#❗-1-18-experimental
1 messages · Page 16 of 1
@feral spire i assume this is u. what u mean by subject to change
It depend on if leaf is gonna suck it up or just use Mojang’s name on no-tick
leaf has been sleeping for a very long time now
i hope he does. i dont like this min 5
is he ok
Thsts the main reason why I didn’t update anything yet
Busy designing CPU
another issue i got is the despawn ranges
if my desapwn is 128 = 8 chunks
my sim dis is 4/5
It’s fine
Read the bumblebee story for potential side effect
But if you want to keep the drop down mob farm still functional
Bumblebee story?
i got a plugin that helps deal with it
mob cap being filled up from non ticking entities in loaded chunks
that effect right?
@feral spire u mention about the ticks per. are you saying that its better for the server to fill those caps so its not constantly checking? so should i put everything back to 1?
To me yes, but people have their own opinions
You seem to know enough so you can test it and come to your own conclusion 
yh i mean i drop some of my settings low and got plugins to adjust it to make it more vanilla like. like for villagers resetting their iron golem countdown when outside of activation range
altho it broke upon 1.18 and i cba to redo all the nms of it
another question is what could cause an idle server to use 150% cpu, i got another server almost identical using 8%
ping in reply

Per player mob spawning is not working for me on 1.18
turn it on
Anyone noticed any strange feature placement errors causing chunks to not generate on the latest paper version?
i dont think so. can you reproduce without plugins?
It happened when I tp'd to the area not running any generation plugins, happens with Chunky and worldborder's generation
your profile picture is off centered. circle doesnt match circle
Oh is it?
is that.... is... discord light mode!!!
yea deal with it
i guess you're lagging, then?
Hey guys, i have sometimes problem. That the chunks are not loading.
Maybe? ill pull the logs on it brb
1.18 chunk loading is slower, world gen is more demanding and aalso theres like 2x the height
Mhh, idk for 2 houres the day. when i write in chat, in my game i dont see the feedback, but in my console i see it.
After 2 minutes i get kicked
"losing connection"
does it say server overloaded?
sadly not
or maybe something related to connection reset, peer closed connection, keepalive, something network? => your connection is trash
idk i have a ping to frankfurt from 5 ms and a 100k download and 25k upload
i dont think its because my internet ^^
Seems I'm getting overloaded with a large structure generating, the thread dump has alot of "structure starts"
100k means 100 mbit
ye
12,5 mb
1.18.1 is supposedly coming out december 10th yea?
that's what it shows on the freeze logs? it might just be what you got stuck on when it finally gave up, not sure if structures are the origin of your issues. give more info, like show timings and stuff
I don’t think there’s an actual date yet
oh ok i had never heard of such a thing
That’s because that isn’t true. 100k means 100 kbit
Not entirely sure, but thats what it seems im scrolling through the thread dump from the overload errors.
i was to sleepy to write 100000 download ^^
The bugs above are fixed for the pre-release version, which is currently only available on PC, but is destined for a live release "at the end of next week" — the 10th of December. - mojang
can you fucking use units in your numbers. i don't understand you. what is 100000 download. how is 100k download equal to 100 mbit/s. how is that related to slow server world generation. i can't help you
i had just heard it somewhere, but apparently there's plenty of articles on it
AFAIK Mojang hasn’t released a date yet
there's no official date thougfh
So guess we’ll see on the 10th
well that would be why i said supposedly december 10th
lol
im more surprised they upheld their nov 30th deadline
wish they'd push it sooner, i would love to have that cylindrical fog
same
such a great feature, goes great with the new world
not even thinking about updating to 1.18 until they release 1.18.1
because spherical fog is atrocious
You can use sodium extra on fabric to remove it entirely btw
Been doing it since forever
same but they moved it consideraly farther away and now it doesnt show up vertically, might actually look good now
most players on my server use optifine or vanilla instances lol
so i test in those clients
although i swear there was a native no fog feature anyway
Fabric is better. Tell them
did they remove it for 1.18?
Yes optifine has it too
optifine is so hostile, wish people would move to sodium :(
ah i didn't see it then.
Maybe it’s removed but I no longer use optifine
Especially with iris now, sodium is the way to go
So I don’t know
honestly in my experience i've gotten 10-15 more fps on optifine than sodium
not a crazy amount, but in more laggy areas it means a lot
@normal pecan I made this for my community but if you want you can share it. This is a switching guide for monke. https://eternity.community/index.php/minecraft-client-optimization/
thanks!
you're missing the other sodium-like mods such as phosphor, lithium, hydrogen etc
nah had em all
vs just optifine
that's a bit sad but oh well
I noticed a similar fps difference when sodium was new, but it has become more consistent than optifine for me over the course of it's updates
I noticed improvements on my low-end device from eternity's guide vs optifine, but like integrated graphics go brr
this curseforge modpack has all of the performance and beauty mods https://www.curseforge.com/minecraft/modpacks/fabulously-optimized
(how come discord didn't show up those huge previews for that link? also i dont really trust modpacks lmao use at your own discretion)
My community has a lot of people with weak PCs and they see a 60 fps to 300 fps increase
It’s crazy
I mean I cap at 60 since refresh rate
also the one thing I miss on fabric is optifine's always-on FPS display
sodium extra has that i think
(also I have a laptop)
It’s on sodium extra yes
Error Im getting is a Chunk not there when requested error
Ill slow down the generation process and force enable it see if that changes anything and report back
It also might be at my border
Yep that's what it is
has to be something to do with the padding, thanks for the help guys 🙂
im having a hard time finding java 17 for linux server
read this ^
ty
indeed exactly what i was looking for tyty
readthedocs.io <<<---- point taken lol
distro?
They found it already
okey
Has anyone been experiencing disconnection issues.
Logs only show that the player disconnected, which is frustrating.
Usually 1 player gets kicked then 10-15 seconds later the rest do as well.
I've got some plugins, and datapacks so there is multiple potential causes to the issue but wondering if others are setting it.
The disconnect message ranges from
- Connection Reset
- Incorrect Header Check
- Packet Sized to Large
That's all I've seen so far.
Yep, i believe it's a vanilla issue fixed in .1
Ahh
As long as its not me tbh
Cause trying to figure out the cause would have made me go even more bald
Is it proxmox? Or Opnsense?!
im actually really still having a hard time and just noticed im on Ubuntu 20 on that system java version indeed returns 17 but when running the jar im still getting the error that i need a min of 17
ive tried the generic method and the Ubuntu method i think im going to have to find 16 and remove it
i had to compile 16 from source for 17 ... who knows where is actually is 🙂
This is awesome, thank you
Do you guys think I’d be able to run simple things like Essentials and Optifine on the experimental branch
have you tried update-java-alternatives?
or maybe not that exact command
no i have not i will try that next
something something java alternatives
my servers are on attached raid im almost wondering if it would be easier to just wipe the os
i doubt it
https://www.megajason.com/2021/06/24/how-to-use-a-different-java-version-on-ubuntu/ this seems like a good article
``` this might actually work
search it first
End cities aren't loading
running essentialsx just fine on paper 32
was also working just fine with paper 10
oh god it's 48 already
how are you running the server? start script? ptero?
Alright I’ll do some self tests then today on my server I’ve been running with friends, just wanted to know if other people weren’t having too many issues
I also realized after I sent that message that Optifine was a mod lol
i don't use many essentials features but the ones i need are there
i dont think stuff should be broken
also works on debian. basically you can install as many java versions as you want. alternatives lets you set the default, but if you run multiple things, just use alternatives to see the path of the java binary, and use that specific binary needed to run your jar.
if you need jdk17 on debian buster, I was able to use the amazon corretto build by adding them to my apt sources.
a similar system exists on every linux distro/derivative, i came from fedora and googled a similar system in debian based ones
so not just debian
yea, just referring to the specific instructions.
I'm using paper 1.18 and end cities aren't loading, I deleted the end world file to reset the end, after that we're unable to find an end city.
well... thats the end then.... huehuehue
@hot zenith have you tried using the /locate command to see if anything shows up that way
^
i had to do some messing about with the directories but that link got me going ty
good to hear have fun
i was pretty convinced i should just take the 20 min to wipe:)
pretty sure ubuntu would contain a version of openjdk ootb and it won't be latest
me waiting for paper 1.18 to come out
i actually find using multiple java versions easier on *nix than my windows box, heh.
well, i do it all the time as a java dev by trade 😛
i had to try a few but this is the one that worked for me on 20 openjdk version "17.0.1" 2021-10-19 LTS OpenJDK Runtime Environment Corretto-17.0.1.12.1 (build 17.0.1+12-LTS) OpenJDK 64-Bit Server VM Corretto-17.0.1.12.1 (build 17.0.1+12-LTS, mixed mode,
depends on use-case. if you are running a jar, you can do same as nix, and use exec path for java.exe
but if the application is a wrapped jar inside an exe, you have to specify java via a config file usually. my eclipse IDE is a pain in the ass in this regard.
i see
or change the path.... but gross.
i have a really hard time compiling a jar in linux
i actually specifically leave java out of my path so i dont use the wrong version by mistake.
I just have java home in my path
And then change java home if needed
Using a script that's also in path
to be clear: as someone who works in java every day for my day job I hate it with a passion, but I love the money! 😛
So if I wanna use an old java version in a terminal, I just enter java11 into terminal and be done
thanks everyone im going to go check out paper 18 i appreciate you all helping me get patched up
Also, as a developer you should just have latest installed, you can target older java versions with that
IF only it actually worked like that.
It works like that for me
don't try to use removed and deprecated functionality then 😛
legacy code maintenance is a bitch in the enterprise.
You can literally just set your source and target language level per project
I work for an insurance, don't try to teach me about legacy code, lol
there is functionality literally removed that doesn't exist in java 11+, for GOOD reason I might add.
hah, I work in insurance too 😛
I interact with code that was written before generics were a thing
something about Java and insurance companies 😛
insurance besties
Java is just the most popular language in enterprise
tis true
(for many good reasons)
rip solaris
I always see conversations about Java stuffs breaking compatibility in enterprise. What horrifying things are they doing in their code that breaks so easily?
usually its just bad code.
It's fud mostly
or code that hasnt been updated since the dawn of time.
I managed to upgrade everybody here to java 11 and most of the pain points where people using old gwt stuff and needing to add some libs to maven that have been removed from the JDK
java makes it pretty easy (but not as easy as javascript) to do really dumb things (as a developer).
I think we at 95% java 11 now
The last ones are code by 3rd parties
Like we still in process of updating our CRM
(I am on the java governance board and can push for stuff like forcing people to update their shit to modern java versions, my apps where the first running java 11 in prod)
once i finally got our old Base64-using code converted off the deprecated (and removed in 1.11) calls, i was able to get everyone up to 11 last month.
you have a governance board? haha, nice.
pretty sure they would have to form a committee to decide if we should have a board around my shop 😛
Idk how old that is, my boss was on there, was out of office one day and asked me go instead and everybody agreed I should stay for all future meetings ^^
But yeah, we are ~650 people in IT so gotta have some kind of place where we can discuss stuff like that
We force everybody to at least update deps and stuff 4 times a year
oh no
our company culture is garbage in IT. basically 100 different silos that don't communicate
"platform updates"
turf wars abound
Oh company culture we always were great, this is cologne after all so most of the stuff you manage by just walking over to a coworker and getting it done instead of going thru some proper process and shit (Compliance hates that but everybody hates them so who cares ^^)
I'm ready for Java 17 and the project as well. Now we downgraded to Java 11 because the customer uses redhat images and there is no LTS image with Java 17 support. 
The plugin miniblocks stopped working on the latest version paper 1.18, previous versions 1.18 it worked. https://www.spigotmc.org/resources/miniblocks.85569/
Makes sense. I'm in the good 'ol MURICA. Middle management cares about one thing: profit-loss ratio 😛
cool thanks for sharing
I guess you will try to enforce java 17 now ?
IT is separated from the rest of the company, I don't give a shit about profits, not my thing to worry about
I feel this in my bones.
Pipeline team agreed to commit to supporting a new LTS version in Jenkins and the rest of the pipeline tooling within 6 months
They super busy with a bunch of stuff sadly
Hah - our shit so fucked up. Every division has their own IT teams for software stuff. The "corp" it is just for desktop/server and networking/security.
I mean that's cool they all agreed on this :)
Did you try grabbing the snapshot version of NBT-API that is meant for 1.18?
But yea, in review, most people who complain about java (like me) - it's really not java at fault. just bad code/environments/culture/etc
All of which can be fixed
I heard that check24 has such a concept. 
Be the change you wanna see in the world
- starts punching windows 10 users *
Not only them, a friend of mine works at Otto and they also let every IT team do whatever, I extremely disagree with that philosophy
"in theory" 🙂 - Oh and since we are Silo'd, I do tend to run my little silo with an iron fist. I ruffle feathers of the potatos that surround me, but we are probably one of the better among the company. thus actually migrating to java11 from 1.7 (yea.... lol) this year. still not perfect, but meh. like i said, money is money.
If you want them to move to Linux, you gotta make Linux actually usable for normal users, see LTT vids, Linux Desktop is still a clusterfuck
Yes, I used 2.9.0 snapshot
tbh i agree with you, i think its easier to patch invasive windows 10 features than to use linux
Been watching this, and as someone who uses debian desktop about 30% of the time - they are hitting the nails on the head with this "daily driver linux" series of vids.
Higher costs but improved individuality. A new proxy account doesn't take 2 sprints then. It will be done on the short way.
I actively try to not being a head monopoly, that's just a painful situation to be in, having to worry about shit hitting the fan if am taking vacation for example, or constantly being annoyed by people about a project and not able to delegate to focus on something else
good thing i completely gave up with linux and moved back to windows 
at this point I am most surprised someone like RedHat hasn't tried to just monetize an REALLY REALLY good linux desktop distro or something.
I mean there's fedora
Also redhat is part of the bad guys now
What they did to open shift is criminal, they can burn in hell
what happened
Insert obligatory fuck IBM here
Oh for sure. Like I said, our stuff is a shitshow - but at the same time, I turn off the "care" button when I log out for the day. I am 100% mercenary for now. With kids and such, it's easier just to take the paycheck and enjoy the relative flexibility. I can worry about job satisfaction when they are grown. Until then, working remote and not worrying about commuting to say, seattle, is a huge boon.
There is also a certain advantage that comes to being the smartest person in the room (although its bad for personal growth, so I have to get that elsewhere).
My friend who works @ redhat here in the states kinda hates reality these days since the IBM buyout.
I solved that by hiring a new college and having her sit in my office, I have never seen anybody deep dive into my shit code and actually understand it like she does
She's like me, but since she's 10 years older she's a bit wiser, does a bit more due diligence and stuff and it's rubbing off on me, lol
Heck, I even write unit tests now, use feature branches and merge requests 😂
no need to brag now 😛
that sounds like sensible development practices. can't have that.
Just to know what's the proportion of Paper core members and maintainers using Windows(+WSL) and full linux ?
(If you know)
We don't keep spread sheets on that, lol
I guess xD
i hope most devs use real linux
This would also be the wrong channel to ask that in
mac, of course, being fake linux runs
as an RDU of paper, I prefer fake linux myself.
Cat is Mac, DenWav is some Linux distro, am window no WSL, Kenny is windows wslg, kashike uses Windows only for gaming, jmp leaf Mari all also use Linux for dev I think
I see :)
kick cat
anyone else's tps plummet to 9 in the latest experimental version
Send a timings report
sure 1s gonna see if it's maybe worldedit or worldguard
Aikar's Timings Viewer - View Timings v2 reports from Paper and Sponge
@peak pulsar
only thing that's changed is paper version and worldguard and worldedit
but worldguard and worldedit were removed and it still happens
I'll try downgrading paper version
This timing is not very accurate
yeah it was only a 3 minute one
And read this guide to Config your stuff if you are running weak hardwares
Too many entities/mobs? @timid wolf
shouldn't be
It was listed on the loaded chunks lagging in the timings report. Something around 1750 tick-entities.
Restarted server and this happened around where a player was. All the blocks are bugged and you cannot place anything in the chunks. You just violently shake up and down when you walk into it
my bad yall
any idea on a fix? apart from deleting the region file
Holy...uff
Restore from backup
Make a backup first
3 is normal @timid wolf 😆
yeah lol
I wanted my crops to grow fast for a second
it was like 3 hours ago when I did it
You know bone meal exist???
well since this is #❗-1-18-experimental im going to experiment and join 179 clients because im bored
strangely fixed by sending everyone out of render distance and returning
no region fuckery required
fog manipulation, possible since 1.18
dope
i think mojang fixed that loading issue in 1.18.1
chunk loading does get glitched when in low fog distance sadly
is this done by sending a fake render distance for the client?
yes
neat
there is something seriously wrong with Paper 1.18
ay can i get a link for the top right
seriously no lag compared to 1.16 and 1.17
ive not had any issues
Use fabric sodium + sodium extra
You can do it on client side without doing anything on server side
Is the 1.18 good to start a server on or should i wait a little bit longer?
Wait for stable if you can
it works fine enough if you accept that it might crash sometimes, keep many backups
personally i'd wait for 1.18.1 (preferably stable) builds
yeah thanks for the update!
if you're just setting up i suppose it should be fine
just wait to release till there are stable builds
have them run out and load chunks in spectator
Are you trying to stress test ur server or something?
im procrastinating on homework.
Ah, been there.
well, the moment these 159 players start running away the server lags, client goes to like <1 fps and almost all of them get kicked
so i guess paper 1.18 isn't ready for 160 players.
or your hardware isn't
Well do note, all of these are close together as well
This isn't a totally accurate representation of 160 actual players
You will need to move player far away and actually make them do stuff
Otherwise this test isn’t going to tell anything
i can make them jump, attack, mount stuff, follow other players and break things
realistically i can barely make them walk forward because the server lags too much
[18:57:11 INFO]: Could not spread 160 entities around 0.5, 0.5 (too many entities for space - try using spread of at most 0.00)
:(
make them fly lmao
A
go do your homework
well, spreading 160 players to new chunks crashes the server on a 12 core machine with 8 gb of ram.
😭
yeah you probably shouldnt do that
Told you. It doesn’t cost much if all player in one chunk
This also happens before on that worldQL thingy. And the demo is just a lot of Steve’s 
Huh, that looks like your JVM and/or OS config was bad
Or maybe you don't have enough RAM to generate 160 players worth of chunks at once 😄
do you have any suggestions on what i could do to properly fit 160 players?
- Show more of the logs
- Show more of the logs
Can't say for sure why it died from what you gave
Are you planning on having 160 players generating new chunks constantly?
right now i'm installing viaversion so i can run the bots on 1.16, that way the script i made doesn't chew through my cpu/ram
yes, im trying to get 160 bots to fly in a straight line using spectator mode to stress a server
i have like 20 of them running in circles, attacking each other and digging holes but that is working perfectly fine, since we're all standing in the same chunk. this is useless.
They’re not even bots, they’re packets
whatever kid, discord bots are just http requests with some if()s and you still call them bots
You can demonstrate the same thing in 20 minutes with protocollib and a server with a gig of ram

but like, real players do a lot more than just flying, they build stupidly big redstone machines, huge bases stacked with block entities, etc
so its only a subset of the server ur actually testing, chunk loading/generating
All my benchmarking back in the day was 20 NPCs randomly spread through the world just standing around being invincible (but not creative so mob AI would still consider them)
I was trying to optimize the base level of things, chunk lookups, block and entity ticking, etc
Mining stuff is whatever, redstone machines you can nerf/remove, and you can pregen worlds
how do i add this?
it's a bit more than that bro
Can you send the crash log?
You could give a look at https://github.com/Eoghanmc22/rust-mc-bot for stress testing
i made my own
How performant is it? In our case we do tests involving hundreds/thousands of connections
does the rust-mc-bot move around?
Yep, it is actually the only thing they do
build #50 of 1.18 paper 
granted like the last few have been a lot of reverts for changes but still
8 steps to complete the 1.18 to do
51 coming soon ™️

is it safe to use waterfall 1.18 rn?
yes
Take a look for velocity btw 
yes, you are on the paper section of the downloads page
click on "waterfall" at the top for waterfall downloads
400 builds omg 👁️
that's not just 1.18, the waterfall build number never resets
as there are never two versions maintained at the same time, it supports 1.8-1.18
Lol BTW there is no eta behinde eternity's name
wut
Things happen! ;p
- Spigot made changes 2) They caused an NPE 3) we test out a fix that doesn't work 4) we revert fix 5) we revert original spigot change
Things did indeed happen.
Whats NPE?
error
Just Bret Hart doing some code review.
Oh
How convenient.
why does this look like slack
Ok, good to know it's not just me
goats suck
paper lets you configure the tick rates of mob behaviors, so you can nerf that goat behavior and none of your players will likely care
kind of funny that goats are still laggy though, pretty sure I saw goat optimizations mentioned at least once or twice in the 1.18 snapshot changelogs somewhere along the way
lmao yeah pretty sure that's the exact behavior that's the problem too, so overkill
checking every possible pos sounds exactly like something mojang would do
tbh I'd wanna see a spark there
Yea I am gonna do a profile and see rn, actually
But it's iterating through each block given here and checking stuff
BlockPos.betweenClosed(i - this.maxLongJumpWidth, j - this.maxLongJumpHeight, k - this.maxLongJumpWidth, i + this.maxLongJumpWidth, j + this.maxLongJumpHeight, k + this.maxLongJumpWidth);
does it actually check every possible position before jumping? can't it just...check one random position, if not valid, check another, etc?
it should be able to fail fast checking the outside first yeah
This is the logic
Iterable<BlockPos> iterable = BlockPos.betweenClosed(i - this.maxLongJumpWidth, j - this.maxLongJumpHeight, k - this.maxLongJumpWidth, i + this.maxLongJumpWidth, j + this.maxLongJumpHeight, k + this.maxLongJumpWidth);
PathNavigation pathNavigation = mob.getNavigation();
for(BlockPos blockPos2 : iterable) {
double d = blockPos2.distSqr(blockPos);
if ((i != blockPos2.getX() || k != blockPos2.getZ()) && pathNavigation.isStableDestination(blockPos2) && mob.getPathfindingMalus(WalkNodeEvaluator.getBlockPathTypeStatic(mob.level, blockPos2.mutable())) == 0.0F) {
Optional<Vec3> optional = this.calculateOptimalJumpVector(mob, Vec3.atCenterOf(blockPos2));
optional.ifPresent((vel) -> {
this.jumpCandidates.add(new LongJumpToRandomPos.PossibleJump(new BlockPos(blockPos2), vel, Mth.ceil(d)));
});
}
}
or something like that
this seems like a no-brainer but like, mojang 
oh god yea ok there's no return
it checks every possible one
👏
🤯
making getJumpVector lazy might do it
i'm guessing it randomly grabs from jumpCandidates later?
yep
It does
is that uh, cached/stored around?
if (this.chosenJump.isPresent()) {
if (l - this.prepareJumpStart >= 40L) {
mob.setYRot(mob.yBodyRot);
mob.setDiscardFriction(true);
Vec3 vec3 = this.chosenJump.get().getJumpVector();
double d = vec3.length();
double e = d + mob.getJumpBoostPower();
mob.setDeltaMovement(vec3.scale(e / d));
mob.getBrain().setMemory(MemoryModuleType.LONG_JUMP_MID_JUMP, true);
serverLevel.playSound((Player)null, mob, this.getJumpSound.apply(mob), SoundSource.NEUTRAL, 1.0F, 1.0F);
}
it's a short lived behavior pretty sure
Yea, so i am not thinking that this is the expensive part here/
could easily been done by making that first loop lazy and choosing random positions instead of iterating in order
how so?
they check a random pos in the tick, and remove it it's not valid
The valid check is done in the start method, no?
yeah I'm looking at it closely, I'm not 100% sure whether that would change behavior or not
@Override
protected void tick(ServerLevel serverLevel, E mob, long l) {
if (this.chosenJump.isPresent()) {
if (l - this.prepareJumpStart >= 40L) {
mob.setYRot(mob.yBodyRot);
mob.setDiscardFriction(true);
Vec3 vec3 = this.chosenJump.get().getJumpVector();
double d = vec3.length();
double e = d + mob.getJumpBoostPower();
mob.setDeltaMovement(vec3.scale(e / d));
mob.getBrain().setMemory(MemoryModuleType.LONG_JUMP_MID_JUMP, true);
serverLevel.playSound((Player)null, mob, this.getJumpSound.apply(mob), SoundSource.NEUTRAL, 1.0F, 1.0F);
}
} else {
--this.findJumpTries;
Optional<LongJumpToRandomPos.PossibleJump> optional = WeightedRandom.getRandomItem(serverLevel.random, this.jumpCandidates);
if (optional.isPresent()) {
this.jumpCandidates.remove(optional.get());
mob.getBrain().setMemory(MemoryModuleType.LOOK_TARGET, new BlockPosTracker(optional.get().getJumpTarget()));
PathNavigation pathNavigation = mob.getNavigation();
Path path = pathNavigation.createPath(optional.get().getJumpTarget(), 0, 8);
if (path == null || !path.canReach()) {
this.chosenJump = optional;
this.prepareJumpStart = l;
}
}
}
}
``` this is tick
and this is start
@Override
protected void start(ServerLevel serverLevel, Mob mob, long l) {
this.chosenJump = Optional.empty();
this.findJumpTries = 20;
this.jumpCandidates.clear();
this.initialPosition = Optional.of(mob.position());
BlockPos blockPos = mob.blockPosition();
int i = blockPos.getX();
int j = blockPos.getY();
int k = blockPos.getZ();
Iterable<BlockPos> iterable = BlockPos.betweenClosed(i - this.maxLongJumpWidth, j - this.maxLongJumpHeight, k - this.maxLongJumpWidth, i + this.maxLongJumpWidth, j + this.maxLongJumpHeight, k + this.maxLongJumpWidth);
PathNavigation pathNavigation = mob.getNavigation();
for(BlockPos blockPos2 : iterable) {
double d = blockPos2.distSqr(blockPos);
if ((i != blockPos2.getX() || k != blockPos2.getZ()) && pathNavigation.isStableDestination(blockPos2) && mob.getPathfindingMalus(WalkNodeEvaluator.getBlockPathTypeStatic(mob.level, blockPos2.mutable())) == 0.0F) {
Optional<Vec3> optional = this.calculateOptimalJumpVector(mob, Vec3.atCenterOf(blockPos2));
optional.ifPresent((vel) -> {
this.jumpCandidates.add(new LongJumpToRandomPos.PossibleJump(new BlockPos(blockPos2), vel, Mth.ceil(d)));
});
}
}
}
(the start is where the canidates are caculated)
er, what's start?
you already shared the start snippet I think
That is the first time that the behavior is ran
well, the relevant part
mhm
oh, start is part of the behavior. tick is the entity tick
tick is the behavior tick, sorry
oh
basically on tick it polls from the locations, and if it's valid will try to jump to it
uh yea^
on the next tick
and it stops once it jumps
so yeah, I think not selecting positions and removing at the start would slightly change behavior
but not in a noticeable way
Optimized AI for Axolotls, Goats, Hoglins, Zoglins, Piglins, Piglin Brutes and Villagers.
1.18 changelog, clearly goats are quite optimized
making that method lazy would be easiest way I think
This was for attacking mobs basically
making a list a map, and reordering booleans to favor less expensive operations first
btw lithium has a optimization for the jump stuff apparently: https://github.com/CaffeineMC/lithium-fabric/commit/fcbf706c259d13edf378e825a344b697056ccc9f
oh that's interesting
I'm not convinced that's 100% no effect on behavior but I would bet you can't reasonably tell any difference
disclaimer have not looked super closely
err are you not able to run "createMojMapPaperclipjar" directly?
Oh, it's generated elsewhere
ic.
Yea so
it defo does not like doing all those collision checks x)
even after adding the collision patch, geez
🥴
alllll the doubles
there's gotta be a z later in the method
I see a y
oh that is the end

Is it maybe possible to narrow the range of angles to test based on the distance from the goat
like with rougher math than they use I mean
ie slightly less optimal angle
maybe even hardcode some known values 
or maybe precompute the angles for each position, relative to the goat?
Well it's called in this
Yea
shove that in a big static field
I'd think it would be interesting to see the differences in the above optimization tho, because yea this is quite a mess x)
🐐 goat go jump
Yea just idle, they take a stupid ton
looks like similar issue, statcked baby chicken grow up cause server no response, not sure vanilla bug or not.
That generally is a separate issue, this only applies to goats.
er, wat
thread stuck at net.minecraft.world.level.CollisionGetter.findFreePosition(CollisionGetter.java:110)

Can you show all logs?
I saw that method in CollisionGetter
and was like oh boy
this is called from mobs tick

it's new in 1.18, separate issue
is there an updated anti-xray config for 1.18 with deepslate ores and such?
Ah
I believe the gist is updated
Anti Xray configuration guide: https://gist.github.com/stonar96/ba18568bd91e5afd590e8038d14e245e
@ancient sun ^
default Optional<Vec3> findFreePosition(@Nullable Entity entity, VoxelShape shape, Vec3 target, double x, double y, double z) {
if (shape.isEmpty()) {
return Optional.empty();
} else {
AABB aABB = shape.bounds().inflate(x, y, z);
VoxelShape voxelShape = StreamSupport.stream(this.getBlockCollisions(entity, aABB).spliterator(), false).filter((voxelShapex) -> {
return this.getWorldBorder() == null || this.getWorldBorder().isWithinBounds(voxelShapex.bounds());
}).flatMap((voxelShapex) -> {
return voxelShapex.toAabbs().stream();
}).map((aABBx) -> {
return aABBx.inflate(x / 2.0D, y / 2.0D, z / 2.0D);
}).map(Shapes::create).reduce(Shapes.empty(), Shapes::or);
VoxelShape voxelShape2 = Shapes.join(shape, voxelShape, BooleanOp.ONLY_FIRST);
return voxelShape2.closestPointTo(target);
}
}
``` 
that covers stuff like netherite too right
Yay Minecraft!!
It should, look in the config.
yes
You know, if this wasn't Java that might actually be efficient
so with a reasonable expectations can i expect a playable experience
thread dump only
dump.txt by @burnt portal: https://paste.gg/56f6383dbc0c4f08abe876a148fcfe8a
How so?
Well, not algorithmically but yeah
yup
But since this is Java someone should see how much faster it is to turn that in to normal loops
Update, 42 was bork somewhere iirc
if my world is named world2, should the master key still be world-settings?
What part of the language here slows it down I’m curious?
it sucks now tbh, huge cpu spike when joining in post 1.12.2 versions, breaking doesnt work in 1.17+
the issue is their collision code
but the streams don't help
I'm gonna try and make that method use leafs collision code
and remove the stream
Yea Minecraft loves using streams in hot code
Their collision code isn't that bad, iirc
Doesn't it just iterate through the blocks in an inefficient order?
But from a higher level the algorithm it uses makes sense
and generally have 0 optimisation there
Like, the big O for it probably looks good
They're writing code like they're working on a spring app, basically
Seems like it
Literally a joke, have they looked at their game in a profiler? There are some obvious places that instead paper has to fix themselves? It’s kinda funny.
That’s true
Part of the issue is that they don't really test at scale
Like, most of what mojang cares about is SP and "can it realm"
Honestly I think at some point they stopped even testing with more than 1 person
When DB and Grum first joined they were at least sometimes doing like an office server or whatever to get some load
I keep saying that I wish that they'd have like an hour or so a week where they fire up a server, get a few dozen people on it and see how it performs
how would i go about pulling an experimental build from the api
same way you do with any other build
er, wat?
api doesnt discern stability types
At some point DB decided he'd rather focus on game design and Grum only writes in enterprise Java and Haskell pretending to be Java 😄
so just latest?
yep
There is nothing special about 1.18, just a pretty banner on the site
dont autoupdate 1.18 builds, it will break at some point
Fucking lenses
at what point is that point
(stupid question)
If we knew they wouldn't break
Yea, I mean, that's the joys with stuff like MC
who the uwu~ knows, it says experimental everywhere just asume its fragile
ight
I love writing code and shit, but, sometimes it's like, I wanna toy with other stuff
autoupdate is always a bad idea, just because they stop saying it's experimental doesn't mean they're telling the truth 😛
i just want to use an old plugin i wrote that used the adventure apis and im too lazy to port it to bukkit-platform
Like, especially with MC, theres so many areas where it feels like it would be trivial to substantially improve random shit here and there with little thought
It's always possible to have a build where an event doesn't work right so WG fails to block things or chests get wiped on chunk unload or mob spawning spews 10,000 creepers on a single block
That’s how I felt on a rather large project I used to work on
I would always go in wanting to do one thing, only to have done a bunch of different little things.
Like, in a month those kinds of builds could still happen
It’s why I’m also enjoying contributing to paper rn since there is just so much more to do.
Some testing happens for PRs but usually only focused on what the PR says its changing, not on whether the rest of the game still works
i firmly believe mojang doesn't care about small improvements. if we, random devs, take that path of "fix ur shit" we'll end up making stuff like optifine/sodium and that's just not a proper fix, the game is still broken
(I Hope it ain’t getting spammy 🥴)
Yea it’s hard, because there is so much.
I still try to work on shit with paper but long gone are the days when me and aikar would be tag teaming updates
What ever happened to aikar
busy with work and shit
You should never blindly update to a new paper build, always have backups
Makes sense
updates are much bigger now I'd assume
Think in part he just kinda burned out with all of the shit of mojang too
Even build #308 could be broken
Only because Paper has so many more patches
Well I’m glad ur still powering through it
Well, we've only gotten 3x the patch and more complexity as the game has gotten shitter
Never did I think that we'd be at the point where we pretty much needed to start rewriting entire systems
Uh, you were at that point before the project existed
What’s stopping them from just integrating some of papers major systems?
That's why I quit 😛
No, I helped trigger the license mess after getting burned out on maintaining a pile of junk on top of a pile of junk
lawyers
The license mess was the last straw though, not the reason I was done
We've already told mojang that they're free to treat our patches as MIT and we're happy to sign on that
But are they too afraid to deal with the legal side?
Ah, profuctor optics is what I was thinking of earlier
The lawyers don't wanna touch it, basically
A lot of what Paper does is what Grum would consider lipstick on a pig
and everything like that has to go through legal, basically
He wants to replace those systems, not patch them up as-is
Yea
Probably asked a thousand times, but is there a current time est on the full 1.18 release?
There is no eta
Even though as he replaces stuff it needs even more patching sometimes 😄
piss off
tbh like I just got done explaining to someone else there is no difference between these builds and any other paper builds
It’s a shame because in the end it’s just upkeeping for you guys
You should always have a backup, always be ready to roll back to and lose progress, and always test a build before you put it live
The only change is that builds that break your shit are somewhat more likely right now
what amaranth meant to say was you should auto-update, run java 8 and never keep backups. #yolo
it says "experimental" everywhere :( why cant people use common sense?
Then how does a hard fork intend on happening
does that mean just stop getting patches from spigot?
that means what we drop spigot as an upstream and forget they exist and be happier
Yeah, that means Paper updates everything on their own
That'd be nice
"nice"? that'd be excellent, outstanding, marvellous
And then as they rewrite the entire API they also reimplement everything as mixins and eventually merge with Sponge
what does spigot do that's so bad?
Don't give us ideas
Hard fork 100% means your plugins will break
That's the whole point
So, it's not all rainbows 😛
Spigot is a 1 man band where the lead basically generally gives 0 shit about the community and it's opinions, etc
i think everyone will stick to paper and drop the bukkit/spigot stuff lol
stagnant API
thats why it should wait until Hangar is ready so there is a place for paper plugins
also see: bungeecord
Spigot is literally only still relevant because paper is binary compatible with it
Only reason spigot is considering supporting custom biomes in the API is because they've got all of the shit ready to start bytecode hacking outdated plugins to still work
(mostly)
wasnt there a page with a % of how many servers are paper vs others?
Should I just calm down my ocd about this ? every time I update it (this is a fresh download so 52) it's NEVER the latest version:)
INFO This server is running Paper version git-Paper-"c837002" (MC: 1.18) (Implementing API version 1.18-R0.1-SNAPSHOT) (Git: c837002)
INFO You are 6 version(s) behind
INFO Download the new version at: https://papermc.io/downloads
INFO Previous version: git-Paper-"a42917d" (MC: 1.18)```
you are 6 commits behind not 6 versions. thats the latest build
iirc it's actually counting git commits or something
It’s honestly terrible lol
why do they continue then?
hobby project?
bStats
yeah bStats combines all versions
I suspect Spigot continues because spite and money
oh yea
what does it mean to make bytecode hacks to support custom plugins? why can't those just be patched in?
He must get all donations
The Bukkit project probably existed for an extra year due to not wanting to let Spigot "win"
Legit modifying plugins byte code
If methods and such break, spigot fixes them!
wa? the bytecode hacking modifies plugins as they load in order to try to mitigate the binary changes
is any of this in prod now?
Rather than just promoting an environment where "Hey, shit moves forward, we might deprecate stuff here and there, be expected to need to update stuff"
spigot's been doing bytecode manipulation on plugins since at least 1.13
Don't
It is usually used when stuff like Minecraft renames materials, like GRASS_PATH -> DIRT_PATH
ok so it will just not show "You are running the latest" until devs are done tweaking it I assume?
see: legacy (-1.12) plugins
where can I find the source for this bytecode manipulation, im curious
instead bukkit somewhat created and spigot cemented an environment where deprecating API is literally last resort
Renames the constants…. At runtime….
Spigot stash
It will show you're running the latest version when you are, in fact, running the latest version
yep. also the commits are going to keep happening and you will never be in the true latest version
This was just explained. It counts how many commits there's been since the jar you're running was built.
And then removing it 10 years later x)
Bukkit used to break the API all the time right up until the version we took over
So revolutionary!
What ever happened? 😩
Yea, I mean, bukkit was more "we shouldn't break shit for the sake of it"
Which, I salute
you can build all commits from git master if you want latest latest
spigot took that to mean, "We shouldn't break stuff, eeever"
But EvilSeph and Wolv wanted API stability and we figured it wouldn't our place to change shit just because we didn't like it since we didn't start the project
and thats where you get shit like the 1.13 update where md yeeted an entire API right on release more or less
Wolv was the first one do to bytecode fuckery to maintain compatibility
Isn't spigot and bukkit both maintained by spigot?
He just did it to the server
ye
@hexed flower @polar quarry
I'll just not type /version for a while.. nah I'm not THAT freak I'll let devs do it they're better than me :)
thanks for explaining :)
Bukkit as a project is dead
they don't own the bukkit/craftbukkit project in any capacity
Spigot has a thing they call Bukkit that is separate still for some reason
Wasn’t that all about the fun legal drama
No?
Im confused
bukkit called it quits before all of that drama
Bukkit as a project was going to be dead either way, tbh
Even without the legal drama we all quit and DB and Grum weren't going to maintain it longterm
They might have done a 1.8 release (I kind of doubt it though) but after that it would have died anyway
Interesting, yea I haven’t rly been here for long… at least for majority of the time I was a clueless mc player using the vanilla server.
It's not like md5 would give up control over his own platform and go just be a random dev working on a Curse and Mojang owned project where all he could do is git commits and forum posts
And the rest of the community that tried to pick up the pieces decided pretty much right away to do a radical new API instead of trying to make something Bukkit compatible or something familiar to Bukkit plugin devs
So where were you situated here? As a spigot maintainer or what here? I’m curious
I was, for most of the time between 1.3 and 1.8, the person doing most of the work on CraftBukkit
Hey I helped 😦
Ahh
🤣
That’s crazy y’all still stick around here!!
I wasn't alone and for a release I disappeared and other people took over
part of what I like about having a bigger team, lol
But I was probably doing about half of it
Like, it wasn't too long ago that updates where basically a tag team between me and whomever else was around, usually aikar and Z would chip in here and there
But, that was when we could go through all the patches in 2 days, with spigots mappings
Can confirm amaranth would spend unreasonable amounts of time working.
Yo electron, paper 1.18 is running on my server with 40 ppl on perfect
ok
I’m glad it’s working well for ya 😄
guys do you get donations
is the opencollective enough to pay for server rent and stuff?
I mean a lot goes to paper itself and to its maintainers individually
nice that's what i want to hear
The project one I don't think any goes to individuals
It's for hosting costs and otherwise just pools waiting for a use if there is any extra
Some of the devs are getting money directly to them via github sponsors or twitch streams or paypal or whatever though, yeah
You can actually see the costs on opencollective, its very transparent
I think the only people making enough to live on though are md5 and people working at hypixel and such
Dunno about md5 these days but I assume so since the project is still going
Crazy how hypixel is such the monopoly over all lol
Using what, some very old spigot base?
Pretty much reworked I’m guessing
iirc they're based on MC 1.6 for most of their stuff
no
Dunno if that started life as CB, CB++, Spigot, or what
this is correct
afaik their SMP (realms) stuff is running the latest versions though
I imagine some of the games run newer versions than others too
I see server providers sponsoring some individual maintainers, ig that messed me up.
iirc some of their games you need 1.12+ to play so probably that version
btw, where is the "comment" option on opencollective? the page says I should enter my discord name in a comment, but there was none :I
contact aurora
alright, gonna do that tomorrow
they are in the eu, so maybe leave them a dm now
or I think we have mod mail somewhere
Updates to Paper do not have any sort of estimate for when they release, ever. Any and all updates will arrive when they are ready, and the only thing to do is wait for them patiently along with everyone else.
im in EU as well and its 4am, thats why I thought I wait 
oh ic
s a m e
i should get to bed
in 5hrs I need to be on a train
it's all about coffee naps
Well, I mean, Ive heard that they're great but coffee doesn't work for me \o/
online uni, so no sleep needed lol 
For real
fuck i never did my homework
thanks for reminding me
I gotta see if I can get diagnosed and maybe see if I can et some aderall or something and see if that works
aderall nap 😄
It was so nice just being able to get up and say “I’m at school now!”
I lived a 5 minute walk to uni, tops
I can count the number of times I was on time on my hands that year
Lmao
Same, either offline or online 5 min max 
Any clue why DataResult and Codec under com.mojang.datafixers and com.mojang.seralization would be not found when compiling with maven? I've included the minecraft-libraries dependency in POM but no luck
We don't support maven, basically
you would have the same issue with spigot
you're best off using paperweight userdev
otherwise have fun manually adding libraries
and idk how you're gonna remap
I guess running tiny remapper manually could work
is there any good resources to switch from maven to gradle? I'm not massively in the know with maven or gradle but heres to learning right
gradle has some docs
If you're doing it just for internals, see the link I shared for a basic setup, but yeah gradle docs are pretty decent for basic stuff
for the most part, for a simple project, the gradle init auto convert stuff does a fairly decent job
anyone know why my debian 10 server keeps downgrading java from 17 to 16 and shutting off the mc server?
im following paper tutorial for java 17
auto update overwrite binary?
?? k just wondering lol
and shutting off the mc server
what do you get in your logs when it shuts off?
unless you just mean it's defaulting to an older version
maybe
in which case, run update-java-alternatives
Linux loads binaries into memory, you can delete files that are still running and they won't stop running
otherwise, when installing, make sure to read what it says it's gonna do, especially with apt
or rather, it doesn't protect against deletion/overwrite for running binaries
er
korb@mc-server:~/mc-server$ java -version
openjdk version "17.0.1" 2021-10-19 LTS
I mean, that would maybe suggest you had some cron or timer task thing to auto upgrade
dont have crontabs at all
ill do some more troubleshooting tomorrow, was just coming here to see if anyone else has had this issues
well, if packages where changed, it'll be in the dpkg/apt logs
some distros have timers setup to auto install security updates
try running apt dist-upgrade and seeing what it says
updated 270mb of stuff
anyone have an idea? is there a guide or something, how to fix "Paper Async Chunk Task Thread ERROR"?
I mean, we can't comment with 0 info
You failed to provide logs or info on what you even screwed with
i can post it, its very long but all the same case i think, at first something like :
[Paper Async Chunk Task Thread #1/ERROR]: Could not de-serialize chunk data for task: Chunk task: class:com.destroystokyo.paper.io.chunk.ChunkLoadTask, for world 'world_nether', (-23,-17), hashcode:52284421, priority: -1
java.lang.ArrayIndexOutOfBoundsException: Index -3 out of bounds for length 18
at net.minecraft.world.level.chunk.storage.ChunkSerializer.loadChunk(ChunkSerializer.java:228) ~[?:?]
at com.destroystokyo.paper.io.chunk.ChunkLoadTask.executeTask(ChunkLoadTask.java:106) ~[paper-1.18.jar:git-Paper-52]
at com.destroystokyo.paper.io.chunk.ChunkLoadTask.run(ChunkLoadTask.java:38) ~[paper-1.18.jar:git-Paper-52]
at com.destroystokyo.paper.io.QueueExecutorThread.pollTasks(QueueExecutorThread.java:105) ~[paper-1.18.jar:git-Paper-52]
at com.destroystokyo.paper.io.QueueExecutorThread.run(QueueExecutorThread.java:38) ~[paper-1.18.jar:git-Paper-52]
and then there is something like:
[Paper Async Chunk Task Thread #0/ERROR]: Could not de-serialize chunk data for task: Chunk task: class:com.destroystokyo.paper.io.chunk.ChunkLoadTask, for world 'world_nether', (13,-7), hashcode:1970547576, priority: -1
java.lang.ArrayIndexOutOfBoundsException: null
i also can post whole log, but there are 15k lines
When this happened to me many moons ago, it was because there were over 1 million llama spit entities in the chunk. in the end i had to delete the region files. it would crash the server if anyone entered the chunk.
Your chunk is busted
First time running on 1.18 and had my server shut off randomly 3 times. No error codes. No nothing. Any idea or is this known?
That error specifially kinda implies that you changed the world height or something and that's basically unsupported
okay, could i change it back in level.dat?
you'd revert whatever you messed with in there
but, any gone chunk is gone, and no idea what it will do with the chunks generated since then
i could paste backuped chunks after i tried to fix level.dat .... the whole story starts with my first try to import my old 1.17 nether, but i got some problems with bed/water/sunrise settings in nether dimension, also creation of portals was different location and scale .. so now i created a whole new one with same seed, but changed DIM1 folder with the old one from 1.17 -> now nether bed/water/sunrise settings are ok, but chunks are not detected correctly
we know something about change of nether height between 1.17 and 1.18 or something like that?
not really possible to say anything without an error. maybe the system shutting it down because its out of memory?
also, check the crash-reports folder as well
nothing changed there, nether still has the bedrock ceiling at y=128 (but is 256 tall in total)
okay so there must be some other change which causes that chunk error
maybe you messed up something while copying files around? I would recommend restoring from a backup and then doing it again and then testing after every step if everything looks alright
if you changed the world height you could also try changing it to the correct value again, look at the chunk data (NBTExplorer), count how many sections are in there and then calculate the height the server expects from that. that might work
okay thank you for your help guys, ill try to compare those 2 at every setting ... till yet, i only checket the WorldGen Part, but may there are some other important points that could cause those chunk errors
is there a working minecraft:aether_test dimension datapack? I don't think the server likes loading worlds with dimensions that don't exist
in 1.17 there was
upgrading with data packs which provide custom biomes, etc, do not support server upgrades
if i fix this name to world_nether, i also get chunk error
made slightly better in 1.18, but, coming from 1.17 won't work
is vanilla able to upgrade worlds with datapacks?
No
well, nice one mojang
Why do you think mojang had the warning saying that you wouldn't be able to upgrade if you had the C&C pack
yeah, but wasn't that about height datapacks only? or all world gen?
custom biomes, etc
in fact, if you used custom biomes with spigots API there is a chance it could perma damage your worlds biomes
then that world is probably lost. you could try using MCASelector to copy the chunks out of it and to replace the custom biomes with vanilla biomes (not sure if it can do that, maybe with the filter feature)
you think WanderingTraderId: 4Integers with -1509730306 -8305610 -1302592036 -1743011540 in it is very important in level.dat? 😬
no, it just saves what the trader is doing
good to know, i would give it a try if i dont get it fixed another way
headless java?
It's not that it couldn't find the class, it's that it bleew up tryna init it
the fact that that's stems from another error, er
It was tryna log an error already in pollTasks
this is with paper or a fork
Is anyone recognising bugs from changing 1.17 -> 1.18
only loss of stability
send change view distance packet to the player
https://wiki.vg/Protocol#Update_View_Distance
however it might cause issues when the player moves too far away
random crashes etc etc
is it ok to use --forceUpgrade --eraseCache if there is rain everywhere in water/ungerground/houses/caves?
ive heard many times not to use forceUpgrade on paper
any other idea how i could fix rain problem
what about --eraseCache .. could this fix rain issue? or is it also dangerous to use with paper .. or unnecessary
if the heightmap/light data stuff is borked, yes
but, many times that's plugins screwing with data sent to the client
here is example of my rain/snow indoor problem ._.
i only have this in one dimension, normal overworld is ok
most times i import my old overworld/nether the way, that i create a custom dimension(data pack) with overworld or nether settings, and paste the dimension data into dimension folder
cause those custom dimensions functions doesnt seem to work on paper 1.18 i try it with settings in level.dat
got many problems with nether and end, but the only known problem with my second overworld it this rain going through everything in some spots
yeah you will need to do that forceupgrade thingy but only do it with a backup
okay thanks, ill try - ill have to do it with whole server, cant do it for only one selected dimension right? i backup everything ;-;
if i destroy my server that way, would i note it early at next startup? 😅
its doing something .. sometimes ArrayIndexOutOfBoundsException:null but its also going through many chunks
still rain everywhere, but i think it didnt break anything
it’s safe if you wear your mask
Alright, thanks.
jk
Next question. Who knows under what update number 1.17 became the stable version and the red warning sign disappeared?
No there is no eta
Version number doesn’t have any relation to how stable they are
It’s just how many changes are made and pushed
No, I mean, when did this red sign with the 1.17 warning disappear?
typo?
No
I just changed in the page code. I thought there was a similar warning sign for 1.17.
Oh, okay.
Yes we put out warnings for 1.17 too
We always do that for earlier builds because we don't know how stable it is
Once we think enough people tested it and nothing came up we will remove the warning
But you can run paper 1.18 without any issues, just be careful and take automatic backups
It is quite stable for me :/
Thanks a lot for the explanation, but I still haven't found out when did you remove the red warning sign for 1.17? At least approximately after what number it was.
Well I dont think this matters cause time for a "stable" version is different every time :)
Updates to Paper do not have any sort of estimate for when they release, ever. Any and all updates will arrive when they are ready, and the only thing to do is wait for them patiently along with everyone else.
Regardless, I would still like to know this.
Then spend your own time looking through builds on gh
Whatever is happened on June 26th will be the one
Thought it wasn’t written in assemblies, sorry.
Thanks.



