help-development

[sly topaz]

nevermind, I was looking at the wrong javadocs

getTo is nullable

[crisp sleet]

https://prnt.sc/v5N5Tlg84Npnno work

[sly topaz]

that screenshot was removed

ah you forgot a space

[crisp sleet]

odd

[sly topaz]

try it in git bash

[crisp sleet]

https://prnt.sc/5y-pQ5pRCK_0

[crisp sleet]

never used git bash

[sly topaz]

just click this arrow and open one

[olive lance]

can u think of an example when it would happen?

[sly topaz]

when getTo returns null

[crisp sleet]

https://prnt.sc/Z4nLgHxwcPFm

[sly topaz]

var newLoc = event.getTo();

if (newLoc == null)
  return;

// do stuff

do that instead

[sly topaz]

try to execute buildtools from there

[olive lance]

right i will do that but I just am wondering when getTo would ever return null and why

[sly topaz]

this command

[sly topaz]

it is NotNull in Paper so I don't know honestly

probably an annotation forgotten in an upgrade

[olive lance]

interesting note

[wet breach]

if they move and teleport at the same time is an example

IE, standing in a nether portal and you move at the time you teleport, getTo would be null

there is other scenarios where getTo can be null

[sly topaz]

that would be for the teleport event maybe but not for move event

[crisp sleet]

no work
https://paste.md-5.net/vixuhutozu.coffeescript

[wet breach]

right but what happens if both packets come in?

because packets don't just stop once sent

so if the client says, hey they moved lets let the server know, and then rightt after the client goes oh times up time to teleport lol

[sly topaz]

delete the stuff that was generated and try again, it is complaining that you did it first on powershell and now you are doing it in git bash

[olive lance]

doesnt the server tell the client when it time to teleport or no

[crisp sleet]

ok trying again

[olive lance]

speaikng from laggy experiences

[sly topaz]

packets are handled by a queue so I doubt it'd be a problem, though I haven't checked exactly how Spigot handles them

I doubt it'd be all too different from Paper

at most, a desync in information can happen and the getTo would just outright be wrong, but for it to be null? I don't know about that

[river oracle]

some edge cases

its nullable for a reason

make sure you check for it

[river oracle]

There are cases where it can be nullable again

I think they gotta go back

[sly topaz]

ugh I don't wanna clone Spigot just to check this lol

welp

[river oracle]

check java docs

I saw machine maker say something the other day about making it nullable again

#925530366192779286 message

[sly topaz]

the javadocs say nothing about why it is nullable

or are you talking about paper's

[river oracle]

ah its because of teleports

I'm speaking of why paper is returing to nullable

idk what every edge case for spigot is tbh

[sly topaz]

but by the sound of it, it looks like more of an API conflict more than an actual behavior issue

[crisp sleet]

it worked now

thank you

[wet breach]

even if this was the case you still end up in a scenario where both happens

so if they teleport, and a move packet comes in, the getTo couldn't be set because they are in a different world

and getTo would refer to an invalid location

therefore its best its null

like technically you could say its accurate but it wouldn't be if getFrom is one world and getTo was another lol

I think having it null is valid, but some people believe nothing ever should be null lol

personally I rather have data absent then be given incorrect data

simply because someone hated nulls

[sly topaz]

data shouldn't be incorrect to begin with, the data having to be null is a symptom not the solution

[wet breach]

it is a solution

you thinking otherwise is your opinion sure

but the problem arising with wrong data is that now you have to check the data that it is correct

[sly topaz]

no, if the server can't properly process the packets sent by the player, then it is a server issue

[wet breach]

no

in a multi-threaded environment you simply can not avoid clashes when two things can happen

[sly topaz]

that is just not true, and what part of the packet processing in the server is multithreaded anyway?

when two things happen at once and you don't handle it, that's called a race condition

[wet breach]

well you are advocating that data should be stuffed into it, when it potentionally could be incorrect data that no longer applies. Which sure it works but it doesn't remove the need for checks which is what null would require anyways, just it is easier to detect no data vs incorrect data or data that no longer applies

[wet breach]

it isn't that it is not handled, just one is dropped in favor of the other

[halcyon hemlock]

is spigot single threaded

[wet breach]

not entirely

[halcyon hemlock]

mostly?

[sly topaz]

almost everything related to the world is handled in a single thread

[halcyon hemlock]

isnt spigot made relatively recently? like 10 years ago?

[sly topaz]

only chat and login proccess isn't

[wet breach]

not sure I would call that recent

paper is newer then spigot is

[halcyon hemlock]

recent enough for multi core cpus to exist innit

[wet breach]

yes

[halcyon hemlock]

im just curious as to why they went with single threaded instead of multi threaded

[wet breach]

but spigot doesn't drastatically change the way the game works or the code of it

[sly topaz]

because the vanilla server works that way

[halcyon hemlock]

spigot is an abstraction over bukkit right?

[river oracle]

spigot is bukkit now pretty much

[sly topaz]

spigot is the continuation of bukkit

[halcyon hemlock]

and paper is continuation of spigot?

[wet breach]

at this point in time you can consider bukkit and spigot one in the same

[halcyon hemlock]

or parallel?

[river oracle]

Paper is a fork of spigot

[sly topaz]

paper is a fork of spigot, they work in parallel to each other yes

[wet breach]

paper is like bleeding edge software built on top of spigot

[river oracle]

Spigot = Bukkit at this point

[halcyon hemlock]

so is paper multithreaded?

[sly topaz]

it isn't, folia is though

[halcyon hemlock]

minestom too right?

minestom is recent afaik

[sly topaz]

I don't know about Minestorm, never used it

[halcyon hemlock]

ye same

[river oracle]

Minestom will never be as viable as bukkit derivatives

[halcyon hemlock]

ive seen people use it

[river oracle]

sure but you won't get a good solid survival server out of it

[halcyon hemlock]

minestom is just an API to build your own server innit?

way more control

[river oracle]

unless you're literally insane

[halcyon hemlock]

if you want survival just use minecrafts own server innit

[wet breach]

one of these days I will release my own implementation

[halcyon hemlock]

the so called "nms" is bukkit or actual minecraft server?

[sly topaz]

vanilla server

[river oracle]

nms -> net.minecraft.server

[halcyon hemlock]

is bukkit an abstraction over the actual vanilla server?

[sly topaz]

yes

[halcyon hemlock]

and spigot even over that

[river oracle]

not anymore

[halcyon hemlock]

i see, java in real life

[river oracle]

number of spigot patches decreases every year pretty much

[halcyon hemlock]

so spigot is dying basically?

[wet breach]

no lol

[river oracle]

patches are just being moved to CraftBukkit

because we maintain craftbukkit

[halcyon hemlock]

ah I see

[sly topaz]

craftbukkit is the server modifications to make the bukkit abstractions work

[halcyon hemlock]

I was just curious since im making my own server framework, and dont want to make the same mistakes as bukkit or others

[sly topaz]

if you want to make a server platform from scratch, I'd recommend talking to the Sponge guys

[halcyon hemlock]

i dont think itll ever compete with spigot in terms of plugins shit

since entirely different langs

[wet breach]

not sure what goals you have, however my implementation if I ever get around to making it won't contain the majority of mojang code 🙂

[halcyon hemlock]

maybe provide a scripting language like lua or some shit

[halcyon hemlock]

yeah same, mine is a clean-room implementation as well

just looking at wiki.vg

[wet breach]

also, I am going to turn it into a proper server setup

[sly topaz]

I mean, it doesn't matter even if it is the same language, look at fabric, minestorm, sponge. They barely get any adoption

[river oracle]

clean room implementations won't ever really be viable unfortunately

[wet breach]

IE separate some components into their own processes

[river oracle]

it takes too long to update

its definitely a cool project though

[wet breach]

mine isn't going to be entirely clean room

[halcyon hemlock]

🤔

[wet breach]

and you simply can't anyways due to login stuff

[halcyon hemlock]

login stuff? encryption?

[sly topaz]

what do you mean by that

[halcyon hemlock]

everythings available on wiki.vg AFAIK

[river oracle]

he means you eventually have to delegate to mojangs servers for auth

[sly topaz]

the whole protocol is very well documented, so clean room is possible

[wet breach]

tampering with player login functionality is considered removing DRM

[sly topaz]

ah but I mean, that's entirely different

[wet breach]

yeah that

[halcyon hemlock]

yea thats just using an external API provided by mojang to authenticate users

[sly topaz]

could go nuts like the fosscord guys and implement the whole mojang backend like they did for discord lol

[wet breach]

anyways, the reason I am not going with a complete clean room implementation is because there is some things I am too lazy to do myself

such as pathfinding etc

or entity logic

I ain't going through and making all that myself lmao

[sly topaz]

that's the easier parts, what hurts is world and chunk gen

[halcyon hemlock]

pathfinding is easy innit

[wet breach]

those are easy

[halcyon hemlock]

chunk gen, if done in proper steps shouldnt be that hard

would still take time though

[sly topaz]

a proper, efficient implementation is no walk in the park

[umbral ridge]

Yes

[wet breach]

well the thing about making your own implementation in regards to those things is that you don't have to follow how mojang does it

you can choose whatever method you want

[halcyon hemlock]

yeah thats true

[wet breach]

the downside of doing that is that seeds online won't work the same lmao

[halcyon hemlock]

for world gen you can start with perlin noise and come up with something

[wet breach]

otherwise you can use whatever randomization method you want

some are more optimal then others depending on what you are making

[halcyon hemlock]

have you worked on your impl? @wet breach or do you have to start?

[wet breach]

kind of a bit of both. The first thing I need to work on is separating the components

[halcyon hemlock]

cuz if u gonna start why not work together

[wet breach]

my implementation will require 3-4 processes

processes as in server processes running

[sly topaz]

I did an impl of the server myself, not clean room tho and for 1.8

[wet breach]

I am going to have a login server, chat server, world server and some others

[halcyon hemlock]

huh

why

[sly topaz]

splitting the logic in processes sounds like a bad idea

[wet breach]

because splitting it like this means I can dedicate the resources to the things that really need it

a chat server doesn't require much and same with a login server

you can easily have a slim VM for that

then you have your world server which you can spread out over many VM's and delegate resources as needed

[halcyon hemlock]

you could have green threads innit

for every task

right now every conection for me is handled on a different thread

every system (keep alive, connection receiver etc)

[sly topaz]

they are looking for an arch which will allow for horizontal scaling

[wet breach]

and the upside of all this is I can reboot the world servers without players getting kicked

[halcyon hemlock]

fair

[sly topaz]

I mean, you can do all of that without splitting into processes

the only upside of process splitting a server impl is horizontal scaling, but at the cost of IPC latency

[wet breach]

not sure what you are going on about

there won't be latency issues

chat is not dependent on the world

also, the way I intend to scale the world server is by doing something along the lines of sharding

[sly topaz]

that's what horizontal scaling is

[wet breach]

ok, I don't really care not sure what issue you have

sharding would allow both vertical and horizontal scaling

vertical scaling just has to do with adding more resources or power vs using more nodes

but given technology today such as VM's becoming a whole lot better your horizontal scaling can also be vertical at the same time

nothing says VM's have to be on different hosts

anyways, the reason for sharding is because it would be more efficient to have a separate node for an area of a map. Since I like survival that is what my implementation is geared to

each node is assigned a piece of the world, and if a node goes down another node can easily take over if need be, or you can add more nodes if a shard starts becoming overwhelmed.

one of the main reasons I want the chat server to be its own thing

is because I am fan of allowing people to connect with just a chat client

other then we don't need the main server processes handling that stuff, as well as if something happens in the world chat won't freeze either

I also intend at some point in creating a custom client as well

I do not like the vanilla client

[river oracle]

now

will he actually start it

I'll give it a 0.01% chance

[glad prawn]

is there any way to completely disable the cobblestone generator?
cancel the event just makes liquid not flow together, i don't want that.

[wet breach]

lol

[alpine narwhal]

nah 50% he starts, 0.01% its finished

[river oracle]

I'd say the odds are more reversed for frost

[wet breach]

y2k is most likely correct XD

however never know

[rough ibex]

I've had one of those 'holy shit this bug was so easy to fix' moments

my computeIfAbsent lambda was modifying the map, not returning the value NotLikeThis

[buoyant viper]

whats default getWalkSpeed of Player again

0.2f?

[drowsy helm]

Yep 0.2

[dawn flower]

actually it's 0.20000000298023224 🤓

stop spreading misinformation!!11

[buoyant viper]

i like Action.PHYSICAL's documentation

[blazing ocean]

love that

[rough ibex]

who wrote that

[blazing ocean]

this ones better tho

squeeehkkk

[shadow night]

The people had some fun

[rough ibex]

easter eggs

[buoyant viper]

does LivingEntity#damage(double damage, Entity source) modify damage based on source GENERIC_ATTACK_DAMAGE attribute or does it deal the direct value, and does it take into consideration the entity being hit's armor protection attribute?

im just having this really weird issue where its dealing correct? damage to all other mobs except Zombies

ie.

Player player = getPlayerSomehow();
Mob mob = (Mob) getVictimSomehow();
mob.damage(mob.getHealth(), player);

and with nothing in my hand, this will not insta kill the Zombie, however if mob is anything other than Zombie, it will

[rough drift]

try doing negative health

[halcyon hemlock]

I added reflections to rust

what are we rating this

frfr

its compile time too

instead of run time

unlike java

[hazy parrot]

How would compile time reflection work lol

[halcyon hemlock]

derive macros

its in rust

so when compiling code, its basically a macro to loop through each file and look for specific type of structs

the system is pretty clean tbh, the serialization

[blazing ocean]

doesn't serde do that

[halcyon hemlock]

serde is slow

the overhead

i compared with 3 other libs and mine is still faster, for no reason

i didnt even do stupid simd shit

[blazing ocean]

i do love me some rust macros

[halcyon hemlock]

derive macros are just awesome

dont know what id do without them fr

[blazing ocean]

real

also extension traits my beloved

[buoyant viper]

this sorta just ended up doing nothing to... everything

like now nothing takes damage lol

kinda funny

[halcyon hemlock]

extension trait?

pub TraitA: TraitB?

[sly topaz]

literally what they are called

traits that extend structs/other traits

would be the equivalent of extension functions in kotlin

[blazing ocean]

trait StringExt {
    fn push_newline(&mut self);
}

impl StringExt for String {
    ...
}

let mut string = "Hello, world".to_string();
string.push_newline();

[sly topaz]

it's just syntax sugar for a wrapper. Maybe a static method that takes the object is a better description

[halcyon hemlock]

ah yeah

[rough drift]

doesn't zig do that by default

like it has that builtin

[halcyon hemlock]

it has a builtin nbt serializer?

damn who built the language

[sly topaz]

lmao

[rough drift]

no like

the reflections

smh

[sly topaz]

they are talking about comptime

[blazing ocean]

comptime is weird

[halcyon hemlock]

ic ic

a camera as well

aww man you cant name nbt compounds as crabs

isnt it utf8

[sly topaz]

what is

[blazing ocean]

now do that for dfu

[halcyon hemlock]

dfu?

[blazing ocean]

datafixerupper

mojangs serialisation lib

[halcyon hemlock]

its not rust

[blazing ocean]

yea

[halcyon hemlock]

how would I bench it

[ivory sleet]

You could rewrite it in rust

[halcyon hemlock]

theres that

[blazing ocean]

jni
create a jar that uses it and run it in rust

[halcyon hemlock]

too much effort

let me try bench my new approach, maybe its way slower now

[sly topaz]

minecraft's is most likely more efficient given the scale

[halcyon hemlock]

fr?

[blazing ocean]

not sure since it's mojang code

[halcyon hemlock]

tf does it even do

its so big

mine is barely 2-3 files

[sly topaz]

code size barely matters in terms of performance

[blazing ocean]

well it has pretty much all the serialisation from minecraft

[halcyon hemlock]

they have fucking papers on it bro

i aint competing with them

[blazing ocean]

damn mojang got docs?

crazy

[halcyon hemlock]

im surprised they didnt obfuscate their open source library

thats a first

[blazing ocean]

how

brig ain't obfuscated

[halcyon hemlock]

https://media.discordapp.net/attachments/1128087376820977825/1129342197997776911/ED6EFBC5-A68E-48C6-9103-B123B1ECC22D.gif?=&width=507&height=120

[ivory sleet]

[blazing ocean]

https://media.discordapp.net/attachments/193073071802941451/1063698188575440970/Screenshot_2.gif

[halcyon hemlock]

thonk

[sly topaz]

you guys do know that a big GIF appears on these right

lol

[blazing ocean]

doesn't for me

compact mode

[ivory sleet]

Discord on phone doesn’t show that unfortunately

[halcyon hemlock]

https://cdn.discordapp.com/attachments/700675087812132964/1265580531719082116/spoder.gif

https://cdn.discordapp.com/emojis/1099380698764300360.gif?v=1&size=48&quality=lossless

[blazing ocean]

https://tenor.com/view/shower-soap-friendly-reminder-you-stink-gif-15157732979346099491

[halcyon hemlock]

wtf

if I do Vec::new() it takes about 350ns as average

[blazing ocean]

[halcyon hemlock]

if I do Vec::with_capacity(2048) it takes 150ns

lmao???

[blazing ocean]

because without args it has unlimited capacity?

[halcyon hemlock]

prolly cuz it keeps increasing size

and allocating

[blazing ocean]

yes

[halcyon hemlock]

hmm how could we guess the size 🤔

time to use unsafe allocation 😋

[blazing ocean]

why do you even need a 200ns optimization

average rust user

https://cdn.discordapp.com/attachments/957661529447202906/1268207751952076942/ELX21fp7.standard-1.gif

[halcyon hemlock]

cuz this is a relatively small struct

a chunk maybe a thousand times bigger

[blazing ocean]

me when kotlin

(goat)

me when

i love null chaining

ye

s

[halcyon hemlock]

bro getting excited with the network handler

[blazing ocean]

!! means "it's not null trust me bro"

basically assert != null

[halcyon hemlock]

ye ik

like typescript

and dart

[blazing ocean]

kotlin does

rust format!()

println("$myInt + 1: ${myInt + 1}")

[halcyon hemlock]

a single u8 takes 34 nanoseconds, no way

my benchmark is definitely broken lmao

[blazing ocean]

stop trying to get nanosecond optimisations and build shit

[halcyon hemlock]

i agree

so real

[sly topaz]

sadly they withdrew the string templates JEP, see https://youtu.be/c6L4Ef9owuQ?si=CuyKyG83f694ikp2

[fossil fiber]

i got an idea on what plugin I need but idk how to code.. anti lava and water from forming cobble/obby flags, need worldguard for dependencies,

[tardy delta]

What tool are you using? Perf?

[halcyon hemlock]

criterion

[sly topaz]

I mean, that kind of benchmark isn't even measuring real performance of anything, it is just playing around with the benchmarking crate

[halcyon hemlock]

thats a lot of 0s

anyways imma go do actual shit now

[blazing ocean]

what have you done

[halcyon hemlock]

make my code slow on purpose cuz its too fast, duhhh

sleep(Duration::from_millis(10));

[sly topaz]

premature optimization is the root of all evil people, remember

[halcyon hemlock]

the thing is. I haven't done any optimizations

ok well cya

[sly topaz]

I'd just sleep on it if I were you

otherwise you'll get burned out and never finish the project

when I am doing dumb shit inside a project, I know it is time to get away from the keyboard

[umbral ridge]

yes

[halcyon hemlock]

nothing to do

¯_(ツ)_/¯

[sly topaz]

go for a walk

play with a pet

[umbral ridge]

or its time to check on your cat

[halcyon hemlock]

i dont have pets

[sly topaz]

I can't recommend having one with a good concience, I love animals too much but I am too broke to have them

[umbral ridge]

I love my cat

[sly topaz]

and they are basically babies that never grow up

[umbral ridge]

XD

[halcyon hemlock]

i dont want additional responsibilities

i already have too much shit to do

[sly topaz]

mine is sleeping right above me right now

[ivory sleet]

Wasn’t mainly just put on ice since they didn’t know how to redesign it?

[halcyon hemlock]

how many lookups are you doing per tick?

[eternal oxide]

Do NOT do db access on the main thread

[ivory sleet]

Yes

[sly topaz]

26 seconds, jeez

[ivory sleet]

Likely will

[sly topaz]

you also may wanna check your queries, that's too long for any connection

is your db hosted locally?

[eternal oxide]

hosting is always shit

may not even be on the same box

[sly topaz]

well, then you won't ever notice latency issues that way

[ivory sleet]

Well @blissful crow dont be scared, string interpolation is likely to come back in the distant future

or “string templates”

if you now will

[sly topaz]

kinda, yeah

[ivory sleet]

very so

[rough drift]

hey I do that and it's fine! Ignore that it is lazily loading db files under 10kb

[sly topaz]

Java 8 has long not been industry standard

[ivory sleet]

You’re on thin ice saying that

[rough drift]

I will pasta la vista you

[sly topaz]

I mean, it isn't fine, but like everything done in a single minecraft tick, it'll work out

[rough drift]

yeah it's done once mostly at startup

or on the first occurrence of something

then everything else is done async

[sly topaz]

did you do it to avoid the context switch penalty or just found it easier to not care about it

[rough drift]

I did it because the data can't be loaded later than exactly when requested

so it has to run on the same thread

plus it loads in <5ms

[tardy delta]

"yet we should not pass up our opportunities in that critical 3%" is the part people tend to forget

[sly topaz]

that just means your design isn't designed around the idea of being non-blocking, normally you deal with that by using completion handlers everywhere

[rough drift]

yes but that would be slower

[sly topaz]

not necessarily, just a lot more tedious than what it's worth

[rough drift]

var tag = tags().get("tag");
tag.has(UUID)
...

vs

tags().get("tag").then(tag -> {
  tag.has(UUID);
  ...
});

I like the first one better

[tardy delta]

laughs in STR."hello \{name}"

[rough drift]

you may have more variables to pass in

that causes new instances to be made

[ivory sleet]

😔

[sly topaz]

hopefully these fancy concurrency features such as scoped values and structured concurrency APIs get out of preview soon™️ so it becomes a whole lot easier

[tardy delta]

tbf i saw them being used in minestom

[rough drift]

scoped values are a bit funky imho

[ivory sleet]

Lol really now?

[tardy delta]

for things like minimessage, mightve been a personal lib

[sly topaz]

I mean, it exists in JDK 21 as a preview, so I wouldn't be surprised

[ivory sleet]

I mean both the struc conc apis are in their third preview, hopefully for the sake of love and god we get those fully delivered in j24

[sly topaz]

nah, they are going to wait till 25 if I remember what I read in the mailing lists correctly, which is next LTS

[rough drift]

java is getting some W updates

[ivory sleet]

ah, that’s interesting

[rough drift]

but honestly the one I'm most excited about is List<int>

[blazing ocean]

yea i hate that

so much

[ivory sleet]

Well at least its on lts then

[blazing ocean]

why couldn't they just be normal

and be like kotlin

[rough drift]

because of some safety thing that was outlined in the JEP

it's apparently safer

[blazing ocean]

🤨

[sly topaz]

it was something about performance, but they later discovered it wasn't necessary and thus the JEP fell apart

[ivory sleet]

valhalla is gonna be great

well good enough

[rough drift]

valhalla is going to be a W

[ivory sleet]

Yup

[rough drift]

imagine spigot with valhalla and 401

major W

[sly topaz]

we are 1 month away of Java 23 too

[rough drift]

yo??

what they addin

[sly topaz]

https://openjdk.org/jeps/455

this is a good one

[rough drift]

holy shit

[blazing ocean]

kotlin when vibes

[rough drift]

can you shut the fuck up about kotlin

[blazing ocean]

no as in

that's literally what kotlin has

they are just copying kotlin shit

[rough drift]

yeah I am aware, a fuck ton of languages have this

maybe java is just updating and adding modern features?

[blazing ocean]

shocker

[ivory sleet]

Its also interesting gen zgc by default is just shipped fully w little preview or anything, in j23

I mean I suppose its been tested enough to outperform normal zgc

[rough drift]

I can't wait for java 100 🔥

[tardy delta]

are we using zgc by default already

[rough drift]

gen zgc

[ivory sleet]

^^

[sly topaz]

https://openjdk.org/jeps/473 is one I've been waiting to get out of preview for a while too

[rough drift]

Yeah

A lot of my optimizing code will need to be updated for all of these JEPs

but they just end up giving me more data to optimize better so W

[sly topaz]

with this I can stop doing dumb shit like calling map twice and abuse the hell out of method references kek

[ivory sleet]

Its not been out for too little long, doubt that’s the last preview but we can hope

[sly topaz]

it is going to be ready for 25 most likely, it isn't a big one

https://openjdk.org/jeps/469 however, I hope it can just get out of incubator lol

thing's going through so many phases

[ivory sleet]

Yea as long as things can be fully delivered on the lts version its pretty accepted

[rough drift]

def can be

they are pretty minor

[rough drift]

Eighth incubator

lmao

[tardy delta]

more simd intrinsics?

[ivory sleet]

Yea, although fully makes sense why it went through a lot of stages

[rough drift]

yeah

[sly topaz]

it's just waiting on valhalla now

[sly topaz]

yep, with this you can do a lot of fun stuff

[rough drift]

valhalla is so W

but I also like most other projects

like panama

[ivory sleet]

Doubt valhalla will be fully delivered by 25 tho

[sly topaz]

Java has a lot of auto-vectorization code but it didn't have many utilities for manual vectorization until that API arrived

[rough drift]

extremely doubtful

[tardy delta]

dunno why i would need simd in java tbf

if i would, i wouldve written it in c

[rough drift]

It is so fucking fast

and it's super easy

[sly topaz]

the same reason you need simd anywhere else, to make things faster

[rough drift]

it's like huge speed boosts for very little

[tardy delta]

working with __m128 is also super easy 👉👈

[rough drift]

DoubleSpecies.MAX_DOUBLE or smt and you can make instances of these ones and just sum n values together in one op

[sly topaz]

the good thing about it being in the JDK is that it's going to eventually be fully cross-platform, so you would be able to write very performant SIMD code that works on various architectures, not like now where you have to be very mindful of it with lower level languages

and the JDK devs are very competent so they'll make something that works well out of the box, anything else will come from library makers

[rough drift]

tbh why write in other langs than Java when you aren't in embedded envs

if done right esp with the releases that are coming it is just a productivity boost

[tardy delta]

love it when ike reacts to his own message

[rough drift]

you gotta commit to the reaction

[buoyant viper]

hmm

[dry hazel]

the jep has been withdrawn

they are incorporating some changes

[rough drift]

W

[sly topaz]

https://blog.jetbrains.com/idea/2024/07/is-java-still-relevant-nowadays/

might be of interest to you

it isn't talked about a lot, but people are using Java a bunch in the enterprise side of things

[halcyon hemlock]

rate my IDE

[tardy delta]

rust 🗿

[sly topaz]

RustRover huh

[tardy delta]

is there like a reason to use rr over vscode or smth

[sly topaz]

not really

rust-analyzer (the lsp), is really good

to the point you could use notepad for what is worth as long as it supported LSPs

[tardy delta]

just learn vim

[halcyon hemlock]

cuz its way better and literally meant for rust?

[tardy delta]

better than what

[halcyon hemlock]

you can enable it in rustrover as well

[halcyon hemlock]

vscode

[tardy delta]

the vscode extension is also literally meant for rust

[halcyon hemlock]

i just dont like vscode

simple

i have all jetbrain ides for everything i do

they do best what they're meant to do

vscode seems too basic

[tardy delta]

average jetbrains suite enjoyer

😂

[sly topaz]

vscode is too basic

but you don't need anything fancy for rust

I don't think jetbrains can offer anything of actual value to me as a dev in RR that I wouldn't be able to achieve in my neovim setup

[tardy delta]

same

[halcyon hemlock]

its free

¯_(ツ)_/¯

i would rather use (neo)vim than vscode

[blazing ocean]

rustrover 🗿

zed >>>

[tardy delta]

zed looks fairly limited

[blazing ocean]

it's pretty good

or just use neovim

[buoyant viper]

how small is ur monitor

[halcyon hemlock]

i tried zed

its not mature enough tbh

[halcyon hemlock]

1920x1080 why?

it aint that small 😢

[buoyant viper]

just looks crammed idk

[halcyon hemlock]

i have big font

[blazing ocean]

i mean

it has all you need

what more do you need than terminal, lsp, editor and files

[quaint mantle]

Zed is written in rust 🗣️

[blazing ocean]

yea

https://cdn.discordapp.com/attachments/957661529447202906/1268207751952076942/ELX21fp7.standard-1.gif

[tardy delta]

isnt that.. .the resolution

meanwhile all other fast langs:

its just llvm

[halcyon hemlock]

being comfortable

[blazing ocean]

huh?

[halcyon hemlock]

the screenshot is 1920x1080

[halcyon hemlock]

also i had to compile it myself

[blazing ocean]

binaries exist

[halcyon hemlock]

not for windows innit

[blazing ocean]

windows 🤡

[halcyon hemlock]

ok

I like playing games

All linux shit is already in wsl

¯_(ツ)_/¯

[wintry dagger]

anyone good w java generics please dm

[blazing ocean]

?ask

[undone axle]

If you have a question, please just ask it. Don't look for staff or topic experts. Don't ask to ask or ask if people are awake or available. Just ask the question to the channel straight out, and wait patiently for a reply. Make sure you use the right channel regarding the topic of your question. Create a thread in case the channel is already in use!

[wintry dagger]

i cant send images

[tardy delta]

?img

[undone axle]

Can't send images? That's because you're not verified! Use !verify to complete verification.
Alternatively, you can upload screenshots to any image hosting site and share the link.

Here's some screenshot utilities that you can use to upload images.
Lightshot: https://prnt.sc
Imgur: https://imgur.com/upload
Flameshot: https://flameshot.org

[wintry dagger]

!verify

[undone axle]

Usage: !verify <forums username>

[wintry dagger]

I see the problem

I’m too stupid to see the solution

[drowsy helm]

field is not typed

you will have to do TicketField<T> then cast it as that again

unless getInput takes a class to implicit cast

[rough drift]

which is a downside

rust is just an automatic downside

[tardy delta]

its readable

[rough drift]

joke of the year award goes to:

[blazing ocean]

you're a downside

[rough drift]

FourteenBush! With their joke "Rust is readable"!

[tardy delta]

lol

[wary harness]

anyone can recomend good discord api for bot integration

[tardy delta]

jda

[wary harness]

I got shop plugin which I want to hook in to discord

[rough drift]

You LITERALLY use kotlin and rust unironically

[blazing ocean]

d4j

[rough drift]

java.net.URL

[tardy delta]

never heard of

[wary harness]

jda or d4j ?

[blazing ocean]

jda

[wary harness]

which one is better and easier?

[blazing ocean]

just use jda, more popular one

[wary harness]

thanks

[wintry dagger]

field type isn’t always = data type, hence the second if check comparing the classes so i cant just cast it

unless i just put in a try catch

[drowsy helm]

probs the best option

[wintry dagger]

try catch ye

figured it might be

[lost matrix]

You wont be able to catch an unsafe cast

[drowsy helm]

wdym, doesnt it throw?

or you can do two TicketField variables

one type ?, other type T

check type then if applicable, cast

[lost matrix]

There is probably a way to make this kinda type safe. But i have no idea what this is.

[drowsy helm]

public <T> boolean addInputData(String type, T data){
  TicketField<?> field = this.category.getInput(type);

  if(!data.getClass().isInstance(field.getType())){
    return false
  }

  TicketField<T> castField = (TicketField<T>)field;
  //do stuff
}

[lost matrix]

If you keep the type checking encapsulated with an implicit relation, you can use unsafe casts but be technically type safe.

[drowsy helm]

not semantically type safe, but techincally type safe

[lost matrix]

Pretty much

[wintry dagger]

This prob work tbh i’m not on pc rn will check later

[lost matrix]

Eg if you have a Map<Class<?>, SomeObj> and stricly limit access to this map, you can safely cast later.

[drowsy helm]

yeah, just a java limitation atp

dont think theres any other way of enforcing it?

[lost matrix]

Hm, i cant point my finger at it, but i feel like isAssignableFrom should be used here somehow

nvm, ignore me, i just got up

[drowsy helm]

that would work i think

[lost matrix]

I feel like there is a better solution. The TicketField<T> could have a Class<T> getFieldType() method.

What even is String type ???

That doesnt look right

[drowsy helm]

inputData is a map im assuming

with key being data type?

[wintry dagger]

Ye key is the name of the field, a string, value is well, object

like am just messing with input fields similar to what discord has with like NumberInputField, StringInputField etcetc that provides a universal interface for all field types

but generics / that might be the wrong approach even idk

[rough hinge]

If in inventoryclick the player clicks on the topInventory inv, will topinventory.getcontents return the updated items?

@EventHandler(priority = EventPriority.HIGH)
    public void onInventoryClick(InventoryClickEvent event) {

        Player player = (Player) event.getWhoClicked();
        Inventory inventory = player.getOpenInventory().getTopInventory();

        System.out.println("antes");

        if (!inventory.getName().equals("Backpack")) {
            System.out.println("TOMAA" + inventory.getName());
            return;
        }

        System.out.println("depois");

        Backpack backpack = plugin.getBackpackService().get(player.getName());
        updateBackpack(backpack, inventory);

    }```

[lost matrix]

The event is fired before any changes where made to either the cursor or the inventories

[rough hinge]

then what can i do?

[lost matrix]

Hmm, i would recommend having a coupling to actual classes.

public <T> boolean addInputData(Class<T> type, T data) {

And if you need them deserialized from Strings, you could call Class.forName()
But i would need a bigger picture

[lost matrix]

First of all, you should never use the name of your inventories as identifiers.
People can just rename a chest, place them and then basically spoof all of your guis.

[lost matrix]

What are you trying to do in the first place?

[rough hinge]

not have nametags

[rough hinge]

backpack

basically if a person put a item on backpack inventory, the backpack#getContents() will update and update on database too

[lost matrix]

Backpacks should be straight forward.

1. When the player opens a backpack -> Load the content and put them in an inventory
2. When the player closes the backpack -> Go over the inventory and save everything

There is no need to sync/save the backpack on every click. Just open a plain old inventory and let the players
throw their stuff in there.
Then save it on close and load it on open.

[wintry dagger]

the reason for the string type is because of the input field’s configurable name, it’s not related to T or classes

Like I could have a IntField with the name Salary, and addInputData parameters would then be addInputData(”Salary”, 24000)

[rough hinge]

if the server closes and the player is editing the backpack?

[rough hinge]

Do you recommend comparing the arrays in inventoryclose before saving to the database?

[lost matrix]

Then you need a Registry mapping Strings to classes. But at some point you need to convert Strings to something more type safe.

[lost matrix]

No, just save the entire inventory and override whatever is in the database before

[wintry dagger]

Bad idea having a string name as key yeah?

[lost matrix]

User input always results in Strings

Honestly i have no idea how your system works and who can register new types.

[rough hinge]

Could you tell me why? What if the server restarts with the backpack open?

[lost matrix]

Make sure to close all open inventories when the server stops.

[rough hinge]

if server crashes -.-

[lost matrix]

Then its gone

[rough hinge]

But how do I close all inventories when the server closes? Doesn't this happen anymore? Do I have to go through all the players and close the inv in the ondisable?

[wintry dagger]

Validators are predefined (URL, Number) etc.

Fields are configurable. The validators should simply serve as a universal standard for adding new fields of a specific type, e.g. a Url Input Field, because it’ll validate the user’s string input later submitted and check if it’s a url for instance

[lost matrix]

Pretty much

[rough hinge]

Bukkit.getScheduler().cancelTasks(this); i need make this on disable?

[rough hinge]

then?

[lost matrix]

Sadly, for some reason, spigot doesnt kick all players before stopping.

[tardy delta]

no, happens implicitly

[rough hinge]

Should I kick all players or close the inventory?

[lost matrix]

Really depends on what plugins are loaded at this point. If its a single plugin server, then kick them.
This will result in the PlayerQuitEvent being fired for all of them. Alternatively you can manually save all of your Managers live data in onDisable.

[rough hinge]

In this case, the only thing I have to save is just that, the rest is all synchronized.

and it is a unique plugin

I'll choose to kick anyway

[lost matrix]

Btw, this has a tricky implication.
You should never run IO operations on the main thread. (Accessing Files, Databases, Web endpoints, etc)
unless you are in your onEnable/onDisable. This means you need a flag that you can set, to switch between sync/async access.
The inventory should be saved in the InventoryCloseEvent, but this should happen async, unless it is fired as a result of your onDisable kicks.

Just keep that in mind. Running IO in events or other parts of the main thread will result in lags.

[rough hinge]

Oh, I disagree with you.

[lost matrix]

Which part

[eternal oxide]

Don't be a fool

[rough hinge]

First of all, why would I use an asynchronous database if I have to use the CompletableFuture's .join method? It will even use more resources and stop the main thread for the same amount of time!

[eternal oxide]

You don;t understand Futures

you do not call .get on a future

it defeats the whole purpose

[onyx fjord]

whats the event for sapling changing its stage?

[lost matrix]

You should not .join() your futures. Ill write you an example which lets you load the data async, and open the inventory sync afterwards.
The main drawback is that you might need to wait 1 or 2 ticks. brb

[rough hinge]

Imagine, if I use .thenAccept or something like that, let's consider that it takes 5 minutes for the response to be returned. A get is used that returns the player's coins and it takes 5 minutes to return the response that will return 1 million coins. However, the player has 0 coins and sells an item in the store for 100 million. The player will receive the 100 million in the cache and will set it in the database, then after 5 minutes the database will overwrite the cache and the player will only have 1 million.

So in most cases I'm forced to use .join, which makes it completely useless and just wastes more resources.

It uses more resources because several tasks that will be executed in another thread are rewritten, and it will end up overloading the system as well.

[wet breach]

your imaginary scenario is flawed

[lost matrix]

  public ItemStack[] loadItemsFromDatabase(UUID playerId) {
    // Load items from database
  }

  public CompletableFuture<ItemStack[]> loadItemsFromDatabaseAsync(UUID playerId) {
    return CompletableFuture.supplyAsync(() -> loadItemsFromDatabase(playerId));
  }

  public void openBackpack(Player player) {
    BukkitScheduler scheduler = Bukkit.getScheduler();
    JavaPlugin plugin = ...;
    UUID playerId = player.getUniqueId();
    Inventory backpack = Bukkit.createInventory(null, 27, "Backpack");
    // Load items async
    CompletableFuture<ItemStack[]> futureItems = loadItemsFromDatabaseAsync(playerId);
    // With thenAccept you can define some code that should run after the future is completed
    futureItems.thenAccept(items -> {
      // This code is used to get back to the main thread
      scheduler.runTask(plugin, () -> {
        backpack.setContents(items);
        player.openInventory(backpack);
      });
    });
  }

[lost matrix]

If you call .join() on a 5 minute query, then your server simply crashes on the spot.

[wet breach]

when you have a cache system in place, you should storing data in the DB right away before it goes to cache since cache can be lost at any moment. If you follow this, there shouldn't be a scenario where your coins get overwritten

[rough hinge]

**Imagine, if I use .thenAccept **

[rough hinge]

And thenaccept will prevent server crash

[rough hinge]

even if it is in the db before the cache, the problem I mentioned will continue

[lost matrix]

Look at the example ive sent. Lets not deviate from backpacks, because currency is an entirely different topic.

[wet breach]

no it won't unless you are just completely ignoring the DB

[rough hinge]

What do backpacks have to do with this? You said that queries etc. in the database have to be asynchronous and I said that 99% of the cases do not

[rough hinge]

brother, I just sent you an example that asynchronous db will result in data inconsistency

[wet breach]

the reason they should be async is because you can encounter a long query for whatever reason and it will hault the main thread

[lost matrix]

In 100% of cases they absolutely need to be async during playtime. There is no exception because you cant risk your server crashing because your DB has a hickup.

[rough hinge]

but if you use .join it will be worse LOL

will stop the main thread anyway and still use more resources

[lost matrix]

Yeap. Do not use .join() on the main thread.

And lets keep caching out of this discussion for now.

[rough hinge]

because the DB having connection errors is not important at all, right? All the server data will simply not be saved from then on.

and even because nothing is important for the player to keep his data in the database because the guy decided to make the DB asynchronous in 100% of the systems 🤡

[pseudo hazel]

you dont wanna crash the server due to freezing atleast

[rough hinge]

and even because a get in mysql is very expensive 🤡

[pseudo hazel]

the player is impacted by the server not responding for several minutes afaik

[lost matrix]

connection recovery strategies have nothing to do with this. If you have actual connection problems to your database during playtime, then this should not be handled by crashing the server on the spot.

[wet breach]

alright I see we have an inexperienced person and I am just not going to bother. DB tables can be locked while performing operations where you don't want overwriting to occur like in your scenario. If the table is structured properly you will know that the data you are about to insert is now stale and you can implement code around this in how you want to handle the exception that occurs.

[wet breach]

no it most definitely is not

[lost matrix]

ola, you are making the argument that simply crashing the server is a proper strategy to handle connection problems...

[wet breach]

mysql can handle millions of queries a second easily with not a whole lot of resources being needed for it

[rough hinge]

Now let's go to the real context that a query will never take 5 minutes, will milliseconds on main thread affect?

[eternal oxide]

ANY delay on teh main thread is detrimental

[rough hinge]

and if the player can do what I said in 5 minutes, in milliseconds. any player is subject to having rollback at any time Do you think this is good?

[pseudo hazel]

you should always program for the exception, not the rule

[wet breach]

you can never guarantee a query will always be quick

[pseudo hazel]

no but it cant be prevented

[tardy delta]

someone shaking with the harddrives hehe

[wet breach]

if the connection from server to DB dies for whatever reason, your query will hang for 30 seconds since that is the TCP timeout

[lost matrix]

Yes. Even 1ms is already 2% of the servers tick. And DB queries can take several hundred millis, which could be several ticks. And if a lot of players query something in parallel, then you are
running into massive, unrecoverable lags. Each millisecond counts

Thats why i dont even query redis (which is right in memory) sync.

[rough hinge]

So the solution would be to have rollback at any time?

[pseudo hazel]

roll back what

[rough hinge]

I had 100 players with my db sync, do you want me to send you the tps? 20.0

Really weird

[wet breach]

with how many queries a second?

[rough hinge]

Imagine, if I use .thenAccept or something like that, let's consider that it takes 5 minutes for the response to be returned. A get is used that returns the player's coins and it takes 5 minutes to return the response that will return 1 million coins. However, the player has 0 coins and sells an item in the store for 100 million. The player will receive the 100 million in the cache and will set it in the database, then after 5 minutes the database will overwrite the cache and the player will only have 1 million.
So in most cases I'm forced to use .join, which makes it completely useless and just wastes more resources.

[rough hinge]

a moment

[wet breach]

a moment?

[lost matrix]

Completely irrelevant. Dont try to come with anecdotal evidence.

[wet breach]

that is not a number

[eternal oxide]

ola you are arguing pointlessly. Not a single person here will agree with you doing sync db access on the main game thread.

[rough hinge]

https://imgur.com/a/c0gzFDW

[wet breach]

that doesn't answer what I asked. How many queries a second?

[rough hinge]

the whole server is in my plugin. and it's everything, insertions/reading when entering, whenever they modify a backpack, whenever they type /gang create, EVERYTHING

[pseudo hazel]

i think thats the wrong image

[rough hinge]

stay calm, I hadn't sent the message yet

[blazing ocean]

ok?

[rough hinge]

Basically it's a rust server (more similar to factions if you don't know) with a synchronized DB and all the data synchronized, anything the player does (setting up a base, evolving it, will be synchronized)

[blazing ocean]

showing a list of packages helps with nothing lmfao

[eternal oxide]

also TPS is a bad measurement

[blazing ocean]

https://cdn.discordapp.com/attachments/957661529447202906/1268207751952076942/ELX21fp7.standard-1.gif

[lost matrix]

I can assure you that we have quite a few devs on here who worked on actually scaled servers. And i can tell you from my personal experience (if we already throw anecdotes in) that you would
be laughed out of any dev interview, if you even so much as mentioned running database queries on the main thread.

[rough hinge]

but yes, the solution will be to have data inconsistency with the DB, and all players will lose their things at any time.

[pseudo hazel]

when did we say that

[blazing ocean]

what is bro making up

[eternal oxide]

done correctly (async) you will never have data inconsistencies or data loss

[wet breach]

only way I can see them sucessfully running DB queries on main thread with little issue is most likely because the amount of queries per second, is like 1 every so many seconds.

[pseudo hazel]

the solution is certainly not to crash your server on a blue monday because the database wasnt feeling it

[rough hinge]

Just 1 player enters and there is already a query

[wet breach]

this is why I was curious with their statement, how many queries were they performing

[lost matrix]

No, there will be no data inconsistency in your Database. You simply dont understand how to work with data. Thats literally it.
Get you head out of your a** and learn from what we provided you. Show some humility. You will not improve with this attitude.

[blazing ocean]

db host slow: server just dies

[void sigil]

Id not take that

@rough hinge Come back

[wet breach]

so in actuality you have no idea let alone don't even know how measure it or where to find this information

[rough hinge]

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

[wet breach]

again proving my point you don't know anything about DB's

[eternal oxide]

that there is a design flaw

[rough hinge]

So tell me how you would solve the problem I mentioned with asynchronous DB

[wet breach]

I already told you that you can lock tables

[eternal oxide]

get the data in the Async pre join so it is available in the Join

[blazing ocean]

ever heard of AsyncPlayerPreJoinEvent or whatever it's called

[eternal oxide]

you also delay the pre join until the data has been fetched so you 100% guarantee it is ready

[lost matrix]

No need to lock. For currencies you simply run queries like INC and DEC, with field constraints. You can define a DB number collumn to always stay above 0.
Then you run your DEC query, and it will either be successful or not.

[wet breach]

You can do it that way too

[lost matrix]

100% consistency in your values as the DB ensures consecutive calls

[wet breach]

^

this is one of many ways to do it

[blazing ocean]

load data in your pre join and then put it in a concurrent hash map and then do your shit on join

[lost matrix]

So there you go. You got two solutions. Locking and field constraints.

[wet breach]

anyways, @rough hinge I recommend doing more research or learning in regards to DB's. They are very powerful tools on their own and can facilate many things for you if you just knew what it had and in some cases remove code you didn't even need.

concurrency with DB's dates probably before you were born(guessing you are young from your example arguments)

[onyx fjord]

yall 👉 👈

[wet breach]

structuregrowevent ?

oh its stage

[onyx fjord]

i tried, when i cancel it the stage still changes

just it never finishes growing

[wet breach]

lol

[wet breach]

are you able to get the stage it was at?

because you could just reset the stage back

[onyx fjord]

so there isnt a specific event that handles that>?

like BlockGrowEvent for plants

[wet breach]

only one I know of is structuregrowevent

[onyx fjord]

yea im cancelling it and stage changes

[wet breach]

so reset the stage back of the sapling

all you really need the event for is just so you know the stage changed

[onyx fjord]

but doesnt structuregrowevent only trigger when last stage turns into a structure?

not each stage

[wintry dagger]

im gonna run every single database call on main thread

or else inconsistencies

[pseudo hazel]

lol

dont

[wintry dagger]

Ofc not im trolling

I had a good laugh reading the above thread

[pseudo hazel]

xS

[wintry dagger]

ur funny @west shoal

wait wrong tag

@lost matrix

[echo basalt]

I'd suggest informed is the right word there

[hidden spade]

Does anyone here know how exactly the custom crop plugins work, from what ive gathered they use blockentities but from what i can see i wouldnt be able to display the crop properly

[wintry dagger]

Can transient be used like this private transient final JavaPlugin plugin in classes with multiple instances to free up ram/memory usage?

if i need to reference the main class a lot

when dependency injecting

[eternal oxide]

no, ther is only ever one instance of JavaPlugin

[echo basalt]

transient is about serialization not about GC

[pseudo hazel]

yeah it would store references to the plugin class object, not the actual object

[wintry dagger]

ah so like if I have multiple Profiles stored in a List and every Profile has a JavaPlugin field, transient isn’t anything to add

[echo basalt]

It is if you're yeeting profile POJO's in the database

(because you can't really encode JavaPlugins)

But otherwise no

[wintry dagger]

yeah well do you ever store such an object in whole as it is

[echo basalt]

I've done it in the past yeah

And that's the whole purpose of transient

[wintry dagger]

Got it

it’s not a too common keyword eh?

dont think i’ve ever used it in plugin or even seen it for that matter

[echo basalt]

In the minecraft context, no

[wintry dagger]

im still stuck on my generics problem

[echo basalt]

You need to convert your tickettype<?> to TicketType<T>

casting is fine

[rough hinge]

For methods like this where findone is called if there is nothing in the cache, would you make it async as well? https://paste.md-5.net/miwaheqake.cs

[wintry dagger]

Either that yeah or making my T data parameter just an Object?

In the addInputData method

Acc no that wouldn’t work

[wintry dagger]

But don’t know if the ticket type generic is T since it gets from the name of the ticket type, not the validator class

I could cast it to TicketType<T> after first if check?

[echo basalt]

Yes

[rough hinge]

For example, the player types /money, would you run async that returns a completablefuture and then you would use thenaccept? And how would you send the message to the player with the amount of money inside the thenaccept? Would you run a task to just send a message?

[echo basalt]

You can just sendMessage in the thenAccept

[rough hinge]

sendmessage is sync?

[echo basalt]

The way we do it at work (we're a pretty big bedrock studio) is that everything that needs to be done in order is blocking

Our economy service is blocking and every request goes through the database

[echo basalt]

this tells me you haven't got a single clue about how minecraft works

[rough hinge]

yes

[echo basalt]

sigh this will take a while to get across

[rough hinge]

I didn't understand. Did you say to sync this?

[echo basalt]

There are a few keywords that are tossed around and they can often get confused

Mostly we need to know the difference between sync vs async vs multi-threaded vs thread-safe vs blocking

because labelling everything "sync" is just a huge meme

[rough hinge]

sync - be executed in a single thread
multi-threaded - many threads to execute
thread-safe something that may have coherence issues if executed outside the thread
async - could it just be load distribution and not imply a different thread?

[echo basalt]

Nope

A blocking method is a method that takes a while to execute regardless of what thread you're on

For example, an IO operation

[rough hinge]

.join on CF?

[echo basalt]

Not exactly

[rough hinge]

then

[echo basalt]

and stop jumping to conclusions so fast ffs

you're here to learn not to 1-up me

Thread-safe means that a method is "alright" to be ran on another thread, or multiple threads at once

So for example, sending a packet is thread-safe. You can queue up multiple packets on multiple threads at once, and the server doesn't care

Then you have the term "sync", which doesn't necessarily mean "main thread" but it means that a method runs in a synchronized operation to the thread that calls it

It might run on another thread but eventually join back the main thread

In minecraft terms we just refer to it as "running on the main thread"

And "async" doesn't necessarily mean "running off the main thread", it just means that "it runs.. eventually.. somewhere"

It's parallel to what your operation is doing

[mellow edge]

is it a common practice if I make my plugin use reflection and compile it with the java version that is run by lowest supported minecraft version my plugin uses. Then I can make a single jar that is supporting versions 1.8-1.21

[quiet ice]

I mean, you can also use MRJ \s

[quiet ice]

Yeah, the definition of async is often most misunderstood

Which is a shame, because it undermines the beauty of lockless concurrent datastructures and algorithms

[echo basalt]

(it basically just means don't expect immediate results)

Now, what we do at work is that we have a "blocking" bunch of services for our database operations

So for example economy is blocking, where any get / set might run through a cache but usually takes a while to execute

[mellow edge]

me?

[echo basalt]

So you're encouraged to execute transactions off the main thread (because blocking it is stupid)

[quiet ice]

yeah

[echo basalt]

And if you care about constant updates (let's say a scoreboard or something) you can store a snapshot of the last let's say 5 seconds

[mellow edge]

tbh I never heard for that

[echo basalt]

And update the snapshot if any payment data is received on the same server within those 5 seconds

That's one example, there are a lot more complex structures

[quiet ice]

Multi-release-jars. They are more often more pain than worth it, but sometimes can be quite neato - especially once you dip your toes in sub-J8 compatibility

[mellow edge]

I didn't know that even exists so I tought I could write my own tool that does that. But then I remembered that I can make a single jar for every version. But the problem comes that basically I need to refactor every existing nms code to reflection and then add new reflection for newer versions and etc....

[rough hinge]

I'm still thinking. That's what I said right

[rough hinge]

hum ok

[quiet ice]

I mean, you're probably better off using MethodHandles instead of Reflection unless you want to support really old Java versions.
And ideally you find workarounds that don't involve NMS code but I can understand that that isn't always possible

[rough hinge]

Yes, I already understood. But I don't even know if sending a message is a packet sent or not.

[echo basalt]

Sending a message is just sending a packet

Same for spawning particles

[rough hinge]

I wanted to know how in my case I would use ./money <...> and use the findone method being async (async on this case, outside the minecraft thread) in the db

[rough hinge]

I would like to know how, after the response returned from the DB, I can send a message outside the Minecraft thread and from what I understand, can I do that?

[echo basalt]

You can yeah

[rough hinge]

I would like to know everything about Minecraft that is thread safe. Sending messages, particles in the world is threadsafe. updating an armorstand for example is not thread safe, right?

[mellow edge]

I understand and good points. But the plugin's base is already in 1.8 and it was meant to be like that. But I want the plugin to work in modern versions, like the latest ones. And my plugin is forced into using NMS on multiple parts, because on some things I didn't use APIs because there is no guarantee that they will work later and upgrading that would be harder than upgrading NMS.

[quiet ice]

upgrading that would be harder than upgrading NMS.
Uh what

[mellow edge]

like third party APIs***, for example book api or smth

because as u know in 1.8.8 book support isn't as good as in 1.21

[rough hinge]

I have a vehicle plugin that whenever a player presses W in the packetreceiveevent (I used packetevents as the lib) updates the location of an armorstand. But, I update the location of the armorstand outside the game thread, which I believe is not thread-safe, what can I do about this? I tried to execute a task but the server thread stopped because there must be like 1000 threads in a single second. The thread stopped with only one player, which in this case was me using the vehicle.

[mellow edge]

so I tought I could maybe get bukkit package name and extract NMS version from it automatically and with that make it support all versions

[eternal oxide]

jump back sync to apply world changes

[thin frost]

Is it possible to block coal from being consumed in a furnace using spigot java

[eternal oxide]

https://hub.spigotmc.org/javadocs/spigot/org/bukkit/event/inventory/FurnaceBurnEvent.html

oh you mean stop it being used, not consumed

[torn shuttle]

how do I get an enity's body's yaw?

not eye

[eternal oxide]

nms

[torn shuttle]

I thought just getting the entity location and getting the yaw would do it but seemingly no

[torn shuttle]

no way really

[eternal oxide]

API is lacking in that it applies the yaw to the body and adds Eye height for eye location

[mellow edge]

I think I need just 1 single NMS rewrite from 1.8. 1.13 was released together with mojang mappings, so this is the only version that I need to refactor NMS and then 1.13+ versions will all work? correct me if I am wrong. All NMS I am currently using is for packets.

[eternal oxide]

it would be a breaking change to fix it

[torn shuttle]

can't get get an entity.getBodyLoc

[eternal oxide]

from nms you can

[torn shuttle]

or at least an entity.getBodyYaw

damnation

[torn shuttle]

any idea what the remapped way of getting it is

[eternal oxide]

I'll have to open my ide

[torn shuttle]

is it just getBukkitYaw?

[mellow edge]

I think I can confirm this one for myself.

[eternal oxide]

this.setYRot(yaw);
this.setYHeadRot(yaw);```

so read those for yaw

[torn shuttle]

well getBukkitYaw returns yRot

so I think I got it?

[mellow edge]

I can confirm.

[eternal oxide]

yRot0 is also yaw

its what setYRot sets

[rough hinge]

can i use player#sendMessage if player is offline but player != null?

[rough hinge]

reason: https://paste.md-5.net/toyadaseri.cs

[young knoll]

Probably?

[drowsy helm]

I think you can but why

[young knoll]

It just won’t do anything

[pseudo hazel]

I guess

[mellow edge]

nothing happens but it works

[drowsy helm]

Just chexk

[rough hinge]

see md5

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

[eternal oxide]

You can but its a stale object so will do nothing

[drowsy helm]

Yeah just put an if statement

[eternal oxide]

you May get a connection error in console but it should not crash the server

[rough hinge]

It will happen, I just want to know which one has more performance, check if the player is online or simply send the message

[pseudo hazel]

sending the message probably

[rough hinge]

send a message is equals send a packet

[pseudo hazel]

right

[rough hinge]

player.isonline check if player is a map or list idk

[pseudo hazel]

but the player object should already know if its online

[eternal oxide]

no

[pseudo hazel]

its in the Player

boolean

[rough hinge]

the code inside thenaccept can be executed after the player has exited

[mellow edge]

I always just do player!=null

[young knoll]

Why is performance a concern here

The difference is tiny

[rough hinge]

but player is != null lol

[rough hinge]

the code inside thenaccept can be executed after the player has exited

[pseudo hazel]

yes

but then at that point it will either be online or not

like the player object will know if its online

or atleast should

[young knoll]

Plus this is in an async task, so it matters even less

[rough hinge]

player#isOnline check in a list, it's not a simple boolean

[pseudo hazel]

even then

many things will be faster than sending a packet

[rough hinge]

What would you do? Would you do the verification or just send a message?

[mellow edge]

look, if the player object is valid, you can do whatever you want, or you can just store UUIDs and get player by uuid

[pseudo hazel]

but idk the implementation so sendMessage might not even send if no player is online

[rough hinge]

yes but 99% of the cases the check in the list will be redundant because thenaccept is supposed to return 1 ms after the player is still online

[rough hinge]

brother, you are still talking about worse alternatives 😂

[young knoll]

Just do the check

Then you don’t have to worry about something changing in the future

[eternal oxide]

Better. STOP holding Player references and fetch from UUID

[mellow edge]

exactly

[pseudo hazel]

your plugin is not gonna break because you missed a 1 ns increase optimization

[rough hinge]

In 99% of cases the check will be redundant, so I think it has more performance in 1% of cases to send the message to a player who is not even online, but I don't know if it returns an error.

[torn shuttle]

well seems like yrot is body and yrot0 might be head

does that mean it's not even exposed?

[young knoll]

Why do you care about performance

[rough hinge]

it depends on the case lol

[young knoll]

This is a single Boolean check that isn’t even on the main thread

[torn shuttle]

oh wait

it's public

I'm dumb

[pseudo hazel]

if you already made up your mind then just go for it, maybe add a comment or whatever

the idea behind programming is that you are aware of the consequences

[rough hinge]

If the player types /money pay <name> you will call a method with his uuid, then to get the name to get from the database you will get the name by uuid (go through the list once) and then in thenaccept check again if the player is online (go through the list once more)

[paper viper]

If you want to use Player you can use weak references

[eternal oxide]

data shoudl be stored against teh UUID as names can be changed

[mellow edge]

^^

[pseudo hazel]

generally yes, but this is such a non issue in this instance

[mellow edge]

always prefer uuids for any real systems

[rough hinge]

not in my case because my server is not premium

[mellow edge]

what?

[rough hinge]

So like I said, don't generalize

[pseudo hazel]

well that sounds like your issue xD

I usually develop for premium servers

[rough hinge]

well, I'll check if the player is online because it should return an error

[young knoll]

https://offlinemo.de

[mellow edge]

lol

[torn shuttle]

hey does the client do some kind of client-side logic for body rotations because this only "works"

it works when they start moving but now when they're standing still