, thank you so much

#

This is above and beyond what I expected! very helpful

undone coral
#

also has a very responsive maintainer

#

if you like point and clicks there's a very capable asset in the asset store, but it's really idiosyncratic... almost TOO traditional

#

it works almost exactly like the scummvm / LucasArts tools

kindred remnant
#

These are just resources you've built up over time as a part of the game dev community?

undone coral
#

i think so yeah

#

i haven't really ever had to ship anything with a huge team so that has affected my opinion a lot

kindred remnant
#

Awesome. Well, thank you for sharing

undone coral
#

this stuff makes a lot of sense if it's 1-3 people engineering. which is how a bunch of these old traditional games were made anyway

#

it's still Within Scope to make a multiplayer FPS as 1 person, it just takes a lot of time

kindred remnant
#

I mean, that suits me more also. Zero chance I'll be switching industries anytime soon

#

If you ever need a hand with web dev, let me know 😂 🫡

undone coral
# onyx blade

also if you want to reduce your pain as a windows developer and you have a macos background

#

you can download busybox for windows, saving it to C:/Windows/sh.exe, and then run sh -l for "Unix in Windows command line"

onyx blade
#

I avoided the entire unity installed SDK and went through the Android Studio version entirely

#

reinstalled java, set up the path variable (that was appearantly also missing) by asking our sysadmin

#

they are going to roll out this change to all pc's in the network so I can have access to our render/build pc to speed up work under load

dusty wigeon
regal olive
#

Is it true that we can code in c++ now?

lost stag
#

does anyone know a good double-precision vector/matrix/transform library for unity (esp something which implements something like a virtual transform (not tied to a gameobject)?) - I couldn't find one and am contemplating rolling my own (I'm aware of the issues with doubles and performance, we need them because of reasons)

sly grove
#

basically, just use Unity.Mathematics in general

lost stag
#

oh nice. I was mislead by the unity mathematics intro page which seemed to suggest it was just float4, float4x4 etc

slender bolt
#

Hi everyone, I have a GameObject with two scripts, A and B, on it. A is a parent class of B. A has an OnMouseDown() function. When I click the GameObject, OnMouseDown() calls twice because "This event is sent to all scripts of the GameObject with Collider."
I only want OnMouseDown to call once, but I need the functionality of both scripts. Other than copying all the code I need from A and putting it in B so I can make B no longer a child class of A, what solution is there?

thorn flintBOT
#
Posting code

📃 Large Code Blocks
Large code blocks should be posted as links to services like:
https://gdl.space/, https://paste.ofcode.org/, https://hatebin.com/
https://paste.myst.rs/, https://hastebin.com/

📃 Inline Code
Surround code with three backquotes. Not quotation marks.
To get C# formatting the first line should only contain cs or csharp.
Add a comment with a line number if there is an error message.
```cs
// Your code here
```
Do not share screenshots of code unless requested.

muted root
#

Hi everybody, I'm trying to figure out how to calculate the scroll sensitivity of a scrollRect so that it moves by an amount of pixels.

#

For example, I'm creating a fixed height scrollRect with a fixed width of 1000px. I put 15 items inside meaning the width become 1500px but it only shows 1000px. I want the mouse scroll to move 1000px at once so that the second page is displayed

#

How would I calculate the required scrollSensitivity? I can't figure out the relation between the sensitivity value and the pixel value. Plus I read online Mac and Windows interprets this value differently?

slender bolt
stark acorn
spark forum
#

Hey everyone, I'm having a rather surprising problem with Jobs

#

I have this IParrallelForJob that works on two 4-element arrays (I've made them constant size just for testing), and yet the Execute(int index) method gets called by the jobs system with the index parameter being higher than 3

# 
[BurstCompile]
public struct InScanConeCheck : IJobParallelFor
{
    [ReadOnly] public NativeArray<BeaconData> Beacons;
    [ReadOnly] public SightSensorData Sensor;
    public NativeArray<ScanResult> Results;
    public void Execute(int index)
    {
        Debug.Log($"{index} / {Beacons.Length} / {Results.Length}");
        if (!Beacons[index].Alive || !Sensor.Alive) return;
        (...)
    }
}```
#

the debug log consistently returns 4 and 4 for arrays lengths, and indices exceeding their length coming from the Execute parameter

#

the line below throws exceptions

#

IndexOutOfRangeException: Index 16 is out of range of '4' Length.

#

scheduling the job with different interloopBatchCount values changes nothing in this behaviour

stiff hornet
spark forum
#

here you go

# 
        // Set up buffers and verify which beacons are in range and within scan cone
        var activeBeacons = new NativeList<BeaconData>(128, Allocator.TempJob);
        for (var i = 0; i < _beaconData.Length; i++)
        { // copy active beacons from the beacon buffer into a shorter list
            if (!_beaconData[i].Alive) continue;
            activeBeacons.Add(_beaconData[i]);
        }
        Debug.Log($"Active beacons list length: {activeBeacons.Length}");
        var results = new NativeArray<ScanResult>(activeBeacons.Length, Allocator.TempJob);

        var job = new IsBeaconInScanConeJob
        {
            Beacons = activeBeacons.AsArray(),
            Sensor = _sensorData[_currentSensor],
            Results = results
        };
        var jobHandle = job.Schedule(_sensorData.Length, 4);```
#

there's some array copying voodoo here, that i wrote trying to narrow down the problem

#

but the entire thing is rather straightforward

plucky laurel
#

anyone used csx in recent unity versions? share experience

stiff hornet
#

as that will be the max index (-1) since you have set that as the job length in the Schedule

spark forum
#

well

#

thankfully this is just me being blind

#

not jobs malfunctioning

#

let me just verify it works

stiff hornet
#

yeah, you are probably wanting to use activeBeacons.Length or results.Length instead

spark forum
#

and naturally it works

#

sorry for making a rubber duck out of you

#

and thanks for spotting the obvious

stiff hornet
#

hehe, you know for next time to always check that length now :p

ashen yoke
blazing verge
dusty wigeon
icy aspen
#

I have this very weird thing going on with my scriptable objects. Whenever I load data on them and quit play mode, there's two situtations:

  1. When I view my scriptable object in the inspector, the data in the scriptable object doesn't get deleted/cleared.
  2. When I don't view my scriptable object in the inspector, the data clears
    Does anybody know how I can fix this problem because it's extremely annoying for the players in my game.
stiff hornet
#

it is WAD that you can change stuff in them in playmode in the inspector and have it still have the change on exit

icy aspen
#

But how should i do that

icy aspen
icy aspen
stiff hornet
#

working as designed

icy aspen
#

and how do i store immutable data on my scriptable objects?

stiff hornet
#

what are you using them for?

icy aspen
#

to save ghosts when playing levels in my game

#

so the player can see its best run

stiff hornet
#

as a proper save system? like exit application and reload when opened again?

icy aspen
#

yes

stiff hornet
#

scriptable objects created at runtime in builds do not get saved to disk

#

a save system is not a use case for scriptable objects

icy aspen
stiff hornet
#

yes it works in the editor (as you can save the SO's as assets)

#

but not in a build

icy aspen
#

oh ok

#

so should i use json files then?

icy aspen
stiff hornet
#

yeah or something similar

icy aspen
#

ok thanks

stiff hornet
#

just in case you want a source

#

`When you use the Editor, you can save data to ScriptableObjects while editing and at run time because ScriptableObjects use the Editor namespace and Editor scripting.

In a deployed build, however, you can’t use ScriptableObjects to save data, but you can use the saved data from the ScriptableObject Assets that you set up during development.`

thin mesa
#

then explain the issue

#

and maybe try googling it first too

#

good luck getting answers. i won't be baited into writing your code for you. and if you had googled it you would have found the perfectly valid 12 year old thread on the forums that explains how you could do that

lime plover
#

Is having a 2d overlapcircle running on update more/less expensive than a 2d collider?

thin mesa
plain abyss
#

So after getting told in #💻┃code-beginner that we won't write it for you you thought the advanced section would be more receptive to just do it for you with no effort. Good luck with that

dusty wigeon
#

Try ChatGPT. Should be able to help on basic question.

fresh salmon
#

Eh no, especially for beginners

#

It's sometimes right, but it's sometimes very wrong

#

And you can't distinguish between the two

dusty wigeon
#

When the question is basic, it is able to be more right than wrong.

fresh salmon
#

There's still a risk a beginner can't take

dusty wigeon
#

Also, it can help in figure out the general idea of the script.

#

I mean, they either work from nothing or work for something that isnt perfect.

#

At that point, you try it. If it does not work or you are not able to make what you want, you try something else.

fresh salmon
#

And if it works, will the user learn anything about the code they blindly copy-pasted? Keep in mind that when you're new to something, you don't have that ability to determine whether the thing it spat in your face is right or wrong

#

And that's why it's not allowed to ask or answer questions using unverified AI output here

#

Anyway it's off-topic for this server

#

Discussion about whether an AI is right or wrong doesn't belong here

dusty wigeon
#

They can learn. I have use it multiple time to study subject where I'm not really good. You can ask to explain each part of the subject. You then can cross verify with your notebook.

It's not about the AI being right or wrong. It's about if referring someone to an AI is a good or bad idea. I have not seen any rules saying that we cannot refer someone to an AI, if there is such rule, I will obliged by.

fresh salmon
#

You then can cross verify
You're forgetting the context

#

A beginner will not verify because they're confident in the answer being right

#

It's formatted so well it looks correct, even when it isn't

native beacon
humble leaf
#

It's somewhat implied by the rule of not allowing AI code questions and answers. If it needs to be explicitly explained that guiding beginners to use it is also not allowed, then we can update the rules.

dusty wigeon
#

It should, if it is something that you do not want to see, as it is not clear for me that guiding someone to an AI agent is something wrong or is the same as using AI generated answer as it's own.

native beacon
#

Well, you are basically answering their question with an untested answer.

#

That isn't really GOOD even if it wasn't AI.

dusty wigeon
#

For me, using an AI to learn is an alternative way of searching the internet. It is true that the AI might induce you in error, but being cautious (like you should always be given any source), you can get relatively good answer. Even more when your question is simple or when people refuse to answer your question for any reason.

It is not true that I'm answering a question with an untested answer because, as I stated earlier, I have use AI like ChatGPT to learn from a subject that I do not know a lot about and found multiple answer that helped me. In other words, I know that ChatGPT could be a potential way of finding an answer and that by using it someone could potentially found an answer. It is the exact same as saying google it.

native beacon
#

If it is the same, then why not say to Google it?

It is not the same at all. It strips out attribution and accountability from answers, even when it is correct.

undone coral
kindred remnant
#

Might be super elitist of me, but days like this kind of make me wish we had a more exclusive channel for #archived-code-advanced. I usually enjoy reading through some of what's been shared here - even if I'm not involved in the conversation at all - but it can be pretty hit and miss. Today was defo a miss

cold mural
#

Can anyone explain how does this flow shader work?
Specifically, how do they came up with the values for making the texture flow appear curved?
As far as I know, the shader is adding/subtracting time from the UV coords to make the texture 'slide'.
But I don't understand how to calculate the values for the UV coords to make the texture in a desired shape.
Are these just interpolations across the edges?
This is the tutorial - https://catlikecoding.com/unity/tutorials/hex-map/part-8/#7.2
This is where they make the shader - https://catlikecoding.com/unity/tutorials/hex-map/part-6/#8
(I've made the shader in Shader Graph)

Flow_Map_03.PNG
Hex Map 8

A Unity Hex Map tutorial about adding support for open water. Part 8 of 27.

Hex Map 6

A Unity Hex Map tutorial about adding support for rivers. Part 6 of 27.

mortal gust
cold mural
#

Like, what will a UV value of (0.5, 1) do?
I tried testing it but it gives usually confusing results, I can't quite understand

mortal gust
#

When you are saying specific UV values which part are you referring to? The UVs are created to go from 0 to 1 to be able to wrap and use tiling texture to have continuous visuals.

cold mural
#

This one
Those UV values are responsible for making the texture appear curved

Flow_Map_04.PNG
#

With UV values like second pic, the texture appears straight and flows in a straight direction
With UV values given in the first pic, the texture appears curved and spreads out in a circular direction

Flow_Map_05.PNG
#

This is the shader function responsible for animating the texture

Flow_Map_06.PNG
mortal gust
#

It's still just uvs that are lerped as input to pixel shader? If you look closely you can see the lines between the edges in the lower part of first image.

cold mural
#

It's warping the texture using UVs

#

This is the same shader created in Shader Graph

Flow_Map_07.PNG
mortal gust
#

Sorry I don't fully understand what information you're asking for. Do you understand how UVs are interpolated from vertex to pixel shader? If you keep the same UVs on the same vertex (if you're using indexing this is no issue), you will have seamless texture transition between triangles. They are simply nudging the UVs to get a more circular shape.

cold mural
#

Which values should I use to make the texture appear circular in shape

mortal gust
#

The middle square is right now textured in a regular 0->1 way which you would expect the image to come out unwarped.

#

Possibly some sqush in Y direction.

#

Presuming triangles are the same shape, the outmost vertex on the side are 0.5 offset on UV.X from the ones next to them to continue the default visual.

#

Now you have something like this

image.png
regal olive
blazing verge
#

You're the first person I know who says c# is worse then c++

cold mural
mortal gust
#

Now the idea is to pull the lines so it makes more of a circular shape.
E.g. to round the lines in the side triangles you move the UV in the side vertex UV.Y "upwards"

image.png
cold mural
mortal gust
#

Yes, some number above 0.8, depending on what angle you want the lines.

cold mural
#

What about the lower coords

regal olive
#

Does anyone have idea how to convert the following code to the mathematics API:

# 
  axis.Normalize ();
  angle *= Mathf.Deg2Rad;
  Vector3 torque = axis * (angle * s.spring) + s.damper * deltaAngularVel;
  return torque;
cold mural
regal olive
#

The math class doesn't really have good documentation and I couldn't find a way to multiply quaternions or find delta rotation between two rotations

mortal gust
sly grove
mortal gust
#

I was a bit confused at this at first why the middle point wasn't further down instead but realized the default space was based on the middle vertex^^

regal olive
cold mural
#

Thank you, I've been at it for a couple of days now trying to understand how it worked

sly grove
cold mural
dusty wigeon
regal olive
dusty wigeon
#

And why is the torque in Vector3 ?

regal olive
#

I tried this but doesn't seem to work:

image.png
dusty wigeon
#

Not sure how mul() behave with Quaternion and Vector3 multiplication. It is probably an overload of the operator. That being said, I would try to use a Vector4 with 0 as 4th coordinate.

regal olive
#

Any idea how to multiply quaternion and float3

dusty wigeon
#

The following snipped produce the same results for both method. I have tested it.

        Vector3 forwardOverload = transform.rotation * Vector3.forward;
        quaternion quaternion = math.mul(math.mul(transform.rotation, new float4(0, 0, 1, 0)), Quaternion.Inverse(transform.rotation));

        Debug.Log($"[Overload Operator] {forwardOverload} [Mul] {new Vector3(quaternion.value.x, quaternion.value.y, quaternion.value.z)}");
regal olive
#

Actually I found that there's a rotate method in math class

dusty wigeon
#

Even better

regal olive
#

how can I decompose the quaternion to angle axis

#

But no idea about the axis

#

The angle is fine

image.png
dusty wigeon
#

Rotate it with forward.

#

It will give you the forward axis.

#

If you want up, you rotate by up.

regal olive
#

Yeah but I want it as an angularVelocity/torque

dusty wigeon
#

Do the same as I did. Make a small script with quaternion.ToAngleAxis() and quaternion.rotate(q, new float3(0,0,1)) and compare.

regal olive
dusty wigeon
#

I'm not really sure what axis is ToAngleAxis

sly grove
#

not axisangle

regal olive
#

This retrieves the angular velocity required to go to q2 from q1 in 1 sec

dusty wigeon
#

I see, I was not aware of Angle Axis representation of rotation.

undone coral
regal olive
autumn locust
#

every since im using the Unity Test Framework with playtime tests, i get this wierd error:

NullReferenceException: Object reference not set to an instance of an object
UnityEditor.TestTools.TestRunner.PlaymodeLauncher+BackgroundWatcher.OnPlayModeStateChanged (UnityEditor.PlayModeStateChange state) (at Library/PackageCache/com.unity.test-framework@1.1.33/UnityEditor.TestRunner/TestLaunchers/PlaymodeLauncher.cs:113)
UnityEditor.EditorApplication.Internal_PlayModeStateChanged (UnityEditor.PlayModeStateChange state) (at <434908727417408abd02bbd1a0835ef8>:0)
UnityEngine.GUIUtility:ProcessEvent(Int32, IntPtr, Boolean&)

it causes the scene to get reset every time i change something + run the scene

#

error only appears after first running a playtime test

errant plinth
regal olive
errant plinth
#

Ohh right, the reverse. Missed that, thanks!

icy aspen
#

which data path should i use in builds
because Application.datapath doesn't work
System.IO.Directory.GetCurrentDirectory()
this one neither

fresh salmon
#

It definitely "works", whatever that means

#

It changes, depending on the platform you're building to, but it will return a correct path

#

Though if you need to save stuff to retrieve it in a later run, you should be using .persistentDataPath instead

icy aspen
#

does that mean that i can use folders from the assets folder

fresh salmon
#

No, folders from the Assets are bundled into the executable, they cannot be fetched afterwards

sly grove
#

what are you trying to do? You should have started with that

fresh salmon
#

Use Resources or Addressables to load assets at runtime.

undone coral
#

yeah a lot of this stuff is pretty nuanced

sly grove
icy aspen
undone coral
#

what would be helpful? it sounds like you are creating a free-body character controller of some kind

sly grove
#

And streamingAssetsPath if you want to read json files that are created at editor-time

icy aspen
sly grove
#

No

#

you can never use folders from assets

#

unless they are a StreamingAssets folder

icy aspen
#

ok

#

so this should work to read and write

Screenshot_2023-02-08_at_20.35.36.png
#

right?

sly grove
#

yes you can read and write from persistentDataPath

#

assuming there is something there to read

thin mesa
icy aspen
sly grove
icy aspen
sly grove
#

right now you're trying to write to a directory you're not allowed to write to because you're missing a path separator

fresh salmon
#

(notice the lack of / between Documents and the name of your JSON file)

undone coral
#

there are tutorials online for saving games

#

there are assets

#

they'll cover stuff like paths

icy aspen
sly grove
thin mesa
icy aspen
#

I don't understand any of this wtf is this

#

i just wanna use json

sly grove
#

you are in the wrong channel

icy aspen
#

and now i have to use that

sly grove
#

we're going to assume people in the advanced channel can handle basic docs and code examples

icy aspen
native nebula
#

I'm debating on integrating Firebase into my Unity game to handle a lot of entity data for different tiles, AI characters, and structures. I'd like to have a nice database to keep all of these different entities in the same place that I can easily manage rather than writing a bunch of helper functions, using resources.load, and other stuff. Any thoughts on using Firebase + Unity?

undone coral
native nebula
undone coral
#

i see how a database lets you store data... but unity already has many good, robust ways to author and store data

#

if you choose to not use unity to store the data, you will be recreating all the models to represent that data in unity anyway, so you'll wind up with 2x the schema, and going from 1 to N schemas entails versioning, serialization issues, etc.

#

@native nebula does that make sense?

native nebula
undone coral
#

then you want to use those not-inside-unity-authored assets inside unity

native nebula
#

that is completely not the case

undone coral
#

hmm

native nebula
#

hold on

undone coral
#

well the first question really is, what kind of game are you making?

native nebula
#

So basically, I'm building some editor tools at the moment to help me mass-produce generated structures, entities, etc. for my game. I'm building the structure editor, which will allow me to save structures as prefabs then spawn them in-game with different logic and such. Then I have the tile editor I'm working on, which creates entities as tile types, with properties like price, sprites, name, z-index, etc. These tiles are connected to a construction system, which detects these tiles and adds them to their respective construction category for the UI. I also have a ledger system, which works with the construction system to manage the expenses of some of these tiles and entities.

I'm at a point where adding a new entity or tile can be expensive in terms of code and how much time it'll take. I would love to have a system where I can input properties from some fields (ie the tile editor) then the other systems will feed directly off of that and add the new entities accordingly

#

mostly a lot of system design revamps

#

It's for a coffee shop management game, 2d top-down. you build ur own coffee shop from the ground up and compete with other coffee shops in the city you start in. the goal of the game is to buy out all the cafes in your city, then expand to other cities

#

so I'm considering Firebase as a solution to help contain a lot of properties for a variety of different entities, that I can then call api requests to from these systems and handle them automatically

undone coral
undone coral
native nebula
native nebula
undone coral
#

adding firebase to your nice sounding coffee shop sim to manage static data

#

is going to make you hate game development

#

i don't disagree that it would work

#

it's going to be 10x worse faff though than whatever you're doing now

#

it sounds like you are more comfortable making automation tools using whatever firebase expects you to do, perhaps in a different programming language, than in c#

native nebula
#

I need to really seperate and hash out these systems first of all

undone coral
#

i get that. but then just make the tool directly on your computer

native nebula
#

hold on

undone coral
#

and do not involve firebase at all. if you want to emit e.g. templated prefabs, they are just ordinary text files.

native nebula
#

lemme show you the top-level view for the systems i have / plan on having

undone coral
#

if you really don'tw ant to build editing tools in unity, don't. you don't have to

#

okay

native nebula
#

@undone coral systems placed in sectors

ExpressoEspresso_Systems_4.png
undone coral
native nebula
#

okay cool

undone coral
#

i would just caution that simulator games tend to have "fake" decoupling

#

it really depends on how complicated of a game you want to make

native nebula
#

yeah, atm i have a clearly defined scope I'd like to achieve

undone coral
#

if it's more Tiny Tower than it is Puerto Rico

native nebula
#

what

undone coral
#

have you played tiny tower and/or puerto rico?

native nebula
#

ive played tiny tower

undone coral
#

okay in a lot of simulator board games

native nebula
#

im thinking of the island territory

#

lol

undone coral
#

not just puerto rico

#

yeah

#

it's a board game

native nebula
#

i see

#

then no

undone coral
#

but you've played sim board games. like settlers of catan has action cards

native nebula
#

yeah

undone coral
#

if you want behavior like that, your game is tightly coupled

native nebula
#

do you wanna hear about the scope btw

undone coral
#

sure

native nebula
#

so the scope is set in a city, with minimal information such as city laws and demographics that afffect employee pay and customer needs/wants

#

i mean there's not really much else to go off of that

undone coral
#

okay well all these games wind up with a crazy ants architecture

#

my honest to goodness suggestion for you - and i rarely say this - is to paper prototype your sim

#

so that you can limit the implementation to the interesting ideas you actually have instead of totally ...

native nebula
#

well it's already released on steam

undone coral
#

oh

native nebula
#

lmao

undone coral
#

well then what the fuck am i talking about then lol

native nebula
#

well I'm quite literally overhauling everything

undone coral
#

hmm

#

this is a good place to be

native nebula
#

atm it's just about ur cafe

#

and u building it

undone coral
#

standby i'll brb

native nebula
#

I'm widening the scope

#

aight

#

is there an npm equivalent for c# and unity

undone coral
native nebula
#

i c

undone coral
#

for unity specifically there is a format similar to node packages

#

for c# there is nuget, which is hard to use in unity

native nebula
#

not the package asset window thing right

#

just so we're clear

native nebula
undone coral
#

adding things to your manifest.json

undone coral
native nebula
#

uh

#

there's kind of a similar game out there called Espresso Tycoon, however, that game is 3d and unreleased

undone coral
#

basically there are four approaches to making simulators in unity:

(1) gameplay is built over time, decoupled, via GameObjects and scripts on them, with a bunch of Manager classes. tightly coupled game behavior happens on various singleton classes. (it sounds like you are here. rimworld for example)
(2) separate game state (in pure C#), game logic (in pure C#) and a decoupled, via gameobjects and scripts, approach to rendering the game state (what most people migrate to. this is how my games usually work)
(3) ECS with Game Objects
(4) full DOTS

native nebula
#

i thought about using dots at the beginning

#

seemed like a pretty viable option

undone coral
#

well you shipped a game

#

and most people who use DOTS in the beginning don't ship games

#

ECS with game objects is a shippable compromise

#

the full DOTS is really hard

native nebula
#

this game is def. similar to rimworld in anything but game premise lol

undone coral
#

(2) and (3) basically differ in how you write the game state and game logic. in both (2) and (3) you are writing game objects with *View scripts on them, that query the game state and render stuff

#

so you get to use the really robust unity rendering featureset you are familiar with eithe rway

native nebula
#

what option do you think has the "easiest" game-scaling opportunities?

undone coral
#

it's just a matter if you want unity's specific approach to expressing game state and game logic or if you want to do it in pure C#.

undone coral
abstract basalt
#

performance doesn't just mean "scale up"

undone coral
#

for example, NativeArrays

#

yes

#

that's why i'm saying in terms of scaling up to more sophisticated gameplay, i would write it in pure C#

native nebula
#

yeah sorry, for scale up I mean adding a shit ton of new entities haha

undone coral
#

instead of ECS

#

i hear you. like item types

#

consider that Cities Skylines predates ECS and still scales really well

native nebula
#

yeah, shop items, tile data, entity properties

undone coral
#

yeah

#

so i would prob choose 2, and i think you sort of already have

#

one approach you should consider is

abstract basalt
#

Sure, Data Oriented Design (DOD) predates ECS. It's just a ecosystem that (tends to) make the default way of implementing something the DOD way

native nebula
#

data orientated design 🥴

native nebula
abstract basalt
#

I forget myself sometimes, that's why I have to delete half my messages

native nebula
native nebula
undone coral
# 
class Item : GameEntity {
 public async UniTask Use(Context context) {
  this.quantity--;
  context.FireGameEvent(new BeforeItemUsed(this));
  await OnUse();
  context.FireGameEvent(new AfterItemUsed(this));
  if (this.quantity <= 0) { context.FireGameEvent(new ItemDepleted(this)); }
 }
 protected abstract async UniTask OnUse(Context context);
}

class CaffeinateWorkers : Item {
 [SerializeField] private float speedMultiplier = 1.2f;
 [SerializeField] private float durationSeconds = 30f;
 override string description => $"Increase worker speed by {(speedMultiplier - 1f)*100)}% for {durationSeconds} seconds";
 override async UniTask OnUse(Context context) {
  var workers = context.GetAllEntitiesByType<Worker>();
  foreach (var worker in workers) {
   worker.speed *= speedMultiplier;
  }
  await context.GameTimeElapsed(durationSeconds);
  foreach (var worker in workers) {
   worker.speed /= speedMultiplier;
  }
 }
}
native nebula
#

yeah i have something similar to that already

undone coral
#

okay

native nebula
#

youre using decomposition right

undone coral
#

then i think you are in a really good place

#

sort of... you can see that the text of the special items is written in C# code

#

i'm not programming inside the inspector

#

so if you had some one time use item that made your worker speed 20% greater for 30 seconds, you want to do that in C#

#

you don't want a "IncreaseSpeedEffect" and a "DurationEffect" or whatever

#

just write it in code

#

you could have those things, but then you're creating a LISP, and as you'll see in an earlier discussion i've written into this chat, that's hard to do and your main obstacle

native nebula
#

I have a script called EntityBase that a lot of different entities inherit, then properties change. ie:

EntityGrass:

EntityGrass : EntityBase {
   void start(){
      SetEntityName("Grass")
      SetEntityZIndex(0)
      SetEntitySprite(spritePath)
      SetEntityIsTerrain(true)
   }
}
undone coral
#

if you wind up having a lot of items that modify worker speed, you can progressively modify this script

undone coral
native nebula
#

ill show u the entity structure later once im not afk

#

if you want

undone coral
#

just think about how an async UniTask abstract method can be used to let you author the spell-like behavior directly.

#

it makes sense to me right away

native nebula
#

this just gave me an idea

#

"player manages a cafe selling magical potiond to medieval fantasy monsters"

#

that sounds insanely fun

undone coral
#

lol

#

so yeah, i think stick to (2)

native nebula
#

yeah

undone coral
#

no firebase since that requires LISP to work well

native nebula
#

well

undone coral
#

like you can try to lispify the contents of OnUse

#

that's hard

native nebula
#

i might analyze my current system structure and see what i can abstract and simplify

graceful forge
#

Hello! I am trying to make a border shader for hexagons that use decals. I am using URP and the issue is the decals are rendering below water which is in a transparent render queue. I have tried forcing the water shader to use depth write, but the decal still does not respect it. I want the decal to project ontop of a transparent material. Any ideas on how to get the decal to do that?

frail parrot
#

Quick question (hopefully): I have an A* pathfinding program. I want to run it both directions, so that if one of the routes returns that there isn't one quickly, I can stop the other so it doesn't take too long searching all possible options.

Basically I want 2 coroutines and as soon as 1 returns empty I want to stop the other, but if neither returns empty I want to get the shortest one.

How can I solve this? I tried looking online but either my coroutines aren't actually going Async or it just doesn't stop the other, it still halts for a moment when I click an unreachable spot.

#

Pretty sure you shouldn’t post your question to multiple channels

obsidian glade
frail parrot
#

of course, one second

#

this is my current A* code

public static IEnumerator FindShortestAStarAsync(TileManager tiles, Hex start, Hex end, bool flipped = false, System.Action<List<Hex>> callback = null)
    {
        List<Hex> route = new List<Hex>();
        Dictionary<Hex, float> priorityQueue = new Dictionary<Hex, float>();
        List<Hex> checkedHex = new List<Hex>();
        //Hex retrieved from Hex
        pointers = new Dictionary<Hex, Hex>();
        var startPoint = start.cubeCoordinate;
        var endPoint = end.cubeCoordinate;
        var current = start;
        var maxDist = CubeDistance(startPoint, endPoint) + (end.transform.position - start.transform.position).magnitude / 10;
        pointers[start] = null;
        priorityQueue[start] = maxDist;
        while (current != end && checkedHex.Count < priorityQueue.Count)
        {
            current = FindLowest(priorityQueue, checkedHex);
            checkedHex.Add(current);
            if (current != end)
            {
                priorityQueue = checkNeighbours(priorityQueue, current, end);
            }
        }
        if (checkedHex.Count < priorityQueue.Count)
        {
            route = GetRoute(current, flipped);
        }
        if (callback != null)
        {
            tiles.priorityQueue = priorityQueue; 
            callback.Invoke(route);
        }
        yield return null;
    }
obsidian glade
scenic forge
#

Yeah it's CPU bound, putting it in a coroutine doesn't magically make it not block main thread.

frail parrot
#

i thought as much, it didn't seem to do anything asynchronous

#

wdym CPU bound?

scenic forge
#

The limiting factor is how fast the CPU can run through the algorithm.

obsidian glade
#

a "quick fix" is to do some amount of work and then yield for a frame, perhaps you let this coroutine run for 50 iterations per frame

#

it means you'll need to wait for it to finish, but it will allow two coroutines to run for 50 iterations each frame each, and cancel the other when one returns finished

#

shoutout also to the UniTask plugin

frail parrot
# 
int count = 0;
        while (current != end && checkedHex.Count < priorityQueue.Count)
        {
            count++;
            current = FindLowest(priorityQueue, checkedHex);
            checkedHex.Add(current);
            if (current != end)
            {
                priorityQueue = checkNeighbours(priorityQueue, current, end);
            }
            if(count % 50 == 0)
            {
                yield return new WaitForEndOfFrame();
            }
        }
#

so something like this?

scenic forge
#

Basically yeah you would have to insert stop/resume points into your algorithm, and communicate between the two and cancel if one completes.

obsidian glade
frail parrot
#

oh okay so that part was correct

#

alright! just that change seems to have fixed it

scenic forge
#

There are issues with doing it this way though, particularly if the algorithm takes a long time, then you would also want to split it across multiple frames. On weak machines if you only allow it to run a short amount of time per frame (so to maintain FPS) then it will take a lot more frames to complete.

#

The best would be to offload it to a different thread so it can go as fast as it can while not blocking main thread, but often that's not a trivial solution.

frail parrot
#

in this case the algorithm shouldn’t take too long. the area isn't so large that it should cause problems

forest jacinth
#

Hello everyone, i have searched everywhere, talked to chatGPT, but i cannot get the thing i want..
I want an Inventory system where i can store my ScriptableObject spells, and then be able to drag a Clone of the spell onto another panel, and then be able to use the spells from that panel. Basically the same as how World of warcrafts spell system works.
If someone knows how to implement this system and guide me, il pay 30 euro for it.

#

im hella desperate

midnight violet
forest jacinth
#

in game UI

#

@midnight violet

#

or if it would be like Minecrafts inventory so i can use the spells on 1-8 keys

midnight violet
#

Then you have to create a class, that gets filled with the data from your scriptable object. But the fact, that you ask this AND you tried using chatGPT lets me assume, you are not at all related to an advanced problem and rather should take a look at some unity basic tutorials. Check unity learn about UI and also scriptableobjects and craft your knowledge to put the things together

forest jacinth
#

i just thought that this was an advanced problem so i'd ask advanced coders even though im a beginner

frail parrot
# scenic forge There are issues with doing it this way though, particularly if the algorithm ta...

so, it took some testing but now I realised that if it surpasses 50 iterations, it actually doesn't return a solution at all. It most likely has to do with how i'm doing the callbacks, so what am I doing wrong here?

public static List<Hex> FindShortestAStar(TileManager tiles, Hex start, Hex end)
    {
        List<Hex> res = new List<Hex>();
        tiles.StartCoroutine(FindShortestAStarAsync(start, end, false, returnValue =>
        {
            res = (res.Count > 0) ? ((returnValue.Count > 0 && returnValue.Count < res.Count) ? returnValue : res) : returnValue;
            if (returnValue.Count == 0)
            {
                tiles.StopAllCoroutines();
            }
        }
        )); 
        tiles.StartCoroutine(FindShortestAStarAsync(end, start, true, returnValue =>
        {
            res = (res.Count > 0) ? ((returnValue.Count > 0 && returnValue.Count < res.Count) ? returnValue : res) : returnValue;
            if (returnValue.Count == 0)
            {
                tiles.StopAllCoroutines();
            }
        }
         ));
        return res;
    }
  }
midnight violet
midnight violet
frail parrot
#

am I returning before getting an answer, since it doesn't have an answer after the first frame in that case? and if so how do I fix that?

forest jacinth
#

gotcha, well thank you for your time

frail parrot
# midnight violet Can you explain your issue?

so, I explained above, I have an A* algorithm, but I want to start it from both sides so that I can know faster if there is even a path possible, because usually one side is smaller so if the smaller one returns a path of 0 then there's no path possible, and I can stop the larger algorithm earlier for more efficiency. I'm doing this by using 2 coroutines, which are started in the function above. Right now i'm introducing a delay of 1 frame after every 50 searches so that it doesn't run synchronously

#

however, if it has to search more than 50 neighbours, then i don't get a response at all

#

it always returns empty

midnight violet
#

The problem here is, you start an async task (I guess because of the name) but immediately return your res which is a empty hex list, right?

frail parrot
#

yeah I think so, but I don't exactly know how to solve for that

midnight violet
#

make your findshortestastar async itself for example

frail parrot
#

and then? won't this still need 2 frames and thus not return anything on time?

midnight violet
#

The problem is, you want a) something in time, which would be mainthread, but that would block it but b) also try to use a coroutine which can always take longer than expected

#

You gotta rethink your loading I guess.

frail parrot
#

well okay, actually, i don't have to wait for the answer I think. it only calls the function and doesn't need the route after that. the update function just checks if the route exists, so I could use a callback on the main async function to just set the route whenever it gets a result

midnight violet
#

Is this like being called on ONE list or on multiple? Where do you use that return value?

scenic forge
#

Rather than both algorithms being coroutines and having messy logic coordinating them, one approach is to only have one coroutine that acts as coordinator, each frame executes a small chunk and if either algorithm succeeds, signal the other to terminate; if both still pending, yield for next frame.

#

Or alternatively, you could really look into running them on separate threads.

frail parrot
#

the return value is just stored on a value in tilemanager, which is the read on update to use a linerenderer and draw the route

midnight violet
#

And you got any performance issues with it yet, why you want to call it two times?

frail parrot
#

so I don't have to await the value, I can just make the whole function async and then set the value on the tilemanager with a callback so I don't have to wait for it

midnight violet
#

Yep, thats what I would do, if you do not need it in realtime. Just let the list be empty or the last result until you get a new one from async and in callback do your list path line renderer whatever thing 😄

frail parrot
#

so it would lag out for a couple of frames

midnight violet
#

that sounds like you gotta refactor your actual pathfinding maybe? Instead of trying to chunk it down to smaller parts of the same code? I dont know your code tho 😄

frail parrot
#

in theory this should work, but it still doesn't function if I have to check more than 50 neighbours? the problem is probably somewhere here but I'm stuck on what is causing the wait

int count = 0;
        while (current != end && checkedHex.Count < priorityQueue.Count)
        {
            count++;
            current = FindLowest(priorityQueue, checkedHex);
            checkedHex.Add(current);
            if (current != end)
            {
                priorityQueue = checkNeighbours(priorityQueue, current, end, pointers, out pointers);
            }
            if(count % 50 == 0)
            {
                yield return null;
            }
        }
frail parrot
midnight violet
#

Ever thought about using navmesh?

frail parrot
#

i have, but since it's a homemade hex grid and I've recently learned about A* i wanted to make it myself

midnight violet
#

So what if you put your pathfinding in full async code, like a Task.Run() method that actually runs not on the mainthread, maybe that helps already

#

Ah got it, good approach and attitude 🙂

frail parrot
#

why thank you

midnight violet
frail parrot
#

i did try doing that as well but I honestly just couldn't figure it out

scenic forge
#

Ultimately your problem is just having a CPU bound chunk of code that takes longer than 1 frame but you don't want it to block main thread and affect FPS, so you either spread it out to many frames or run it on a separate thread.

frail parrot
#

yeah exactly, but the spreading out over many frames doesn't actually work currently

scenic forge
#

But there are also other issues like "deciding how much work to do per frame" which will have disproportionate effect depending on how strong the client machine is.

midnight violet
frail parrot
#

so what should the function type be?

#

cause it's not just Task

midnight violet
scenic forge
#

If you are doing the separate thread route, you might also be interested in Burst and Job system.

midnight violet
#

🙂

frail parrot
#

alright! I'll do some research into that more then, thanks a lot anyway

midnight violet
scenic forge
#

If you are using Burst you pretty much have to rewrite your code from what I can see; but on the other hand, the performance gain is incredible (I benchmarked CPU Gaussian blur and using Burst was 200x speedup) so you might not even need to multithread or deal with all these splitting frame shenanigans anymore.

undone coral
muted root
#

Hi everybody, I have a problem where destroying a gameobject with a mesh filter and mesh renderer will also destroy the mesh renderer materials and mesh filter mesh references of the original gameobject.
Meaning I have a gameobject A, use it to instantiate gameobject B. I destroy B, and after that, gameobject A has no mesh and materials anymore...

#

How can I avoid that ?

muted root
#

OK, to answer my own question, the problem seems to come from the trilib asset I use to load assets dynamically. It seems to destroy the reference itself when the gameobject is destroyed. So cloning the references after I've dynamically loaded them fixes my problem. Quite peculiar...

dusty wigeon
# frail parrot so, I explained above, I have an A* algorithm, but I want to start it from both ...

At the moment, you are just doing your AStar twice... You need to redefine the Goal, the FScore and the reconstruction of the path. Also, you should look into how many node you expand before doing anything else. I'm pretty sure you will not gain anything from doing an expansion from both side. In fact, it will probably be slower. If you were doing a Breadth first search, you would gain in performance, but not with AStar.

#

If there is no obstacle, you should get to your target in a line.

native nebula
#

So I'm currently in the middle of overhauling a game and expanding its scope. At the moment it's pretty costly to create new items (entities) for the game and therefore it's slower to produce more content. I analyzed why that is and determined three main problems:

  1. The systems in my game are too dependent on what type of entity it is
  • Some systems require type-checking to know what entity scripts to add and remove
  1. There are too many systems that are dependent on the entity type
  • Adds onto the expensiveness of creating new content
  1. The entity scripts can have different methods that help solve 1 and 2, but not completely
#

Any thoughts on reducing dependency on type-checking?

undone coral
# 
class Entity {
 string name;
 int cost;
 int secondaryCost;
 int value;
 int range;
 int positionX;
 int positionY;
}
#

okay, let's say you had a coffee machine that says, "reduce the maintenenace cost of all adjacent machines by 2"

#

you'd stuff 2 into value*

#

you'd stuff the adjacency of 1 into range

#

what if it's an item "extra large back of beans (makes 60 cups)"

#

stuff 60 into value

#

bUt pOsItIoN iS uNuSeD fOr iTeMs

#

so. what.

#

now you have 1 entity type

#

it's up to you to interpret the fields how you want

#

you can make a very, very simple class hierarchy

#

but when you subclass entity to implement something special, you don't need to add more to an entity's fields, which is all you're going to be interested in anyway. you're going to declare a property like adjacencyRange and it will write to range

#

or int numberOfCups => value;

#

does that make sense?

#

here's other way to think about this

#

in ECS, systems are functions, and entities are function arguments

#

every function can take the same set of arguments. but it's up to the function to interpret them

#

some systems require type-checking to know what entity scripts to add and remove
this sounds like a different kind fo flaw

#
  1. There are too many systems that are dependent on the entity type
  • Adds onto the expensiveness of creating new content
    use LINQ's .OfType<T> on lists. don't overthink this
frail parrot
#

@dusty wigeon @native nebula
As I tried to explain, the A* pathfinding algorithm is already very efficient. The problem arises when we try to reach a point that is unreachable. The algorithm with exhaust every. Single. Location. before realizing it can’t get to the point.

Because we know that if a point is unreachable, their areas don’t connect, we can say ‘in most cases these areas aren’t equal’, so if we run the algorithm twice from both sides, if either of them returns that it’s impossible, we can instantly stop the other algorithm, which is more efficient. But only if it can run both ways in parallel.

If there is a route found, we can use the double algorithm to double check that we have the shortest route and just take the shortest of the two.

THAT is why I’m trying to solve it this way

frail parrot
sly grove
abstract folio
sly grove
#

now if you know heuristically in your game this case is more common, then you are right, this might be an actual optimization

frail parrot
#

So it’s always more efficient to run both

#

But only if they can be done at the same time

sly grove
frail parrot
#

Well okay, that’s true

#

But to that I say, get a better pc

sly grove
#

so I mean, I think you have a decent idea here, assuming certain conditions are true

#

but I'd also guess that there are probably lower hanging fruit than this? 🤔

frail parrot
#

Maybe? I honestly couldn’t find anything online

sly grove
#

how big is this grid?

frail parrot
#

Not set size

#

Currently 25 by 25 but could be anything

cedar moss
#

Anyone herecan guide me on the best workflow to implement collisions on a player controller in a 3D enviroment ? Why was wondering if its with capsulecasts and not allowing the player to move in case they hit something or if there is a better way

#

Considering I would have to have capsulecasts in 4 directions

frail parrot
cedar moss
#

well I have Box Collider component, does it fire an event or something or does it stop things from going through each other automatically

dusty wigeon
# native nebula So I'm currently in the middle of overhauling a game and expanding its scope. At...

Having a class "Character" or "Entity" as a God class is something that will always happens. That being said, you should definitely try your best to divide the class in smaller chuncks. One way to fight this is to follow the Single Principle Rule. [1]

Alright, that being said. What can you do to refactor your code ? Unfortunately, most of the time, the only answer you are going to find is: You just need to divide your class ! But, having suffer this situation a small number of time, I can tell you that you can do more than that.

First of all, you most define the context in which the object act. Those context are called Bounded Context [2] and come from DDD (Domain-Driven Design) which is a practical way of structuring your domain/architecture in OOP (Object Oriented Programming). In your case, that would be the system in which the God Class interact. Something like an Inventory System, Animation System, Loading System, etc.

After, you take one context and you try to isolate it. Only one, this is important otherwise you will lose yourself rapidly and will break your whole project. The first step to isolate the code associated to your context inside your God Class is to create an interface that regroups all the function associated with the system. The second step is to create a bridge [3] between your class and new object that will be use to contain every method/data that is link to the given system. Now you have an object with all the data and method extracted, but you still are pointing to your God Class... (I'm out of words/time, if you want more just ask me and I'll write the remaining later)

[1] https://medium.com/@carlos.ariel.mamani/the-god-object-or-god-class-anti-pattern-bfb8c15eb513
[2] https://martinfowler.com/bliki/BoundedContext.html#:~:text=Bounded Context is a central,being explicit about their interrelationships.
[3] https://en.wikipedia.org/wiki/Bridge_pattern

sly grove
#

Rigidbodies and CharacterController

cedar moss
#

well it does work if I add a RigidBody to my character but it shakes uncontrollably if you are hitting something and not letting go of the key

frail parrot
#

Perhaps its best if you look up some simple Unity tutorials for a character controller

#

Those often show the best practices

dusty wigeon
frail parrot
#

Yeah alright, those are probably the simplest solutions, if a bit limiting

#

It’ll do for now

native nebula
native nebula
frail parrot
#

What exactly do you mean by that? “If you have checks for blockers?”

undone coral
native nebula
native nebula
frail parrot
#

Yeah

#

There can be tiles that aren’t reachable tho

native nebula
# dusty wigeon Having a class "Character" or "Entity" as a God class is something that will alw...

After, you take one context and you try to isolate it. Only one, this is important otherwise you will lose yourself rapidly and will break your whole project. The first step to isolate the code associated to your context inside your God Class is to create an interface that regroups all the function associated with the system. The second step is to create a bridge [3] between your class and new object that will be use to contain every method/data that is link to the given system. Now you have an object with all the data and method extracted, but you still are pointing to your God Class... (I'm out of words/time, if you want more just ask me and I'll write the remaining later)

So what you're essentially saying, is to treat the systems as nodes and the class as a large parent node. The large parent node will have bridges, or paths, that point to the systems..?

native nebula
frail parrot
#

Iirc that’s slower than A*

native nebula
#

it is

frail parrot
#

I’m trying to improve it

native nebula
#

flood fill one of the two separated terrains, if the start and end goal aren't in that flood fill, you know they are unreachable

undone coral
#

you have to do that work somewhere

#

somewhere, somehow, there is a type

#

you can use the type hierarchy to make this simpler

native nebula
undone coral
#

or place the functionality that is calling .OfType inside the type itself

undone coral
#

expensive in what sense?

#

can you give a concrete example?

native nebula
undone coral
#

if you have a lot of switches that go through a dozen types

native nebula
frail parrot
#

I mean, since I’m already precalculating all the neighbours I might as well count how many connected points are in their respective chunks. That way I don’t have to calculate anything. I can just compare the amount of points in the chunks to see if they’re in the same area

undone coral
#

you probably want ot migrate that behavior into the type itself

native nebula
#

for some of the properties at least

undone coral
#

for example, if you have something of rht eform

if (target is CoffeeBag coffeeBag) {
} else if (target is Item item) {
} else if (target is GrassTile grassTile) {
} else if (target is ...
#

yeah

#

is this generally for rendering code?

#

or for game effects?

native nebula
#

Current Process

  1. Create new entity script (ie EntityChair)

    public abstract EntityChair : EntityBase {
    void Start(){
        
    }
}

  2. Complete EntityChair script

    Basic Needs

        void Start(){
        SetEntityName("Chair")
        SetEntityZIndex(2)
        SetEntitySprite(pathToSprite)
    }

    EntityChairs can be rotated from the player with different sprites

    void ChangeSprite(){
    if (EntityRotation == 0){ SetEntitySprite(pathToWestSprite) }
    if (EntityRotation == 90){ SetEntitySprite(pathToNorthSprite) }
    ...
}

    EntityChairs could also be connected to each other

    void CheckNeighbors(){
    if (myTC.Neighbor.west.GetComponent<EntityChair>()){ SetEntitySprite(pathToWestConnectorSprite) }
}
#

@undone coral

undone coral
#

yeah

#

so for rendering code

#

use interfaces

#

for example ITileHasDirectionality for tiles with directionality

dusty wigeon
undone coral
#

so you'll check that type, but not by an exhaustive search

#

rendering problems are very often solved by interfaces

#

this is a Good Pattern

native nebula
#

essentially the current process is this:

  1. Create new Entity script
  2. Initialize properties for Entity in that script (including specific, special methods)
  3. Integrate new entity into other systems @undone coral
undone coral
#

usually i advocate against using interfaces (they typically have 1 implementaiton)

#

yeah

#

the integration you can simplify using interfaces, since you're really talking about how many objects are usefully simulated in a type hierarchy, but the "rendering hierarchy" is very different

native nebula
#

i did have interfaces at the beginning actually

undone coral
#

for example ITooltip

native nebula
#

i had stuff like IEntityName, IEntitySprite, etc.

undone coral
#

so if you want something to show a tooltip, it seems natural to use an interface fo rit

#

since you can hover over so many things

undone coral
#

i wouldn't necessarily do it that way

#

th einterface should be the way it is rendered

#

not just a way to share a property

#

so like ITileView versus an ICharacterView, and entities that ordinarily would implement those interfaces could have an IInventoryView for when those entities are in the inventory

#

so instead of your tooltip code concatenating together the IEntityName.name and IEntityDescription.description

#

declare interface ITooltip {string name; string description;}

scenic forge
undone coral
#

once you realize how to combine them, you can

native nebula
#

So ITileEntity would implement tile properties, ICharacterEntity would implement maybe some tile properties, but more importantly AI controls

undone coral
#

no

#

ICharacterEntity would not do ai controls

native nebula
undone coral
#

stick to its traits, not to recreate a type hierarchy

native nebula
#

gotcha

undone coral
#

like ICharacterRenderable... like maybe a Barista and a Plant only have Entity as a common ancestor

#

but they both are rendered as animated sprites with positions

#

that may change over time or whatever

#

so your rendering system would want to be able to query for all the ICharacterRenderables and do that all in one pass or whatever

frail parrot
undone coral
#

heuristic game AIs are like, a whole different beast, especially for simulators. usually all of that code is authored separately. if you have some kind of really abstract approach, I can see how you could do

interface IAI {
 float activeValue { get; }
 GameAction[] availableActions { get; }
}

with many things returning no value or an empty array of actions

#

this is really hard to do

undone coral
scenic forge
undone coral
#

that's still fine

#

that's how you have it right now i think

#

that's a Good Idea

#

it's when you have something that has to look like a tile

#

like a coffee spill

#

but is actually something else - it's not a Tile

#

do you see what i mean?

native nebula
#

tiles and entities

scenic forge
#

Honestly if you are hitting performance bottleneck, I would really suggest look into Burst, unless what you are doing is really not Burst friendly.

fair wolf
#

hi, i'm having a problem with AI. When i record a video of the unity3d screen with a screen recorder software the AI choose almost always the same strategy. what could cause this? a fixed seed?

untold moth
#

What AI are we talking about here? Artificial General Intelligence?😅

fair wolf
#

i think the problem is not related to the AI algorithm

#

because i get that issue when i start to capture the screen

untold moth
#

There's not much we can help you with without knowing the details.

fair wolf
#

it's a markov based AI

#

with a custom trained model

untold moth
fair wolf
#

ok

echo timber
#

Is it possible to combine skinned mesh renderers?

#

I would like to have my weapons "baked" as one draw call after they're customized with attachments.

#

Is that sort of thing possible?

sly grove
#

good luck with your materials though... 🤔

echo timber
echo timber
sly grove
#

SkinnedMeshRenderer is a renderer which renders meshes

sly grove
#

the accessory meshes would need to be vertex weighted too - you might have to do that in code, though It's probably possible to do it ahead of time, not sure.

#

Honestly I wouldn't bother with this though it's probably not worth it

echo timber
#

Oh! Ive already weight painted them.

echo timber
sly grove
#

how many of these weapons are actually being rendered at once?

#

It's probably not a huge performance gain

echo timber
#

Did I mention it's on mobile? So even a few draw calls make a huge difference!

#

So I have 10 players with their weapons

#

Without mesh combining its too many draw calls, with all the weapons and their attachments

untold moth
#

How do you expect these objects to move and rotate(or even just exist) individually if you combine their meshes?

echo timber
#

What do you mean? They will be separate objects for each player? For example rifle A and its attachments will be "baked" to b one object, rifle B will get the same procedure. But rifle A & B are still separate objects.

#

So player A has rifle A, player B has rifle B, and so on

#

Does that make sense?

untold moth
#

So you want to combine a regular mesh with a skinned one..?

echo timber
#

No, my guns with all their attachments are skinned. All of the attachaments are rigged with their gun. So I want to combine skinned with skinned.

untold moth
#

The gun rig with the character rig?

echo timber
#

No. The gun armature has a gun object, and all of its attachments as separate objects. I want to combine meshes under the gun armature.

#

For example:

Rifle armature A
-gun mesh
-suppressor
-scope1
-scope2

untold moth
#

That might be possible, but would probably require some intricate coding.

echo timber
#

I see.

untold moth
#

Google seems to provide some results for combining skinned meshes.

echo timber
#

Yeaaa I saw some paid assets, I wanna avoid those.

untold moth
#

That being said, I'd agree with Praetor about wether it's even worth it.

#

If it's something static like a gun, I'd probably use regular mesh renderers for everything and let dynamic batching combine stuff.

echo timber
#

I think it'll take some time, but the performance gain will be worth it. Because I already have so many dynamic objects in my scenes

untold moth
#

If I were you, I'd check the profiler. Honestly, skinned meshes are way heavier than a few render calls.