[MEGA THREAD] Get Your Code Reviewed or Review Others

[dense leaf]

Well it's actually 384 sprints because-
fuck off annoying PMs 🙃

[junior holly]

You guys would love my father

He's been a project impl manager for like 28 years now

Basically an architect minus being a dev

[dense leaf]

My father works with VB at work

[junior holly]

What is vb

[dense leaf]

Virtual Basic

[late smelt]

Visual Basic

[dense leaf]

same thing

[junior holly]

No clue what that is and I'm already puking

[late smelt]

It’s a programming language

Pretty old

[dense leaf]

pretty shit

[junior holly]

Damn it only made it to 6.0

https://paste.md-5.net/qicuroroqa.java

Do you guys hate this

[late smelt]

Visual Basic .NET only made it to 6.0

Visual Basic is still alive, and is on version 17

[junior holly]

Oh

[woeful pasture]

depends on the internals

oh you posted internals xD

[woeful pasture]

No I don't like this personally, Is their really a reason to put skills behind an object like that? both of the values you passed can be thrown in a function to cause change.

[junior holly]

mmmm

[woeful pasture]

this could much better be done with some sort of registry and more functional approach

[junior holly]

Funny you say that cuz last time me and con talked, he said I was doing "too much functional" stuff kek

[junior holly]

Tbf no, just trying really hard to follow stricter oop 🤷

[woeful pasture]

I think too much of both is bad, but here it makes sense to just not use objects. The thing about mathmatics is that they can inherently be best be described by functions

if skills apply changes to how the disc flies, why not apply those transformations as functions? It only makes sense

[late smelt]

Did someone say registry

😩

[junior holly]

https://paste.md-5.net/yikixibuyu.cs

This is where the skills actually manipulate the physics, I feel like you're probably right tho it feels quite bloated

[woeful pasture]

yeah I feel like not applying the skills as functions makes no sense here 👀 especially since they can be modeled by pretty simple functions

[dense leaf]

Registries my beloved

[junior holly]

I think my biggest thought was keeping serialization separated but then again why put each skill into it's own file, could just toss them all into one call it a day

[late smelt]

I need to make a registry system like mojangs

[junior holly]

I also liked the thought of individual objects in terms of displaying players skill sets, makes it easier in my opinion to deal with an existing object rather than have to create one from thin air

[woeful pasture]

part of me thinks that you should just have a PlayerData class with a Map<Skill, Integer>. You could make more things customizable this way too e.g. the particels behind the disc of each player etc

just my thoughts, I prefer using functions in this case over objects so just my opinion but ye

[junior holly]

I mean I could the same with internal functions within the skill classes

Idk I like the deep encapsulation too

[alpine anvil]

Use pineapples

[woeful pasture]

Mojangs is too overly verbose

[junior holly]

I feel like skills aren't something that should ever be messed with outside of their current and max levels

[woeful pasture]

I never saw any utilitiy of it over something like mine which is just a thin map wrapper with utilities prety much

[junior holly]

And perhaps creating your own skills too... not sure on that one yet

[late smelt]

Yeah but is yours statically accessible

[woeful pasture]

uhm probably? idk what you mean by that

[late smelt]

All of mojangs registries are static

Stuff can also be registered to them in static fields

[woeful pasture]

I always have a Registries class so I can do Registries.THING

[dense leaf]

Brawls BasicRegistry my beloved

[junior holly]

https://paste.md-5.net/ozawerabet.cs
Back again

[wintry fern]

The other issue with these is when using them on stuff you cant guarantee wont be null later and then it throws trap errors

Trap if you dont know is a type of error where you stated in your code that it wont be null or is not null and it ends up being null later lol.

[cobalt trellis]

Idk, I had relatively few issues with that method. On the other hand I understand what the implications of it are

[hasty lotus]

i honestly use it to suppress ide nullability warnings if i know the object is guaranteed to be not null if that makes sense (sometimes IDE's aren't smart enough to tell)

if it is null then its useful cause it tells me before breaking more stuff

and id just import and use the static method so it would be like

final Player player = requireNonNull(foo.bar());

[late smelt]

Just use nullability annotations smh

[random yacht]

Or assert if you wanna be really fun

[hasty lotus]

i usually use checker so i dont have to add in the annotations

cuz it adds them in for me at compile

[random yacht]

I do actually use asserts for some calls, like to ItemStack#getItemMeta() if I just freshly made the ItemStack. It's never going to be null

And assertions are disabled by default so whatever. I'm only doing it to appease my IDE

[hasty lotus]

i do the same but use requireNonNull, but requireNonNull is basically a useless null check at this point

[random yacht]

Objects#requireNonNull, imo, is for streams

[hasty lotus]

People who use Preconditions

https://tenor.com/view/emotional-damage-emotional-damage-meme-funny-gif-24332819

[round fossil]

implicit null checks do be lovely at times, altho it can get rly anti pattern-like

[late smelt]

Muh cpu cycles

[sharp epoch]

as if anyone coding in Java cared about that

nowadays most of algorithms are more memory bound than cpu bound anyway

[night knoll]

xd

ajajaj

[late smelt]

Muh bytes

[round fossil]

shame we always have to sacrifice something

zero cost abstractions when

[late smelt]

If your code can’t run on an Atari 2600

It’s bad

[junior holly]

https://paste.md-5.net/tijejuyoza.java

Notes, questions, concerns, love you guys

Changed this from player.getUniqueId to the id I pass... I was being dumb lol

[round fossil]

its a bit weird

but ehm well for starters

why not depend on Plugin instead of JavaPlugin?

[junior holly]

I assume other people might want to use my round objects

[round fossil]

and if u use ConcurrentHashMap, your type should be ConcurrentMap right?

since the concurrency becomes a requirement rather

[junior holly]

Fair

[round fossil]

no i mean

Plugin is the interface of JavaPlugin

so why JavaPlugin over Plugin?

[junior holly]

Oh well then because I wasn't aware of that lol

[round fossil]

na all good, just curious

[junior holly]

What about the handlethrow method, do you feel it's redundant to have multiple handlethrow methods per object?

ie: One in the round as well as the disc object

[round fossil]

oh

eh it depends

[junior holly]

https://paste.md-5.net/fuwepufeki.java
context:

[round fossil]

i mean it becomes ambiguous

like which is the right type to invoke the method on, and which type is just the intermediate one

alternatively name them better

[junior holly]

Maybe the round method should be called handleStroke

I feel that'd be a bit better

[round fossil]

from a library perspective ofc ^^

yea i mean maybe :>

[woeful pasture]

shouldn't the handleStroke call be under the Hospital's ICU object

[junior holly]

Haven't even implemented teh ambulances yet kek

[woeful pasture]

considering you're from America you might be best off implementing an Uber class instead

[junior holly]

Nah health insurance system is already functional

Just don't be dumb and not have insurance

[woeful pasture]

as far as this method goes it isn't horrible tbh, it def saves repretition so I think its fine

[junior holly]

Well I was thinking, from the context of the round I should invoke a player object just for those odd cases a player might log out, since the disc object's handleThrow method actually uses the player object

[round fossil]

also kat

RoundHandler

[junior holly]

I just mean handle the nullability case for the player in the round method rather than the disc method

[junior holly]

tell me

[round fossil]

why is only the map thread safe

and the sets are not

[junior holly]

Because I am dumb

I forget that's a thing psh

[round fossil]

fair

well u could also ask urself if u perhaps are oversynchronizing rn

cuz just smacking thread safe collections on everything tend to lead to oversynchronization

[junior holly]

I just thought most times people like to handle their "rounds"/minigames async so

Thread safety is a concern most times in the data migration innit?

[round fossil]

that's a nice thought, but in this code u dont deal with async whatsoever

[junior holly]

yeah true

[round fossil]

well read writes happening at the same time

and writes that depend on reads

ofc oversynchronizing is better than undersynchronizing, but as said

[junior holly]

erm concurrency is gonna take me forever to fully understand

[round fossil]

u should primarily focus on writing a structure that follows what you need, you can let your "library consumers" derive your types (<-- open closed principle) and pass their own dependencies to concrete types to allow of modification

i think there's a principle for this

its called open-closed principle if im not mistaken

[junior holly]

I will have a read on it for sure

[round fossil]

yea its weird stuff

[alpine anvil]

its super fun

[junior holly]

Did you mean like this btw?

[round fossil]

yea

if concurrency is a requirement

then def add that

[junior holly]

mmm

[round fossil]

and this principle is called liskovs substitution principle i believe xD

well i havent done high level software for some time

but ya

[junior holly]

Thinking about it more the only "hardcore" math that occurs is the disc physic modifications, still this has no relation to concurrency needs of the round, any other things that have to deal with this data is simple math and other functions

[round fossil]

eh depends

it might

but prob not

[junior holly]

https://paste.md-5.net/olebedirer.cpp

(it's poorly written cuz I was just trying to figure a flight phase system kek) This is really the only math that could be considered heavy

[round fossil]

it doesnt even look that bad

or at least u made it out to look nice

[junior holly]

Well I do certainly appreciate that

backhand and forehand physics are quite similar

[round fossil]

Well I do think u should avoid Math.random()

[junior holly]

basically just swap the +/- gives it the flight path irl

[round fossil]

but apart from that

nice

[junior holly]

this is def subject to a rewrite cuz I need to randomly invoke lift factors

either +/- on those too

[round fossil]

yea anyway kat

for this kind of code, it can be fine to leave some comments on it

explaining what the math is doing yk

[junior holly]

yeah tbf there is a single comment in that whole thing cuz I was trying to explain to a friend (not a coder)

There were more but they are dumbed down lol so I removed them

Also a lot of the math needs to be converted to configurable fields instead of magic numbers

[round fossil]

yea

[junior holly]

Did you ever see the irl reference compared to how it looks now?

[round fossil]

nope

[woeful pasture]

unless I missed something newer here's the comparisson
#general message
#general message

[pine grail]

A)

public void handlePlayer(Player player) {
  if(player.getHealthPercentage() < 0.50) {
    player.setHealthPercentage(0.50); // Sets the health to be at least 50% of the maximum health, convenience method
    player.sendMessage(Text.literal("Healed to 50%").color(Color.GREEN));
    return;
  }

  player.sendMessage(Text.literal("You have more than 50% health!").color(Color.GOLD));
}


public void otherMethod(Entity entity) {
  this.stopTicking.add(entity.getUniqueId());
  this.spin.add(entity.getUniqueId());
}

public void someEventHandler(EntityTickEvent event) {
  if(this.stopTicking.contains(event.entityId())) {
    event.cancel();
  }

  if(this.spin.contains(event.entityId())) {
    var storage = event.entity().storage();
    var plugin = event.context().plugin();
    var id = Identifier.plugin(plugin, "ticks");
    var ticks = storage.getInt(id, 0) + 1;

    if(ticks >= 3) {
      spinEntity(event.entity());
      ticks = 0;
    }

    storage.setInt(id, ticks);
  }
}

vs

public void handlePlayer(Entity player) {
  var health = player.get(Health.class);
  if(health == null) return;

  var messages = player.get(Messages.class);
  if(messages == null) return;

  if(health.getPercentage() < 0.50) {
    health.setPercentage(0.50); // Sets the health to be at least 50% of the maximum health, convenience method
    messages.send(Text.literal("Healed to 50%").color(Color.GREEN));
    return;
  }

  messages.send(Text.literal("You have more than 50% health!").color(Color.GOLD));
}


public void otherMethod(Entity entity) {
  entity.remove(Ticking.class);
  entity.add(new Spinner(3)); // Every 3 ticks
}

public class SpinSystem extends System {
  // Implement this yada yada yada
}

I guess there is also ensureGet, basically Entity#ensureGet throws an exception (with a custom message if provided) if the entity doesn't have that component

Please if you can state the reasoning (cc: @dense leaf)

[night knoll]

ECS way better

[pine grail]

I suppose I did not ask for a detailed reasoning

[night knoll]

Is ECS code real thing in spigot or that was pseudo code?

[pine grail]

wym, the bottom and top one will both technically be my API, I am asking how to model it

it's not Spigot-specific, it works as a sort of translator to multiple platforms, I suppose

[pine grail]

Ah, pseudo code I suppose, not in spigot

[night knoll]

Yup

[dense leaf]

I don't really find it intuitive to begin with, especially the messages, having an ECS component for messages seems really annoying to me since I'd just like to do a simple sendMessage. Then I also don't really understand what the actual code does in the ECS example, e.g., what is otherMethod used for??

[pine grail]

they're all different methods to showcase how you'd do different things, not exactly related

[dense leaf]

Also it'd be nice if I could just access basic player data like health and saturation in the entity object

[pine grail]

The Entity itself is not a Player, nor a zombie nor a wither, it's just Any entity

An entity could very well just be a block

[dense leaf]

...What

Why would an entity be a block

[pine grail]

A block has components:

Touched
Landed
etc

or I suppose not

[dense leaf]

A block is certainly not an entity tho

[pine grail]

It's an entity in the sense of a presence in the world, not a mc entity, but I can limit it to be only entity entities for now

Also

regarding what you said

[dense leaf]

So I'd just rename it to ECSEntity or something then since that can get confusing quickly

[pine grail]

I mean why would I do that? You don't depend on any other libraries

a block would also be part of the ecs

[dense leaf]

I mean do you not have an actual entity entity class then

[pine grail]

Entity entity = ...;

entity.sendMessage(Text.empty());

entity.sendMessage(new FireDamageMessage(5));

[pine grail]

No, there is no "Entity entity class"

each entity is defined by it's components

example

[dense leaf]

Entity entity as in actual minecraft entity

[night knoll]

Bro didn't used game engines before 💀

Entity is literally any game object

[pine grail]

var zombie = ...;
zombie.add(new Position());
zombie.add(new Velocity());
zombie.add(new NetworkIdentifier(Identifier.minecraft("zombie")));
zombie.add(new Health());
zombie.add(new Pathfinding()); // or something

var slimeBlock = ...;
slimeBlock.add(new Position());
slimeBlock.add(new NetworkIdentifier(Identifier.minecraft("slime_block")));
slimeBlock.add(new Bounce());

// you can technically now add the block in the world

Obviously this process is not done by hand

[night knoll]

Idk looks like library for minestom

[pine grail]

Yeah the goal is providing the same flexibility on the vanilla server (ig Spigot / Forge aren't really vanilla but ykwim)

[pine grail]

@dense leaf

[dense leaf]

I saw it yes

[pine grail]

What do you have to say about it

[dense leaf]

idk I'm not really a fan of ECS's

[pine grail]

Why so

[dense leaf]

because muh classes

I just wanna zombie.setHealth(20) or something

[pine grail]

zombie.get(Health.class).set(20);

plus you can also remove health after that

make the zombie immortal

[night knoll]

Make your own health component

[pine grail]

^

You can add whatever functionality

[night knoll]

Kotlin extension functions 🗿

[pine grail]

oh fair yeah rad could just use extension functions 💀

[junior holly]

idk if this was pseudo but funny a builder using get and set in the same line

[dense leaf]

Well because I should stop using Kotlin I shouldn't

[pine grail]

oh yeah fair forgot about that

[junior holly]

Also why would the builder have a get function

[pine grail]

it's not really a builder:

Zombie Entity
get(Health.class) Health or null
set(20); void

[junior holly]

oh kek

[random yacht]

Looks like some shit Sponge would conjure up

I still think Java's a shit language to do component-based paradigms. Languages with templating and reified generics are way better to do it with.

[woeful pasture]

its actually so great cuz then you don't need inheritance! I'm sure it is not confusing design at all!

[dense leaf]

Kotlin has reified generics!

[random yacht]

Kotlin can eat a dick >:(

[woeful pasture]

what is reified generics exactly?

[random yacht]

Compile-time type information from generics, basically

Java has type erasure

[woeful pasture]

ahhh

[alpine anvil]

conclure?!

[dense leaf]

:3?

[sharp epoch]

you mean dynamic dispatch nuclear bomb

[night knoll]

Added pagination
https://github.com/NazarbekAld/XeXeGui/blob/main/tests/src/main/kotlin/PaginationTestCommand.kt

-# Heart attack warning

[dense leaf]

[pine grail]

I mean not really

think about it

You can have both!

(/hj)

[dense leaf]

DFU would benefit from templates A LOT

[desert ore]

I feel like you could just have an Entity class like this which can represent anything, like, if(entity.get(Type.class) != null && entity.get(Type.class).is(Block.class)) {...} or something along those lines

[pine grail]

Ah yeah, so a Type component

which says Entity, Block, etc

[pine grail]

skill issue get good

good call @desert ore btw

(also if anyone can further vote I would like a more definite answer)

[desert ore]

I'd be interested to see a system that is completely and fully component based, maybe even

var server = context.getServer();
if(server.get(Type.class) == null || !server.get(Type.class).is(Server.class) || server.get(Logger.class) == null) return;
server.get(Logger.class).info("hello world");
``` but that seems annoying

[pine grail]

it is very much annoying

the system is component based for the following:

Entities, Blocks, Items

Then

Lifecycles are fully plugin-driven except LOAD, ENABLE, DISABLE, UNLOAD

Events follow in suit

except vanilla events

which are also built in

[pine grail]

you can actually server.ensureGet(Type.class).as(Server.class).get(Logger.class) 💀

I mean that is WILD and I won't allow it

mainly because of structural reasons

[round fossil]

Only annoyance is that it becomes extremely runtime based, like you don’t really know what types hold true, esp when you start adding, renaming and removing componential keys under a longer span

[pine grail]

I took some liberties to make sure no code that runs on this system is capable of affecting anything outside of its own VERY limited area, not even other plugins, it all communicates through events

[pine grail]

Yeah ^

Anyways, do I proceed with the component-based approach?

[desert ore]

In some way, everything being runtime based is even pretty good

[round fossil]

I mean both have different pros and cons

[pine grail]

Yeah, I like the component system for it's flexibility and adaptability

[desert ore]

I personally would love to see it

[pine grail]

You know what, this is my framework

fuck it

[night knoll]

Fuck around and find out

[pine grail]

Yeah

And plus it will allow me to make some very funny things

The trouble...

[round fossil]

but yea, isn’t sponge heavily opinionated this way?

[pine grail]

It will work on Fabric easily

Spigot on the other hand...

ugh kill me

[pine grail]

I mean sponge is... interesting

I definitely want something cleaner than THAT

[desert ore]

I like the concept of sponge, but I'm not sure whether I like the implementation of it that much

[round fossil]

they have different implementations right?

I mean it may make more sense to go with a componential design then

[desert ore]

I believe yeah, but today we were talking about sponge and somebody brought up the itemstack builder

That shit is hell

[round fossil]

https://tenor.com/view/its-fine-dog-fine-dog-fire-meme-gif-16250404

[pine grail]

how should I add duplicate component add

return false? return the old component? throw an error? add both?

what if I add an "allowsDuplicate" flag on the component

[sly jackal]

what is duplicate component add

[desert ore]

Make it a seperate type of component

Maybe a seperate component to store a list of a certain component

[sly jackal]

i dodnt follow the convo, what kind of component is this

[desert ore]

That way you don't face issues with get returning an optional list which is ass

[desert ore]

Maybe those screenshots will help you understand a bit more

[sly jackal]

how are those components even stored

[desert ore]

Probably a list or a hashmap

Map<Class<Component>, Component> seems reasonable

[sly jackal]

well if its a map, dont allow duplicates

easy as pie

[desert ore]

Ye, but I think a ComponentListComponent would be interesting

[sly jackal]

well yes you would have to

otherwise how can you even keep a list of things

[desert ore]

True

Ah I love the idea of components

[sly jackal]

pretty much sounds like you are trying to do nbt with extra steps

[desert ore]

It seems so unnecessarily overengineered and complicated

And I love it

I absolutely fucking love it

[sly jackal]

what are the components even gonna be used for

[desert ore]

everything

Entities, blocks and everything of that type ig

[sly jackal]

oh, well too bad everything is already a component

by the name of Object

[desert ore]

Lol

[dense leaf]

inline fun <reified T : Component> Any.get(): T?

frfr

[pine grail]

not really! It has huge practical usages on a large scale

It is used for example by Unity DOTS to allow for dozens of thousands of entities ticking at once basically

[desert ore]

Damn

Can you apply a ticking component to a non-ticking entity?

[pine grail]

Yup, and vice versa remove ticking from a ticked entity

Also ECS is pretty much the only way you'll get more than a few thousand entities at once without heavily multi threading or pulling other tricks

It can handle thousands upon thousands of entities (even in the 100k+ range) *if implemented correctly

Right now I won't implement it directly in the game, more or less slowly replace vanilla with it

[hasty lotus]

Roast me 🔥 https://github.com/PulseBeat02/MurderRun

More specifically, this https://github.com/PulseBeat02/MurderRun/blob/main/src/main/java/io/github/pulsebeat02/murderrun/game/gadget/GlobalGadgetRegistry.java. I am loading and instantiating all my Gadget classes which I store in an Enum. I first load them into a registry, where their String identifier and ItemStack item in inventory can be just first used for display purposes. But each Gadget has its own function and utility when the game starts. When the game starts, I create a new instance of all the Gadget to make sure it doesn't conflict with any other games that may be running at the same time.

But does anyone know a way to "lazily" load the Gadget I only need at runtime. The thing is, one of my Gadget is a RandomGadget, which is a gadget that gives you a random gadget, and that means I'd have to load the Gadget class when it's activated. Anyone know a good way to lazily load the Gadget classes so I don't have to load hundreds of them each time the game starts to save resources?

[dense leaf]

nice code

[hasty lotus]

Bleuh

[late smelt]

How does an ECS allow for thousands of ticking entities at once

[woeful pasture]

It doesn't that's the great part

[round fossil]

Doesn’t that still compile down to a java bytecode misery eventually?

and java has reified generics, just extremely awkward and uncomfortable to use

[dense leaf]

I mean it just inlines the function body

[jagged rock]

isn't it the T... hack

where you can get the Class<T> back from the empty array

[desert ore]

I wonder how different things would be if java didn't have type erasure

[sharp epoch]

type erasure is a good thing, people just go bonkers over the type system otherwise

[sharp epoch]

kotlin embeds its own metadata into classes to support this kind of functionality, this is part of why it is hard to relocate some stuff with koltin since some of that metadata might contain class references which one also would have to relocate

[pine grail]

well basically, take minecraft: Each entity is processed for something every. single. tick. Even if it can't actually do anything, in an ECS instead each tick there is basically a bunch of queries that are ran (can happen in async), so things that run in tick intervals just get ignored until they should run, and don't count towards any loops (as they are separated by archetypes, basically a set of components)

TL;DR: Just separating entities into buckets and allowing for parallel processing of entities (i.e. all villagers will tick together WHILE all of the zombies are ticking together)

entity <-> entity interactions are scheduled at the end to ensure synchronization

iirc

[jagged rock]

That's if the ECS you're running is concurrent yeah

[pine grail]

I mean yeah I'm doing that

sort of

it will be concurrent

notrn

Right now I'll just do a translation layer (so performance--) but later I'll slowly replace vanilla stuff with it

[desert ore]

You are planning to make it non-vanilla and incompatible with everything?

[pine grail]

I mean, not really I'll just replace ticking

oh wait

Hm

[dense leaf]

lol

[pine grail]

If you're using something that fundementally changes how plugins/mods are made

Should you really be expecting that everything else works with it

custom entities will work

mods that change vanilla ones won't

[round fossil]

Ah that makes sense

[round fossil]

I mean just T extends S on a class type parameter (though u may still lose higher ordered types), and then also you got the type token hack with TypeToken<T>{} (subclassing and passing ur type to the superclass type parameter)

But yea varargs can be polluted w non reifiable params

[desert ore]

So you want a heavily fucked system?

[pine grail]

not really, but if I want to allow changing entities It'll be incompatible with mods that also change the same entity in the same part

[pine grail]

well then it is ECS

[dense leaf]

?

[sharp epoch]

blud just wanted to talk about ECS

[woeful pasture]

Guys what's an ECS

[dense leaf]

extra complicated system or something idk i don't do gamedev

[woeful pasture]

Entity component system apparently

[dense leaf]

yeeah

[woeful pasture]

I wonder if mojang will ever embrace entity components like they do items

[sharp epoch]

it is a pretty standard way to do entities in gamedev, and mojang does seem to be heading in that direction

it wouldn't be exactly in the way Ike demonstrated though, at least I don't think from looking at the current source changes

[pine grail]

I would love the system I proposed lol

[willow bone]

That is how I did it

you can essentially add any <T> as a component for the entity

and also get it back by giving an entity id

it saves 'em in a sparseset

[pine grail]

Tbh that's a neat idea but not for me

[willow bone]

What you planning on doing then?

oh wait let me show you my query system too

it's clean as fuck

[pine grail]

Entities aren't builders, they are constantly modifiable, replaceable, they can cross universes and are not bound to one system, each entity is given n archetypes depending on how many archetypes it matches, it may match two identical archetypes, each archetype has a priority and a system attached to it

[willow bone]

yes ofcoruse

[pine grail]

This way you can have THE maximum flexibility

[willow bone]

the first image I showed is how to create an entity

let me show you how to query and modify an entity for example

[pine grail]

And you rarely interact with the components

[willow bone]

me or like in general

[pine grail]

Me*

Basically: send a text message to an entity, if it can process it, it will! Otherwise it won't. But you don't know.

Send a damage message

Did it get damaged? Probably

What's the health? Does it even have health?

Zero coupling, even between the things acting on the entity and the entity

The entity also doesn't know which messages it can receive

Each component says what it can receive

[willow bone]

what is an 'entity' in your case?

in my case, it's just a usize (u64 / long?)

@pine grail Example of query

the "query" gets all entities that have the player identity + position component

this can be any amount of components and any type

since I need this all to be multithreaded, there cannot be 2 multiple mutable borrows, but infinite immutable ones

this supports 100% multithreading

[pine grail]

A class to give convenience methods, otherwise it's always universe. etc

Otherwise it's just a number to a list of components

[willow bone]

do you save entity -> Component[] or Component -> entity[]?

arrays of structs or structs of arrays

[willow bone]

thats a good idea i should do thatt too. so you can do entity.get::<Position>(state)

[pine grail]

I do

ID -> <Type, Component> for lookup

Archetype -> Entity[] for system processing

ID -> Entity (object, stored to avoid re-creating)

Or actually

Yeah there that's it

[willow bone]

oh so each entity manages its own components?\

[pine grail]

Yup

Fast lookup

[willow bone]

I have HashMap<Type, SparseSet<usize, Component>>

because in systems, you don't care about individual entities

you care about the general population

so it'll be better for CPU cache this way?

[pine grail]

Yeah that's fair enough but then the issue arises of accessing speed, in my case each entity is given a unique archetype ID, this ID is then used to instantly know every entity in the universe without doing any further lookups

I.e. the accessing speed to get every player is O(1)

Same for every say zombie or zombie-like entity

Etc

[willow bone]

hmmm interesting

[pine grail]

You straight up ignore the cache because the arrays are also contiguous in memory, as each component is an ID into the global component table

Act no that was the old one

Current one doesn't have a global table anymore (faster)

But still yeah all of the data is thightly packed together

[willow bone]

thanks for the idea :)

[pine grail]

Haha np :P

[willow bone]

good luck with your idea. you're doing it for minecraft right?

[pine grail]

The state is actually in the entity for me, you do entity.get(Health.class)

[pine grail]

Yup

[willow bone]

huh? how?

wont that take way more memory?

[pine grail]

Nop

I mean "in"

That's not quite true

[willow bone]

even if you have a atomic shit, it still saves it

[pine grail]

It's in the universe but the entity knows what it is from an id

Basically an entity is just two IDs

[willow bone]

heh

[pine grail]

Considering that I could hold 100k entities in around 1MB of ram it's not too bad

Java ram overhead go brrrr

[willow bone]

dang how

mine takes WAY more

my rust skills aren't the best

[pine grail]

basically the issue is

each object has some overhead

around 32 bytes of ram iirc

[willow bone]

how 100b per entity

with or without components?

like an empty entity?

[pine grail]

it would be entities with position and maybe another component such as health

yeah no you could store that in 100b per entity

[willow bone]

position is around 24 and health around like + another 4, so 28 bytes

do you support multithreading?

[pine grail]

no actually, add 32 on top of each comp

[willow bone]

each component has 32?

so 60?

• 32 of an entity

so around 100b?

[pine grail]

actually no I'm dum

That is for arrays

assuming you're on a 64 bit system

entities are 48 bytes, each component is 16 + data, so in this case it would be 32 for position and 20 for health, so exactly 100 bytes

[willow bone]

interesting

mine would definitely take way more

[late smelt]

Smh integer health

[willow bone]

float better

[pine grail]

Fixed precision numbers

[willow bone]

isnt float a 4 byte

thing

[pine grail]

I mean yeah I guess

Fair enough sure you can just use a float, doesn't change the memory usage

Though there is a downside

there is inherently some more storage to take into account to hold the archetypes

And to hold the components

it ends up not being 100k entities, it would be 100k entity objects @ 100 bytes per, but then there is collection overhead

[willow bone]

12.3 MB for 1 million plain entities

tf

12 bytes per entity?

8 bytes for the entity itself

and 4 bytes for uh

[pine grail]

Dayum

[willow bone]

let me try with position

[pine grail]

mine would be around 46.8kb for 1M plain entities

[willow bone]

88 MB with position

[pine grail]

Plus collection overhead but idk how to calculate

I can actually lower this by using the entity object itself as the entity ID

because java already gives me a unique ID per object which is part of the 16 bytes

removing 32 bytes ish

around 15.6kb per 1M plain entities if I do that

[willow bone]

82 bytes per entity with a position (double*3) = (10 million entities took 808 MB ram)

[willow bone]

dang tf

do you support multithread?

[pine grail]

Not yet

It would be insanely hard to support multithreading in mc

but it does support regional collisions

and regional entity lookups

[willow bone]

wtf

regional entity lookups?? how?

[pine grail]

So basically entities are ran through the position system, this system basically sorts entities into an octree so that you can access an area extremely quickly and disregard the rest of the universe

[willow bone]

octtree takes a shit ton memory no??

last time i tried on a 100k or a million entities and it took almost 16 gb ram

[pine grail]

No WHAT?!

An octree is either nullptr or pointer to either another octree or an entity index start - entity index end numbers

each octree element is 16 bytes

[willow bone]

i just realized. a million bytes is about a megabyte. how do you get 15.6 kB?

[pine grail]

I multiplied size in bytes * 1M / 1kb

basically

16 * 1.000.000 / 1024

[willow bone]

that'd be 15625 (kb)?

or 15 megabytes about?

[pine grail]

hmm

lemme thonk

Ah yeah I'm DUMB

it's 15MB

or 68MB per 1M entities with pos and health

[willow bone]

160 for me

yikes

[pine grail]

though I can lower this

considering that mc only uses 8 bytes for position

I can do the same and actively edit the number

and then that would bring it down to

60MB

not bad?

also the issue here is using two diff components btw

I FOUND A SOLUTION

@willow bone

I turned 100M entities into just 36MB with pos and health

[late smelt]

How does mc use 8 bytes for position

[pine grail]

sacrifices the y axis

[late smelt]

Eh?

[pine grail]

https://wiki.vg/Protocol#Position

y is -2048 to 2047

[late smelt]

How does that work for stuff at decimal positions tho

[pine grail]

¯_(ツ)_/¯

actually

mc uses deltas to signify position

they send 6 bytes per delta

and uses 24 bytes per teleport

so actually

still 12 bytes

40MB

eh

[willow bone]

bro what

100m is 100 mb

[pine grail]

rn it is 40 bytes per entity with pos and health

wait NO I FORGOT ABOUT JAVA OVERHEAD

KMS

wait no that's not in the entity

poggers

[willow bone]

nice nice

[night knoll]

can you give me ur ide background

plz

[sharp epoch]

here: https://www.pixiv.net/en/artworks/59406498 (this is the original, you can probably find bigger sizes if you put it in google though)

[night knoll]

gracias

the fact that you literally update it is so crazy and says much about you

[willow bone]

i think its this one

[night knoll]

ty

[willow bone]

np enjoy

[night knoll]

ty

you 2

[dense leaf]

https://github.com/radstevee/readycheck made a silly little plugin for a friend (in java 😨)

[empty wigeon]

is this a fancy way of checking if some number of players are ready before some action can proceed?
what are we getting ready for? 🐈

[dense leaf]

yea pretty much that

[empty wigeon]

||(we love java ☕️)||

[dense leaf]

we love kotlin

[empty wigeon]

anyway, the first thing that stood out to me is ChatListener and how it's getting an instance of the plugin just for registration, which could be moved to the main class itself via some dedicated method that accepts registerEvents(Listener... listeners) or something like that ¯_(ツ)_/¯

[sly jackal]

i love a good plugin class singleton

but I do agree on passing the plugin instance in your chatlistener instead of getting from your main class yourself

which is just coupling that isnt needed

[dense leaf]

singletons my beloved

[dense leaf]

yeah that's fair I'm just not a big fan of that

[sly jackal]

i would just make it a variable passed in the comstructir

idk like maybe in the future you want this listener to check other stuff about a plugin like getting a resource or smth

also really nitpicky but in the ready command I would put the failing conditions at the top, not at the bottom

which in this case would make it an early return

[dense leaf]

yeah true

[dense leaf]

I just realised

I named the class RecollectPlayers instead of RecollectPlayersCommand

[woeful pasture]

Pass the instance or use a static field instead
https://github.com/radstevee/readycheck/blob/master/src/main/java/net/radstevee/readycheck/ChatListener.java#L12-L14

Might be worth creating Permission objects instead here so plugins like LP can pick them up automagically.
https://github.com/radstevee/readycheck/blob/master/src/main/java/net/radstevee/readycheck/Permissions.java#L10-L11

Otherwise lgtm

[dense leaf]

I register perms in my plugin.yml if that's enough for stuff like LP

[alpine anvil]

perms work without needing to be Permissions afaik

[alpine anvil]

call ur main class ReadyCheckPlugin, use real getters (tragedy no kotlin)

[dense leaf]

fluent getters are real getters

[woeful pasture]

Epic is in denial about fluent getters

I always use them for statics lol

[alpine anvil]

fluent getters are for kotlin and records only

and paper

[dense leaf]

they're great

[dense leaf]

wtflip

[round fossil]

The only real argument I’ve heard against fluent command and queries is that with get/set you can prompt your IDE to list all the “traditional” getters and setters (suggestion+/completion), and that it’s technically stricter by specification than fluent command and queries

[sharp epoch]

it is really not that big of a deal honestly

maybe because I'm used to look at the javadocs instead of depending on the auto-completion I guess, even if I get all the getters by doing get, often times I want to read what the get these do as well

[lyric stump]

https://github.com/InfLabss/PlayerKits2-Optimized/

[dense leaf]

this is the reason I hate bukkit commandexecutors

[desert ore]

I'd end up doing this with reflection lmao

[round fossil]

I mean it used to be more useful when people were strictly abiding get/set for data encapsulation, now that record components stray away from this, it has become less of a standard so get/set won't net you all getters, setters, data accessors and data mutators anyway.

[sharp epoch]

I was hoping you'd notice the pun in my message lol

[round fossil]

i prob missed it :')

so I assume you are pro-fluency then?

[lyric stump]

umm I think you guys missed this 😭

[round fossil]

Why did you suffix it optimized? Out of curiosity

[alpine anvil]

its super asnyc optimized deluxe async perforamce ++

[dense leaf]

multithreaded

[lyric stump]

It sounded better lmfao

It's a joke just in case you didn't get it

[sharp epoch]

I am, for the most part

but I do like consistency if anything, so mixing up fluent getters/setters with non-fluent ones is a no-go for me

there are cases where it can't be helped but if you can avoid getting inconsistent naming then better

[random yacht]

Do you think Brig is gonna be any better?

I assure you it won't be lol

At least with regards to this specific problem. You still have to specify command arguments and what they execute

If you don't want to be passing through your messagesConfig and messagesManager to these methods, make them fields or something. Or get them in the methods

[dense leaf]

Well obviously yeah but it's just worse with Bukkit

Brig is much better

[random yacht]

Oh okay Bukkit is just objectively bad for something Brig does the same Gotcha

[dense leaf]

I mean

[random yacht]

If you look at a vanilla command implementation, it's pretty similar, only it uses a builder pattern instead of if/elseif

[dense leaf]

At least you don't need to manually parse with Brig

[random yacht]

Yes but here specifically you're complaining about the if/else chain

So >:( Be specific when you complain

[dense leaf]

hehe

I just hate Bukkit CommandExecutors in general, had to use em recently

Except that I didn't need any argument parsing luckily

[lyric stump]

Lucky

[dense leaf]

I usually use Cloud for plugins, but Brig is also amazing

[hasty lotus]

I used to use Brig but I switched to Cloud

I use annotations because I really don’t like writing command argument parsing code or anything

I just want to write the damn function of the plugin, I don’t want to write command argument parsing logic!

[sharp epoch]

Commands.register(literal("idk")
                .then(literal("reload")
                    .executes(this::reload)
                )
                .then(literal("give")
                    .executes(this::give)
                )
                .then(literal("delete")
                    .executes(this::delete)
                )
                .then(literal("reset")
                    .executes(this::reset)
                )
                .then(literal("migrate")
                    .executes(this::migrate)
                )
                .then(literal("open")
                    .executes(this::open)
                ) 
 // help not needed anymore considering the client will suggest proper args ?)
        );

it looks horrible no matter how you refactor it, I'll give you that

[dense leaf]

In Kotlin you can remove the this and the semicolon

way better ikr

[random yacht]

I'm not sure that is better lol

But yes, Javier gave a good example. You're going to end up with what is essentially an if/else, only it does it with a builder

[dense leaf]

Command registration is always gonna be a PITA

[sharp epoch]

I believe that can easily be solved by making a DSL

[dense leaf]

Cloud!

Kotlin DSL!

[sharp epoch]

Cloud is eh, I always felt it is overkill for anything simple, just too much engineering going on there

[dense leaf]

That's true but that's also great for bigger and more complex commands

[sharp epoch]

don't know what their v2 is looking like right now though, has been a while since I looked at it

[junior holly]

Hate to break it to ya but that is kinda part of the plugin :p

[sharp epoch]

eh, frameworks like cloud exist for that very reason

[junior holly]

Why is everyone so keen on using the wheel instead of reinventing it smh

[sharp epoch]

their annotation-based stuff is essentially for those who just have a very declarative command tree and want to center on the actual command logic

||this is why I just use Skript for commands||

[hasty lotus]

yeah but i'm not spending my coding time handling argument parsing

the frameworks exist to minimize that

[round fossil]

the feeling when you write a command abstraction for the fifty-eleventh time

[rancid fern]

https://tenor.com/view/ah-shit-here-we-go-again-gta-sa-cj-gta-again-gif-20425950

[sharp epoch]

https://xkcd.com/927/

xkcd: Standards

Fortunately, the charging one has been solved now that we've all standardized on mini-USB. Or is it micro-USB? Shit.

https://xkcd.com/2140/ too

xkcd: Reinvent the Wheel

Right now it's a bicycle wheel, so we've had to move to lighter vehicles, but the reduced overhead is worth it. There was one week when a wheel of cheese got dangerously close to the first page, though.

[round fossil]

🥲

[night knoll]

don't

playerkits 2 is good

ajneb's is better

[lyric stump]

Umm, I need it for a server so I gotta add some more things to the API

Also, I'm learning so it's a win for me.

[night knoll]

mhm

[night knoll]

you guys wouldn't mind if I ask you to review my lib which uses both minestom and paper?

[hasty lotus]

Put it here

[night knoll]

I will put in a bit I saw some small issues need to patch it up quick

[pine grail]

maps:

[lunar vale]

why are you doing this in the kit manager

[lyric stump]

U don't think so I did that

I

[lunar vale]

https://cdn.discordapp.com/attachments/1186570213077041236/1261726249030782986/attachment-3.gif

[lyric stump]

Lmfao

Me no english

LOL

And yeh it's probably in the original player kits 2

[untold jay]

"milion" 😭

[late smelt]

Looks legit to me

[sleek shard]

I'd take it

lemme become a skript kiddie

I'm gonna hack em
tree

[pine grail]

color a && tree /S

wait no

yeah no that's correct

[sleek shard]

yeah I mean the 1.8 color codes are just default colors on windows cmd

[pine grail]

lol

[hybrid osprey]

Theyre the CGA colors

https://en.wikipedia.org/wiki/Color_Graphics_Adapter#Color_palette

[dense leaf]

Back in like 3rd grade I almost got expelled for doing tree C:\ in a while loop because they thought I was hacking

[pine grail]

lmao

that couldn't've happened to me because we don't have computers in 3rd grade kekw

[dense leaf]

that was like 10 years ago or something, don't remember what age I was in 3rd grade

[pine grail]

yeah so around 6yo for me then, which again, we only had computer lab next year?

and it was mainly just scratch

[dense leaf]

We had to go there for learning how to use a mouse and shit

[pine grail]

lmao

[sleek shard]

my teachers got mad at me when I was 10... because they had this website blocking software... I bypassed it by setting a proxy on the browser and gave it to my friends... I got detention D:

was fun tho

[woeful pasture]

Lol in my hs it seemed like they didn't care it was pretty well known I had remoted linux on my chromebook and never got talked to or anything

[sleek shard]

ahahah

[woeful pasture]

Like I had IT people walk behind me whole I was coding on Pycharm on an arch OS

[sleek shard]

in secondary/high school I also managed to bypass their security and got admin on one of the pc's... could do literally anything

shouldn't be saying that but anyway

[woeful pasture]

One of my friends got expelled for hacking his schools mainframe 💀

[sleek shard]

LOL

[woeful pasture]

He's got a news story about it iirc

It's funny to read

[sleek shard]

literally on the pc I managed to replace the shift lock with cmd (windows 7, back when I was around 13-14), opened up cmd on the logon menu, did net user, and set myself as an admin

that's the pc I got admin on

[woeful pasture]

Tbf you got security problems of the bored autistic kid hacks in on a whim

[sleek shard]

🤣

[woeful pasture]

You could run a lot on my school computers cuz they had python on em

Anyone with an ounce of knowledge could do a ton

Wasn't secured or anything

[sleek shard]

wtf

[woeful pasture]

I was able to change env vars and registry access from python

[woeful pasture]

I feel like schools always underestimate the kids who like tech

[sleek shard]

nah deadass

in my college (a week ago) I literally just made such a simple program into a class, methods with returns, params, and in/out stuff. Literally basic shit. And my programming teacher was impressed 🤣

c++

[woeful pasture]

Think my favorite thing I've made was some stupid project for one of my compsci entry level classes for java but I wrote the entire thing in Java Reflection

[sleek shard]

oh hell nahhh 😂

I love that

[woeful pasture]

I got 100% on it too even though one of the requirements was the code being concise. I will tell you when the class is 500 lines of reflection and the top is like

@Input("Enter One Number")
@While("number", Integer.class)
@Input("Enter A Second Number"_
...

If I were to take a class right now I'd probably write everything in bytecode using ASM lol

i'd just bootstrap my jar so they could run it

[sleek shard]

honestly one thing is - I've been doing java since I was around 12, but I do say 6 years because I like to keep it humble, as I didn't really start learning proper shit until I was 13-14. 20 now, so the route of programming is going quite well

[woeful pasture]

I didn't start java till 2021 and I kept taking month breaks cuz i was stupid so I kept quitting

[sleek shard]

I mean even now I can't even tell you every nook and cranny about java

only recently I've gotten good at it

within the past 2-3 years I've actually been trying to get better

[woeful pasture]

I've always been great, guess its my inate talent to make Inventory PRs

[sleek shard]

😂

[dense leaf]

I've only been using Kotlin for about a year now but can tell you about I'd say 80% of the languages features

[sleek shard]

I mean I think I know every java keyword

[dense leaf]

oh so you know java? name every class name

[sleek shard]

public - obvious
private - obvious
protected - makes it accessible to sub-classes for inheritance
blank - makes it accessible to package-private, allowing only the classes inside of the package allowed to access it
volatile - make the (primitive) data type visible with updates to all threads
... and some more

[night knoll]

InternalFrameInternalFrameTitlePaneInternalFrameTitlePane

[dense leaf]

ah yes that's every keyword

[sleek shard]

not every one but yknow

😂

I just don't have time to list them all out

and kinda cba

[round fossil]

I meaaan volatile prevents compiler optimizations also

[sleek shard]

oh that's a new one

thank you mr conclube

[round fossil]

:>

[late smelt]

What about strictfp!!!1

[hasty lotus]

native

[round fossil]

non-sealed :D

[hasty lotus]

sealed

[random yacht]

const!

[round fossil]

[junior holly]

https://paste.md-5.net/icadokeyez.java
What do we think

[night knoll]

https://github.com/NazarbekAld/XeXeGui/blob/main/tests/src/main/kotlin/AsyncGuiTestCommand.kt

[dense leaf]

Since you're already using paper, they have ItemStack#editMeta

I usually just do something like this so the scope is this and it returns the stack again: ```kt
inline fun ItemStack.editItemMeta(block: ItemMeta.() -> Unit) = apply { editMeta { block(it) } }

inline fun <reified T : ItemMeta> ItemStack.editItemMetaSpecific(crossinline block: T.() -> Unit) = apply { editMeta(T::class.java) { block(it) } }

[spring reef]

visually i would choose other particles

i would want to see that disc flying, rather than lots of particles blocking the view

codewise i would trim down the methods, reduce boilerplate and put the complex equations into a descriptive method

driver and disc are a bit misleading. i would not know what that should be without looking at the code

personally i would turn ThrowTechnique into an interface, and then make an abstract BaseTechnique or AbstractTechnique which then is the base for further techniques

[dense leaf]

There's gonna be a model and shit for that in the future

[spring reef]

make them available for combat as well, could be a cool technique

[junior holly]

Eh?

Elaborate

This is a disc golf mini game system so I don’t understand what the heck id do with combat

[random yacht]

I'm also not sure any of those suggestions are really beneficial lol. You could maybe compact down your switch case with a more generic method to reduce duplication but that's really about it

The terminology of those classes is fine. The "Driver" is a type of disc. Akin to how there's a "driver" type of golf club

I also don't think the throw technique needs to be overcomplicated into an interface. It's just not necessary. Abstract class is fine

[junior holly]

I assumed they had no idea what the point of this whole system was for so yeah kek

[random yacht]

If you want to get super critical you could maybe argue you could rename Driver to either DiscDriver or DriverDisc, whatever. Some people (myself included) like to link types of objects together by name

That's preference though tbf

private final String technique;

public ThrowTechnique(String technique) {
    this.technique = technique;
}

public String getTechnique() {
    return this.technique;
}

Also not entirely sure why you're passing a "technique" string to a throw technique. Is this better named as "name"? Or "description"? Or is it something else?

[junior holly]

Registry key

[random yacht]

Yeah then maybe "id" or "key" would be better imo

[junior holly]

Sure sure

[random yacht]

Feels weird passing something called "technique" to a technique object :p

[junior holly]

Yeah no for sure lol

[random yacht]

Other than that, code looks fine. Like I said, you could maybe compact down your switch so that each block just calls a more generic method, but I wouldn't even stress too much about that

[junior holly]

Mmm maybe, there’s a lot of techniques so that switch is gonna be duplicated a lot

Though the physic manipulations will be different

[random yacht]

Yeah, was writing out a generic method, kinda sucks ngl lol

It's fine

[junior holly]

kek

Visually speaking, what do we think?

(I pinged you in general with the irl reference)

[woeful pasture]

Looks good

[junior holly]

Why thank you

Past what like 2 weeks or something, been doing so much god damn math

Really happy with how this looks currently

[random yacht]

Visually looks good :)

Sorry, was chatting with family

[bronze crag]

Please, opinion on the idea if it's worth or stupid (the impl is not finished ofc, just wondering if making this makes any sense)

public class CacheWrapper<T> {

    private static final ScheduledExecutorService SCHEDULED_EXECUTOR_SERVICE = Executors.newSingleThreadScheduledExecutor();

    // Some weak key based cache which is going to remove the entry after 5 minutes if the object hasn't been used for all that time
    private static final Map<Object, CacheWrapper<?>> cachedObjectsMap = new HashMap<>();

    private final WeakReference<T> storedObject;

    public CacheWrapper(T object, Consumer<T> cacheAction) {
        this.storedObject = new WeakReference<>(object);

        cacheAction.accept(object);
        SCHEDULED_EXECUTOR_SERVICE.scheduleAtFixedRate(() -> cacheAction.accept(object), 0, 1, TimeUnit.SECONDS);
    }

    public T getStoredObject() {
        return storedObject.get();
    }

    @SuppressWarnings("unchecked")
    public static <T> CacheWrapper<T> of(T object, Consumer<T> cacheAction) {
        return (CacheWrapper<T>) cachedObjectsMap.computeIfAbsent(object, wrapper -> new CacheWrapper<>(object, cacheAction));
    }
}```

Methods which return CompletableFuture aren't not easy to work with, especially when you have too many of them nested
```java
networkInstance.getNetworkPlayerMap().sizeAsync().thenAccept(size -> 
  networkInstance.getNetworkMemoryUsage().thenAccept(memoryUsed -> { 
    sout(size + " " + memoryUsed)
  })
);

Turns into

val playersOnline = CacheWrapper.of(new AtomicInteger(0), value -> value.set(networkInstance.getNetworkPlayerMap().size()));
val memoryUsed = CacheWrapper.of(new AtomicInteger(0), value -> value.set(networkInstance.getMemoryUsage().join()));

sout(playersOnline + " " + memoryUsed)

[bronze crag]

Essentially, this is about objects which can be cached. I was thinking about Hypixel at the moment of writing this.

For example, they have a scoreboard which displays amount of online players on the whole server, let's say that the values on the scoreboard are updated every second and there are 50k players on the server seeing this scoreboard. This would mean, that 50k queries are being processed to the network every second, that's a bit of overkill (I was imagining they use redis and maybe keep players in some RMap).

So, instead of 50k queries/sec to redis, with CacheWrapper we do 1 query/sec instead. I wanted to think of something reusable for other cases and not just one that specific problem, otherwise it could have been done differently

[pine grail]

but why

[remote kiln]

if there is anything heavy what is called frequently, then it can be used with CacheWrapper to decrease load to the servers. Pretty much the example with Hypixel which he's provided

I believe that's what he meant

[bronze crag]

CacheWrapper is to cache objects whose values change over time

[pine grail]

so just react useState

[bronze crag]

wdym

[sharp epoch]

what is the benefit of this over using something like Caffeine with an expiry after access eviction rule

then again, I'm not sure whether there's worth in caching the cache

well, I retract that, since when dealing with distributed caches, there's some worth in having a local one. I just don't usually go for that approach since it makes things unnecessarily complex

[woeful pasture]

Cache<Cache<

[bronze crag]

Is it a concern? https://gyazo.com/58c124acacc47c813f1828a704a79628

[sharp epoch]

I mean, if you are using Redis, it gives you all the leeway to go for more shards if necessary

[bronze crag]

CacheWrapper is supposed to use Caffeine (instead the map it uses), this way you don't need to create a separate class which is scheduled to update amount of players every second, CacheWrapper does it for you in a convenient way

[sharp epoch]

but CacheWrapper would effectively be a single method to abstract out the getting from the cache

I don't find that particularly useful, however there might be more of the picture I'm not seeing here

[night knoll]

What I should add to my gui library? https://github.com/NazarbekAld/XeXeGui/
There's examples of existing features: https://github.com/NazarbekAld/XeXeGui/tree/main/tests/src/main/kotlin

[sharp epoch]

If you are willing to go for a change of focus: in general, if I want to do a non-trivial multi-tiered cache I'd use something like Spring's CA however using it outside of the Spring ecosystem is not really intuitive, so if I have a Redis cache layer then I'd end up using LocalCachedMap, or if I use other distributed cache options (i.e., hazelcast I believe provides a similar abstraction to what you're proposing now). Maybe a way to orchestrate all of that outside of Spring would propose more value, as nobody likes dealing with multi-tiered cache and making that more easier would definitely be a great scope

that said, it is infinitely harder to get that right, so be it as you will

[clever crag]

like my plugin

[bronze crag]

Thanks for an advice, I use LocalCachedMap but that's for statistics mostly, things like player count what is an integer and can't be easily cached in caffeine since it always changes require a little work around, that's why I thought CacheWrapper might be a solution for this problem

[junior holly]

https://github.com/NormalManV2/NormalDiscGolf

Any advice for any area appreciated ❤️

[rancid fern]

• Missing Readme
• Package name should be lowercase
• NDGApi has no real reason to exist? Everything is already present and accessible in the NormalDiscGolf class.
• Separate api from impl and place it in it's own module ^^
• Put test in it's own module (if you plan to keep those commands instead of using unit tests, otherwise move to src/test/java and use something like MockBukkit)
• Optional: Make a builder for your discs, record + record builder recommended :)
• The I prefix in the ITeam class is reduntant
• ITeam interface is not in api package
• DiscThrowEvent and GoalScoreEvent should be part of api

For project structure I recommend looking at LuckPerms

[junior holly]

Any syntax issues if you read that much into anything?

[rancid fern]

Didn't look too much at the code itself

[junior holly]

Ah nw

[alpine anvil]

wheres @woeful pasture when we need him

[round fossil]

centralize services around one api singleton

unless u're working with a really pedantic and complex system, keep it simple and minimalistic when u utilize static state

[junior holly]

Fair enough ❤️

[woeful pasture]

Pointless Abstraction https://github.com/NormalManV2/NormalDiscGolf/blob/master/src/main/java/normalmanv2/normalDiscGolf/impl/team/ITeam.java
Another Pointless Abstraction
https://github.com/NormalManV2/NormalDiscGolf/blob/master/src/main/java/normalmanv2/normalDiscGolf/impl/skill/ISkill.java
Is there even a point in skills sharing a common parent. It seems like literally every single skill is just inherited for some reason, but no actual extension is added. Is their even a benefit in a shared ancestor here?? I feel like a lot of these abstractions are just for the sake of "because OOP" versus actual intelligent abstractions. This entire skill thing could be done with a Map<String, Skill> If you ever wanted to add some form of modifiers to the skills you can literally just compose functions within the skills.

I feel like the whole "ThrowTechnique" Stuff could technically be implemented better but honestly its good enough as is

[woeful pasture]

don't worry I'm here now

[sly jackal]

as soon as you start with "I" for your interface you know you have too much abstraction

[rancid fern]

or you are using C# 🔫

[sly jackal]

c# is not an excuse to have bad naming conventions

[rancid fern]

Having I for interfaces is part of the C# convention

[round fossil]

^

[rancid fern]

All of their interfaces have it

[round fossil]

in java we have extends and implements to tell them apart

[sly jackal]

than C# is a bad excuse

[round fossil]

one could say :>

[sly jackal]

but in any case my point was more paired with teh fact that in java if you use I it means you probably have a class that implements it without the I, as in these cases. Which makes it kinda awkward to have any other implementation

since then the naming is weird by default

like if you have a ExtraSkill, why would it implement ISkill if it can just extend Skill

[round fossil]

well I mean its been proven often that having an over-abstracted system is preferred to an under-abstracted system, especially long term and when you start writing internal library modules and components

[sly jackal]

in fact thats what all the skills are doing anyways

yes, but I would pick your names better

[round fossil]

100%

[junior holly]

Well I moved all the interfaces to the api file anyway, I don’t understand why it’s considered a useless abstraction in terms of the skill objects, I feel like I was being lazy and didn’t want to actually write the getter/setters for each

Renamed the interfaces as well (without the I)

But if I could have a better explanation of why the abstraction is considered useless?

[round fossil]

tbh

i didnt read into ur infrastructure

but writing apis is hard

cuz every class can be considered an api to its consumers/users

every module may have a set of classes that are interfacing the module's functionality to the rest of the system etc

there's the literal interface in java, but api can mean a lot less, and a lot more

for example luckperms

they have an api module, but thats for other plugin devs

then they have their own module apis, that is internal abstractions to make the development within the project easier

[junior holly]

So do a lot of people design in a way that’s like “internal api” versus an external if you will

[round fossil]

esp when u have a lot of contributors and collaborators and you have to meet certain standards etc

[round fossil]

yea

on a package level, module level, system level, project level etc

[junior holly]

Mmm I see

[round fossil]

this is also something programming languages usually dont provide any semantics to us

meaning there's for example no built in annotation to communicate if a package, or class is an internal or external api and to what audience said api appeals to

@ApiStatus.Internal iirc exists in jetbrains annots library

but thats 3rd party

however, prefixing an interface with I is goofy

esp in Java

[dense leaf]

kotlin internal:

(just public to the JVM)

[round fossil]

ah w for kotlin

[junior holly]

Tbf that’s only because they were in the same place, moved now and naming conventions are fixed

It wasn’t meant to be permanent naming, just for the sake of the ide not yelling at me

[round fossil]

yea

push ur updates

and ill read through ur code more thoroughly

[junior holly]

👍 updated https://github.com/NormalManV2/NormalDiscGolf

All the interfaces were moved to the api package, though I feel GameRound could have remained where it was... not too sure on the semantics of api design obviously so

[round fossil]

alr so ur api sucks

[junior holly]

Yeha

[round fossil]

what is it for?

is it for other plugin devs?

[junior holly]

Honestly could not give you a reason... I just moved all the references there because I felt like main was getting bloated kek

[round fossil]

well thats the primary question id ask

whats the point of this api package?

[junior holly]

Eventually the goal was to make things accessible to other devs yes

[round fossil]

okay well, u probably completely wanna hide internals then

that is anything in the impl package

from 3rd party devs

[junior holly]

So in other words, the api I have is useless

[sly jackal]

i guess you mean api for programmers that wanna hook into your plugin from theirs?

[junior holly]

More so from an "extension" standpoint but yeah

[sly jackal]

or is this api intended to just extend your plugin directly

[junior holly]

That was the idea at first, but now that I'm hearing all this, it seems that's not a very uh supported way I guess lol

Like I said eventually the goal was to allow 3rd parties to make their own extensions/adjustments kinda thing

[round fossil]

the point of having an api module/package for other devs is that u can change the implementation of ur plugin without forcing plugins that depend on ur plugin to change

primarily

there are some other nieche things about it also

but else, its not so common to centralize interfaces in one package

usually u put classes in package by functionality or something like purpose

[junior holly]

I was thinking the only 2 interfaces that really make sense to stay in the api is skill and team, but then again the children are internal impl obviously, so does that even make sense?

[round fossil]

barely

[junior holly]

I guess I just don't really understand what should be considered internal/external

[sly jackal]

the api should probably be created separately from the rest of the code

and not also be the core of the implementation

[round fossil]

you should be able to compile the api code without having the rest of ur code on the classpath

[junior holly]

So then again, my api is useless right now kek

[round fossil]

little bit

it just looks like u were doing something for the sake of it

with no real goal in mind

[junior holly]

Don't sugarcoat it, I can take it man... I'm here to learn lol

[junior holly]

Correct lol

My idea of an api was to provide the accessibility to whatever I needed so perhaps I just don't understand what should be put where

public class Team implements normalmanv2.normalDiscGolf.api.Team {

    private final UUID ownerId;
    private final Set<UUID> teamMembers;

    public Team(UUID ownerId) {
        this.ownerId = ownerId;
        this.teamMembers = new HashSet<>();
    }

    @Override
    public void addPlayer(UUID player) {
        this.teamMembers.add(player);
    }

    @Override
    public void removePlayer(UUID player) {
        this.teamMembers.remove(player);
    }

    @Override
    public Set<UUID> getPlayers() {
        return Collections.unmodifiableSet(this.teamMembers);
    }

    @Override
    public UUID getOwner() {
        return this.ownerId;
    }
}```

So when it comes to the "pointless abstraction" bit that miles mentioned, the team object is meant to be used as a disposable object if you will. As for the skills like I already mentioned, I was being lazy and didn't want to write out those bits in each skill

[dense leaf]

why not import Team

[junior holly]

I could see how the team object doesn't need to implement a parent considering the usage goal... still though am a bit confused on the skills aspect

[junior holly]

ide did that automatically when I refactored 🤷

[round fossil]

i mean sure it could be considered pointless

it just depends on what ur appeal of the situation is

I generally abstract over almost anything because I often write unit tests, and Ive found that having interfaces can be really nieche long term for maintainability and reusability

^in the domain of object orientation

[sharp epoch]

I generally try to avoid abstracting anything when starting out a project

just prototype till I get the idea right, then start abstracting

it doesn't have to look pretty from the get-go, it just has to work

[round fossil]

very true

hard to predict what the ideal structure is gonna look like before u write it

[sharp epoch]

then again, everyone has their methods and it doesn't apply for every situation, but it usually helps harnessing the scope of the project, whether it'll be a time sink or not basically lol

[junior holly]

public class Skill implements normalmanv2.normalDiscGolf.api.Skill {

    private int level;
    private int maxLevel;

    public Skill(int level, int maxLevel) {
        this.level = level;
        this.maxLevel = maxLevel;
    }

    @Override
    public int getLevel() {
        return this.level;
    }

    @Override
    public int getMaxLevel() {
        return this.maxLevel;
    }

    @Override
    public void setLevel(int level) {
        this.level = level;
    }

    @Override
    public void setMaxLevel(int maxLevel) {
        this.maxLevel = maxLevel;
    }
}

...


public class Backhand extends Skill {
    public Backhand(int level, int maxLevel) {
        super(level, maxLevel);
    }
}```

My thought in terms of the skills was to enforce the structure by parent... though internally I only use the skill object to determine what skill and what the current level is, perhaps someone else wants to make a skill differing from my impl

Jeez man writing clean code is hard kek

[sharp epoch]

man use actual names that don't conflict with each other 😭

[dense leaf]

SkillImpl

or sth

[sharp epoch]

I know thinking of names is hard but just adding Impl or I at the start should do lol

[junior holly]

Thats so fuckin ugly

but fine

[sharp epoch]

BETTER THAN IMPLEMENT my.long.ass.fully.qualified.name.bc.im.lazy.Skill

[junior holly]

ok

you're not wrong

[woeful pasture]

I'm still confused what the rational for having a skill interface is

same with Team

if you just abstracted it because why not that's probably not a great reason. There is such a thing as too much abstraction. Its not like the interface is good API either

so I'm confused

[junior holly]

I think laziness is the biggest reason

I just thought it'd be best to enforce the structure, but even then I can't find a reason to keep it how it is

[sly jackal]

then dont

[junior holly]

Actually, I feel it just makes me try to think about how the structure of the object should go so to be fair, I think you could consider those interfaces as placeholders if you will. I guess I just like to think of what an object should look like and designing an interface for said object just makes that easier for me

[junior holly]

It's changed lol

[sly jackal]

yeah thats fair

there are also other options like making silly diagrams that arent part of the code

[junior holly]

Yeah it'd probably be a good idea to take to pen and paper before writing the code

Idk it seems like I love to write something out (whether its shite or not) get it reviewed and bum off others experience to make it better kek

Should I think about interfaces like this? An interface provides a concrete class structure in which the implementations may differ method to method
And a superclass like: A superclass provides default implementations and variables for it's subclasses

[jagged rock]

An interface is a sort of "contract" that defines a structure of its implementing class. One of its main purposes is to create a strict boundary between "how it works" and "how you use it"

[clever crag]

https://github.com/MinesonTeam/AdvancedAPI

Check out dev branch

[night knoll]

This guy dm spammed me this api

[junior holly]

Same lol

[rancid fern]

Yeah he likes to do that

[woeful pasture]

this my friends is why you have your DMs off

knowing this though I will probably not be incentivized to actually review this lol

[junior holly]

I mean

how was I supposed to know he was trying to like advertise

[woeful pasture]

hold up the real play here is for him to use Pineapple

the ultimate spigot library soon to be paper exclusive

[junior holly]

kek

[night knoll]

My DMs are just open for wor.

[night knoll]

BRO SAID WOW 💀

[junior holly]

Indeed he did

[night knoll]

Why do ppl use 362 APIs?

Oh, NBTAPI...

I mean just use NMS

You're not gonna need all NBTAPI methods

[rancid fern]

Saves time

[night knoll]

Yeah but uses a lot of reflection

on methods that you are probably not gonna use

Idk, that's my opinion

[woeful pasture]

Reflection is pretty well optimized now

[rancid fern]

True but if you're modifying so much NBT that the performance matters you're doing smth wrong

[rancid fern]

And this

[woeful pasture]

I mean to people just forget about how its backed by Method Handles which can, in some cases, achieve native call performance?

[night knoll]

Yeah, i know, but why do you need NBTAPI?

[woeful pasture]

pre PDC type shit

or if you need to modify some non exposed field

[night knoll]

You are gonna learn a lot more if you do it by yourself

[rancid fern]

Yeah the JVM can inline handles

[night knoll]

Lol

PDC saved me tbh

[woeful pasture]

praise lynx 🙏

[night knoll]

Also, i started reading other people code and i think my plugins is absolute GARBAGE right now

That's why im recoding it

[woeful pasture]

probably because it is garbage right now, but for ever garbage line of code you write you learn a little bit more 💯

[junior holly]

I personally like to spend the time figuring out how to do something myself, not only then do I better understand the underlying concepts but also I feel it helps me connect other dots perhaps when it comes to another system

It's also hard to say that what someone else has made will work for my specific goal

[night knoll]

Yup, i think this is a signal to say "psst! you're starting to learn"