1.21.9 Snapshots

[green saddle]

Looks like mojang did a punctuation pass this snapshot

[surreal wind]

this is both better and worse than the mini kb I'm using for a project

smh phone

[ember lintel]

oh no our model data patches

[onyx plinth]

That Q key gives me pain.

[keen cloak]

Huh? This is completely irrelevant to baked models, it does not affect the model data system

[orchid adder]

why is the Q key so wide

[surreal wind]

they didn't have space for a tab key there so they expanded it to the left

[winter bear]

which part mutates that and when, BE or renderer?

[keen cloak]

The renderer

Where previous versions called render() directly, it now calls extractRenderState() and then render() with extracted state

The plan appears to be to move the render call off-thread at some point

[shrewd dirge]

so the blockentity updates its data first, then your extractRenderState override passes data from the BE to the state, then render runs with the state provided

[onyx plinth]

Anyway the Make and Share functions were attached to F16 and F17 (I forget which is which) and were handled by the OS my Raspberry Pi 3 came with (the OS also felt like a toy but it taught me how to use basic terminal).

[keen cloak]

Yes

[surreal wind]

yes, with the goal of eventually possibly multithreading render calls

[onyx plinth]

Why am I rambling in the snapshot channel?

[surreal wind]

🤷‍♂️

[winter bear]

the only BER I got is for the Railgun, and currently I store yaw and pitch in the BE, but technically only the BER cares about that, so I could move that to the render state then?

[fringe summit]

not could

must

[shrewd dirge]

well, where will the renderer get the data from

[fringe summit]

if they do what I think they have to do, you dont get BE access

in the BER

[surreal wind]

yeah, like entities now

[shrewd dirge]

no no I mean

extractRenderState

[ember lintel]

I think the deadmouse ears are gone

[winter bear]

I meant I don't have to store it in the BE anymore and just have it in the state

[shrewd dirge]

where do you get the yaw/pitch from when you store it in the state

[fringe summit]

the state is likely immutable, that depends on how you organize that in your be

you could create the state everytime it's queried from those two fields in your BE

or cache the state, IDK, haven't checked their code yet

[winter bear]

a couple messages before it said mutable

[surreal wind]

assuming it works like entities already work

[onyx plinth]

*deadmau5

[fringe summit]

oh,I missed mutable

mutable would be a bit out of character 🤔

[surreal wind]

currently entities have a CLASS for their state so you can't use a record

[fringe summit]

okay odd, never looked at it that closely

[surreal wind]

on top of that you receive an object

so you are REQUIRED to mutate

[winter bear]

from the selected target BlockPos and math

[fringe summit]

okay, that is fine, they manage the lifecyle in that case

[ember lintel]

not in code

[surreal wind]

assuming this didn't change in the snapshot

[fringe summit]

as long as you dont hold onto the reference that can be more memory efficient

[winter bear]

👀 labels

[fringe summit]

i think you could probably do a "frontbuffer/backbuffer" kinda deal

for each state obj

[surreal wind]

labels?

[fringe summit]

in case rendering can overlap with querying state for the next frame

[winter bear]

label96:

[surreal wind]

oh

that's decompiled code

every single time I open intellij it "forgets" the downloaded sources

[winter bear]

yeah, I know, still I don't see them often

[surreal wind]

and I can't attach without clicking Download sources

(okay "every single time" is an exaggeration, I haven't open this project in over 30 days to intellij/gradle has graciously chosen to wipe all cached data about it)

(despite me actively DISABLING automatic cache cleanup)

[winter bear]

but looking at this, I wouldn't give it the target angles but the current angles

[surreal wind]

yes

state.value = lerp(current,target,partialTick)

[ember lintel]

they probably don't want unnecessary frequent allocations

[keen cloak]

At least for entities, 25w33a changed it to create a new render state object for each entity being extracted (which makes sense for their goal of pushing the actual rendering off-thread)

[winter bear]

potentially slerp

[surreal wind]

there, the proper sources don't have labels

[orchid adder]

can it become a value class

[ember lintel]

with Valhalla

[surreal wind]

it will be nice when valhalla finally lands

I'll still be annoyed at java for not wanting to commit to the promise of making it a REAL value type and only promising the semantics of it at the highest level

but it will be nice

(like, I get it, java doesn't want the language to dictate what the jvm must do, but rather just provide the declarative tools to tell the jvm what it CAN safely do)

[winter bear]

so it's message sending concurrency

Do not communicate by sharing memory; instead, share memory by communicating

[cyan ether]

don't be hopeful it happens in our lifetime

[surreal wind]

that's not sharing memory, that's synchronizing memories.

[winter bear]

the rust/go/idk approach to multithreading

[orchid adder]

i'll be in Valhalla before Valhalla comes to me

[winter bear]

it's a slogan from the Go docs ¯_(ツ)_/¯

[static dagger]

Sorry i was at work

Well still am but you get it

[winter bear]

you were at work doing work

you are now no longer doing work

[surreal wind]

terrible slogan, no wonder I don't like go :P

[static dagger]

Snapshot 25w36a

Article: https://www.minecraft.net/en-us/article/minecraft-snapshot-25w36a
Changelog: https://misode.github.io/versions/?id=25w36a
Notion: https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2624f070f8888019a837e253c146c41f&pm=c
Snowman: https://github.com/neoforged/Snowman/commit/f23239ea40b6ad5199ca990a657040b07cf0ae35
Primer: <>

Slicedlimes Videos:

• Main:
• Pack:

[surreal wind]

they could have said "Don't share memory, share memories." and it would be so much nicer

[frank patrol]

Wooh, no longer will I need multiple checks for disabling direct attacks, sweeping edge and projectiles (hopefully)

[winter bear]

well, the idea is that you don't share access to a region of memory but send an object containing the relevant data to the other thread

[frank patrol]

Oh, I was scrolled up to the last snapshoot oop

[surreal wind]

hence, instead of sharing the memory (the space in your RAM), you share memories (data objects)

[keen cloak]

Well, shit, key mapping categories are an enum now

[surreal patrol]

Extensible enums!

[keen cloak]

Yes, but still annoying that the vast majority of mods will have to do that

[thorn sable]

Mannequins? Huh

Sounds like fake player, is neos fake player ded now?

[keen cloak]

Nope

[thorn sable]

But they are fake players, or am I misunderstanding

[keen cloak]

They abstracted a few player properties out into an Avatar class which both the player and the mannequin extend

[onyx plinth]

It's basically like the NPCs you see on servers like Hypixel.

[thorn sable]

Ahh interesting, I assume they look like players, could you give them someone’s skin, could be a nice tool for videos and such

[winter bear]

they are just entities that look like players, they can't interact with stuff aiui

[onyx plinth]

They apparently have a giant NPC label above their heads. So more like a mapmaker tool.

Though I haven't seen maps being made in a long time, not that I've actually checked.

[thorn sable]

I like the sound of them tbh wouldn’t have thought mojang would add something like this but neat nonetheless

Wonder if it’s something slicedlimes been playing around with in his snapshot vids

Would be a good way to test them tbh

[onyx plinth]

I remember on bedrock I managed to hack in a spawn egg that spawned the player entity. This just reminded me of that.

And by hack in I mean a save editor.

[thorn sable]

Heard it here first Minecraft players are hatched from eggs

That’s the new lore

[onyx plinth]

Well, if you wanna get really technical…
But this is a family friendly server, right?

[thorn sable]

yeah very family friendly-christian server

just ignore all the shenanigans that happens in #squirrels-🦊

[onyx plinth]

Yeahhh….

[thorn sable]

psst, updating my notion page, should be able to grab any missing links from there
https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2624f070f8888019a837e253c146c41f&pm=c

[surreal patrol]

So many key changes

[keen cloak]

I really like the input handling changes

[keen cloak]

They aren't, it's just hard-coded in a different place now and the flag passed around has the generic name showExtraEars

[surreal patrol]

I love the name InputQuirks

[keen cloak]

Hehe

[winter bear]

specialMagicHackyFocus vibes

[thorn sable]

oh we might be getting a 36b?
https://bugs.mojang.com/browse/MC/issues/MC-301749

[orchid adder]

sounds hotfix-snapshot worthy

[plain zenith]

ecj: Permitted type CompoundTag does not declare net.minecraft.nbt.Tag as direct super interface

[thorn sable]

just play singleplayer, there fixed the crash

[keen cloak]

I'm really starting to like this snapshot. BER render state extraction happens before the frame graph is built, so we can re-introduce the BE extension for using the glowing effect in BERs which had to be removed when the frame graph system was initially introduced \o/

[fervent maple]

No friends, no bugs

[winter bear]

isn't $$5 a float

wait no

I read that wrong

it's $$5 in extractVisibleEntities, not shouldRender, it's trying to call shouldRender but fails

[keen cloak]

The crash is in EntityRenderDispatcher#shouldRender() when calling EntityRenderer#shouldRender() because the renderer is null

[surreal patrol]

I think particles have been added to the feature render system now

[keen cloak]

Indeed

[fervent maple]

b

25w36b

[ashen bison]

b snapshot is out?

[fervent maple]

Yes

[vernal sinew]

New version detected: 25w36b.

Minecraft Snapshot 25w36b

Here is Snapshot 25w36b to fix some common crashes.Fixed bugs in 25w36bMC-301745 Placing an ender chest in the world crashes the gameMC-301749 Multiplayer Crash....

Sep 2, 2025, 4:05 PM

[fervent maple]

https://www.minecraft.net/en-us/article/minecraft-snapshot-25w36a

[thorn sable]

Snapshot 25w36a/b

• Primer: https://github.com/ChampionAsh5357/neoforged-github/blob/update/219-or-22/primers/1.21.9/index.md
• Article: https://www.minecraft.net/en-us/article/minecraft-snapshot-25w36a
• Notion: https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2624f070f8888019a837e253c146c41f&pm=c

Changelog:

• 25w36a: https://misode.github.io/versions/?id=25w36a
• 25w36b: https://misode.github.io/versions/?id=25w36b

SnowMan:

• 25w36a: https://github.com/neoforged/Snowman/commit/f23239ea40b6ad5199ca990a657040b07cf0ae35
• 25w36b: https://github.com/neoforged/Snowman/commit/af69b4ba9f50958b1da8d225796502463614b7a6

SlicedLimes Videos:

• Main: https://www.youtube.com/watch?v=x_TLVuM5FUM
• Pack: https://www.youtube.com/watch?v=jDYmkVMtE9k

-# New pin to include 25w36b

[shrewd dirge]

we've had snapshot but what about second snapshot

[plain zenith]

Also what is the jetbrains annotations versioning

Of course the b snapshot broke a client patch

[winter bear]

ah, right, that's why I got confused

[surreal wind]

the 404 page is neat https://www.minecraft.net/en-us/article/minecraft-snapshot-25w36b

[thorn sable]

part of the main 36a article

much like 34a/b

[surreal wind]

I know

I just clicked on the other thing first

the bot message

and noticed the 404 page

[thorn sable]

oh yeah their 404 is real nice

[winter bear]

used to be different

less interactive

also, they're clearly hinting at dynamic lighting

[orchid adder]

do you get the cave or the stronghold

[winter bear]

stronghold and then on reload cave twice

definitely more cave, but both

but the old page used to be just steve and alex looking into a cave iirc

MC-301745 is funny
guess they reused the chest renderer for enderchests but forgot to add the check

I do wonder how long it's gonna take for someone to make a mod that spawns a mannequin with the same equipment every time a player logs off

[thorn sable]

while i like multi letter generics for readability
mojang could you please atleast prefix with T - TAvatarLikeEntity
T meaning type

[surreal patrol]

The particle changes reuse so much of the old system but replaces their intentions that I'm probably going to need two or three separate examples just making the particle itself

[keen cloak]

Reminds me of Ark putting you to sleep when you log off 😄

[gritty swan]

I can remove my hacks now \o/

[thorn sable]

CM hacks i assume, tbh didnt know CM dimension had a worldborder
or is it hacks to remove/make it stupid big

[gritty swan]

It's hacks to reset it if you make it tiny for the overworld

[thorn sable]

ah yeah, playing 1 chunk in overworld you prob dont want that in CM xD

CM just spawn you out of world border and kills you

[surreal patrol]

I thought for a second that they tossed out derived level data, but no, they just turned the world border settings into its own saved data

Wait no, what?

Oh ok, I'm right. I was just super confused since they have to handle legacy support

[thorn sable]

oh screens got a slight overhaul
no longer are raw int, double for input stuff passed around
screens now get new objects KeyEvent MouseButtonEvent etc containing all the necessery info
plus some very nice helpers InputWithModifiers#isConfirmation (both the above implement InputWithModifiers)

[surreal patrol]

It's really nice, even though this update is just a giant rewrite central

Almost every rendering system has been touched at this point

[thorn sable]

much better than checking for the keycode myself tbh

public boolean keyPressed(KeyEvent e) {
   if(e.isEscape()) return false;
   // ...
}

oh with this the static screen helpers for if shift/alt/control are pressed
are moved into Minecraft (not static) theres also variations in InputWithModifiers (again not static)

so anyone who was using those static helpers, rip

but should be a quick fix, for them

[crystal ermine]

oh my god end me

is it a sealed InputEvent interface or completely separate?

[thorn sable]

KeyEvent and MouseButtonEvent are records which both implement InputWithModifiers
InputWithModifiers is a regular old iface consisting of mostly helpers and around 2 methods int input() and int modifiers()

[crystal ermine]

is int input() the mouse button?

[thorn sable]

yeah for KeyEvent that maps to the key code and MoustButtonEvent maps to the mouse button

[crystal ermine]

is doubleClick gone?

[thorn sable]

KeyEvent and MouseButtonEvent ofc have additional info about those specificlly
like MouseButtonEvent has the mouse position
and KeyEvent has the scan code

[thorn sable]

the bool flags passed to the input methods in screen are still present

they were not moved to the new input objects

[crystal ermine]

separate from the MouseClickEvent?!

that's silly

[crystal ermine]

this feels very half baked.

[crystal ermine]

it should be iface MouseEvent, then MouseClickEvent, MouseDragEvent, etc

and MouseEvent extends InputWithModifiers

[thorn sable]

yeah could be better, but this is still alot better than what it originally was imo
checking for key codes and mouse buttons is blegh

then you also gotta check modifiers and stuff too, no
having all the nice utils to do it for it is very nice

theres also this class which the utils make use of

[fervent maple]

https://mojira.dev/MC-301782 lul

[winter bear]

wrong AABB

one could check in neo

probably dev only tho

don't know if those commands are available in prod

[fervent maple]

it is

but the bug has already been fixed in NeoForge

https://github.com/neoforged/NeoForge/issues/874

[winter bear]

ok, fair, guess they just implemented the culling without double checking the AABBs

[cyan ether]

ah great, does that mean we might have to rename our events

[keen cloak]

That's the issue: they don't use AABBs, they use the particle's position (which is seemingly intended to be the center or bottom center of the particle) to check for frustum-point-intersection

[thorn sable]

dont think so, ours are called InputEvent.[MouseButton|Key] etc
but we should update them to take in vanillas events imo

[winter bear]

but they got collision detection?

which use AABBs iirc

[thorn sable]

maybe also have our events implement InputWithModifiers similar to vanilla
so that our InputEvent.[MouseButton|Key] exposes all the same utils as vanilla

[fervent maple]

https://www.youtube.com/watch?v=x_TLVuM5FUM

[ember lintel]

ok it's time, what are your guesses for how many patches will be broken in this snapshot

[tiny wing]

more then 1

[surreal patrol]

A lot

Anything in particle world is probably angry

[sleek shore]

NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

i've been using em

[gritty swan]

Mojang, again: 👁️

Gander is so far in the grave it might reincarnate into vanilla code soon

[thorn sable]

we joked about mojang yoinking monica
but no, they just yoink the code instead

[winter bear]

aren't there just version with different parameter orders but the same parameters overall

[ember lintel]

well it's definitely a bit more than 1
here's the client package

[fierce badger]

for rejects???

[thorn sable]

well congrats fin, you were right, it was > 1

[ember lintel]

it is 30 files in total

screm I think they changed weathering they use the WeatheringCopper interface for lightning rods now

[ember lintel]

26 left

[keen cloak]

I'm gonna dive into the client rejects

[ember lintel]

I did the widget ones

which ones are you working on?

[keen cloak]

I'm currently going through the screen ones and then I'll look into the renderer rejects

[ember lintel]

spyglass is now a use animation

[fierce badger]

ummm always has been

or it is as of 1.21.1

[keen cloak]

I'm gonna leave the KeyBindsScreen reject alone for now

[ember lintel]

in the last snapshot there was still a hardcoded item check in the player renderer

[fierce badger]

oh must've been a recent change

its a UseAnim in 1.21.1

the hardcoded item check was to change the transform to head for rendering

[ember lintel]

nevermind it just moved

[marble widget]

Interesting, so they opted for performance over correctness

Makes sense given how many particles are in a world at once and the fact that most particles are very small

[keen cloak]

Yeah. At least with the campfire smoke I couldn't get it to break, so ¯_(ツ)_/¯

[ashen bison]

mojang spelling mistake

report to gegy if yall wanna

[surreal patrol]

Take your dang primer (both 25w36a/b): https://github.com/neoforged/.github/pull/29/commits/f11f58c41da70b2aa599613aed82da04b8b87531

Basically more block entity submission rewrites, the particle feature addition, and input consolidation

[fierce badger]

ClientMannequin - The client version of the Mannequin entity.

wtf is that

can they be client only?

[surreal patrol]

Basically, just imagine ClientPlayer except for a mannequin

Not everything needs to be handled by the player on the server

Such as walk cycles and the like

[fierce badger]

they cant walk tho

[surreal patrol]

They're living entities, so they can be moved. The hurt animation plays part of the walk cycle anyway

[winter bear]

They can

[mellow rivet]

pt1212

[Reference to](#squirrels-🦊 message) #squirrels-🦊 [➤ ](#squirrels-🦊 message)https://packaged-media.redd.it/quvohmgi1smf1/pb/m2-res_1080p.mp4?m=DASHPlaylist.mpd&v=1&e=1756872000&s=c341db1b12f4b6dad550cfa9268cf9188c04be99

platinpython referenced · Sep 2, 2025, 11:02 PM

[surreal patrol]

You know, I'm sort of curious to see the old adventure and horror maps updated to use mannequins

[runic crypt]

We finally have a built in "Fake" player

What does it extend?

[heavy iris]

It wont help modding unless they go and replace all Player parameters everywhere with Avatar

[heavy iris]

Avatar

[runic crypt]

And what does that extend?

And what does Player extend?

[heavy iris]

LivingEntity -> Avatar
Avatar -> Player
Avatar -> Mannequin

[runic crypt]

Hmm, sounds good enough

[heavy iris]

very blegh

[runic crypt]

As as the Mannequin instance can be passed thru into a func that wants a player

[heavy iris]

No, it can't?

[runic crypt]

Avatar doesnt extends Player?

[heavy iris]

no

Avatar is the new super type of Player

[runic crypt]

Ok that is what I was asking

Ahh

Player extends Avatar?

And Mannequin extends Avatar?

[heavy iris]

Avatar extends LivingEntity
Player extends Avatar
Mannequin extends Avatar

[runic crypt]

Ah

So it's more of an NPC

rather than fake player hrm

[heavy iris]

so hopefully they demote all current Player params everywhere to Avatar

Yeah, its an NPC

[runic crypt]

Ah okay

Still useful for servers and whatnot for NPCs atleast

[heavy iris]

100% for server-side-modding

[runic crypt]

Yah

[surreal patrol]

server side modjam part 2: avatar boogaloo

[runic crypt]

I wonder how hard itd be to make a Fake Player 🤔

[heavy iris]

Imean, we still have FakePlayer, so..

[runic crypt]

Yeah

[heavy iris]

its a solved problem, just would be nice to yeet

[runic crypt]

Well yeah

[plain zenith]

For this to happen you seem to need:
a source file that will attempt to load the class implementing the interface after permitted types would normally be checked (like a record with the class as a component)
the pair of sealed interface and implementing class
1999 other source files alphabetically between the source file and the sealed interface and implementing class pair that do not reference either the class or the interface

when ecj loads the implementing class it catches it up to the current step which causes it to also load the sealed interface which it also attempts to catch up to the current step but because the class is already being caught up it doesn't initialize the interfaces of the class before checking the interfaces

[ancient dawn]

How the... you know, sounds like ecj. Lovely

[heavy iris]

ecj doin ecj things i see

[plain zenith]

https://github.com/eclipse-jdt/eclipse.jdt.core/issues/4331
It has already been reported

[ancient dawn]

Only 2 weeks ago too; that's... surprisingly recent? Wacky

[broken nest]

mannequin update

[ember lintel]

mojang please un enum the key mapping categories

render types for particles also come from an enum now

[runic crypt]

Mojang please remove enums for stuff that modders could expand on, thanks! 😁

Prob won't happen

[winter bear]

No more enums for anything

More chest types

[runic crypt]

But asking is worth it

And please Mojang

Please, make more cute mobs such as Copper Golem!

And add "secret" behavior, such as said golem being gifted a rose by the iron golem

[winter bear]

I'm asking for eldritch horrors beyond human comprehension instead

[runic crypt]

😂

[ember lintel]

so... the warden

good news all automated tests pass

[gloomy sable]

make Direction a registry!

[blazing vector]

cant u just expand enums

[runic crypt]

Nope

Not without asm and hacks

[gloomy sable]

you can with bytecode modifications, but its hacky and has required a lot of effort on neos part to make it work nicely

[blazing vector]

ya but its works nicely

[runic crypt]

Enums are Immutable by default

[fading dew]

That will be useful for all the mods that add other spatial dimensions.

[winter bear]

Not enough beyond human comprehension

[runic crypt]

No it doesnt

[winter bear]

Good start tho

[runic crypt]

It's horrible

Everywhere an enums is being used and has a switch case without a "default" means ALOT of patching

To make said enums expandable

So enums requiring asm and hacks makes it take more work

[winter bear]

I might work nicely at the end for the user, but it's not nice to implement

[runic crypt]

Yeah

And what if two mods decide to have an enums with the same exact name?

[blazing vector]

path will still defer

[runic crypt]

Alot of conflicts and various other issues can occur

And that's why namedspaced registries work better

[runic crypt]

No, it crashes.

Can't have two enums that are the same

Jvm doesn't know how to handle that

[blazing vector]

oh didnt know that

[runic crypt]

And what if the mod uses different parameters then the other mods that also uses it

I did say enums are immutable

[blazing vector]

should probably rename my enum then

[runic crypt]

By default enums are immutable

[blazing vector]

instead of LightState

[runic crypt]

So yeah

Should do modid_enum

[blazing vector]

ya

[winter bear]

If it's your own enum class, that's fine

[blazing vector]

its my own enum class

[winter bear]

It's specifically for two enum values being the same

[runic crypt]

Then it's fine

[blazing vector]

but what if some other mod makes it with same name

[runic crypt]

We aren't talking about enums you create

I'm talking about enums that already exist in the code

[blazing vector]

oh ok

[winter bear]

Which wouldn't compile normally, but via the framework that's not checked until runtime

[runic crypt]

But if you're using the enum extension system, it can be a problem

[blazing vector]

so 2 people expanding same enum and they somehow get same values for 2 different entry

[runic crypt]

Or different values

For the same entry

Conflicting

[blazing vector]

ya

so 2 different value for same key

[runic crypt]

Ye

[blazing vector]

thats more likely to happen then case 1

[ember lintel]

me and my friend chilling in the new snapshot on neoforge /j

[blazing vector]

lol etho mannequin

[winter bear]

Make direfall loop

[blazing vector]

can u make mannequin do still like attack or mine?

using mods

[winter bear]

Also, did you mod out the "subtitle"?

[blazing vector]

as technically its just a living entity

[ember lintel]

no it's distance based obviously the real Etho

[winter bear]

Ah, and the name got a higher distance?

[blazing vector]

i should back port them

[runic crypt]

FakePlayer works

[blazing vector]

true

[runic crypt]

Waste of effort to back port unless you just want NPCs

[winter bear]

looks at CustomNPCs

[runic crypt]

😂

[narrow palm]

Enum extensions require a namespace

[shrewd dirge]

submitBlockModel A pose stack, the render type, block state model, RGB floats, light coordinates, overlay coordinates, and outline color
why not ARGB

feel like rendering transparent blocks from BERs is going to get even more annoying than it already was

[surreal wind]

https://www.youtube.com/watch?v=jDYmkVMtE9k

Etho, Mumbo, PhoenixSC, no idea, no idea, Grian

[orchid adder]

IIRC, from left to right: Etho, slicedlime, Mumbo, Grian, PhoenixSC, evbo (from the hit series Parkour Civilization)

[surreal wind]

ah. I don't know slicedlime's skin from memory

[orchid adder]

I think my mind labels it as "white-haired Link"

[surreal patrol]

Sometimes I look at those snapshot videos and go, man it must be nice not having to deal with anything behind the scenes

[fervent maple]

Snapshot day

[surreal wind]

[runic crypt]

[ember lintel]

T-1h

[orchid adder]

i wonder what they'll release this time

moar wrecking balls to rendering?

[tiny wing]

New tree gen

[orchid adder]

ah yes, the tradition

[tiny wing]

Spheres

Jiggly physics

[shrewd dirge]

jiggly frogs?

[ember lintel]

hmm I just checked, mojang switching to java 25 seems unlikely before LWJGL4 is out

[fringe summit]

It's not that LWJGL3 doesn't run in J25

At least not to my knowledge

The Microsoft OpenJDK LTS has to be out first anyway, although that might be quick

[ember lintel]

well yea but LWJGL 3 will not get Panama, that's scheduled for LWJGL 4

https://github.com/orgs/LWJGL/discussions/609

[fringe summit]

Sure, but Panama isn't the only benefit to J25

[thorn sable]

T-10mins

[ember lintel]

It's the biggest tho (at least for vanilla minecraft)

[shrewd dirge]

https://images.squarespace-cdn.com/content/v1/582be63e5016e168ebdbfe45/1541497081443-8TPJRWJ1BB804BHBPLL7/hopkin04.png?format=750w

[fringe summit]

compact object headers

[keen cloak]

Flexible constructor bodies

[fringe summit]

I was gonna bunch that together under "various language improvements"

• general JVM performance improvements
• GC improvements
• JFR cooperative sampling / depends on their plans for the JFR integration

[ember lintel]

isn't that just memory stuff... I guess it would reduce the overhead of blockstates

[shrewd dirge]

flexible constructor bodies is huge for mc mods

[keen cloak]

Yup

[fringe summit]

"isn't that just memory stuff"..... significantly reducing memory use for their players and servers is not "just memory stuff"

well. i guess it is just memory stuff 😄

[ember lintel]

class file api is gonna be huge for us (we can yeet asm)

[keen cloak]

We can't

You'd first need to rewrite mixin almost from the ground up to no longer use asm

[ember lintel]

well then @mumfrey

[keen cloak]

If anything, it would be Fabric mixin that is switched, upstream mixin doing so seems incredibly unlikely to me due to the java version compat range it provides

[orchid adder]

I'd think upstream Mixin would be unlikely mostly because of how slow it moves/updates

[ember lintel]

well then poke the fabric mixin devs to move to the class file api

[orchid adder]

did someone say new project /j

[ember lintel]

T-0
any moment now

[orchid adder]

i'll grab the flight tickets to Sweden and the picket signs if there's none /j

[ember lintel]

article: https://www.minecraft.net/en-us/article/minecraft-snapshot-25w37a

[thorn sable]

Snapshot 25w37a

• Primer: https://github.com/ChampionAsh5357/neoforged-github/blob/update/219-or-22/primers/1.21.9/index.md
• Article: https://www.minecraft.net/en-us/article/minecraft-snapshot-25w37a
• Changelog: https://misode.github.io/versions/?id=25w37a
• Notion: https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2694f070f8888085abbcfbc5d6b631b3&pm=c
• SnowMan: https://github.com/neoforged/Snowman/commit/6185955ea9df2576c4b70b05602abdb80c855d8d

SlicedLimes Videos:

• Main: https://www.youtube.com/watch?v=Bu1O2wH0BCg
• Pack: N/A

||<@&1067092163520909374>||

[orchid adder]

it's here

[ember lintel]

OpenGL 3.3 is now required to play the game

[tiny wing]

OpenGL 3.3 is now required to play the game, up from 3.2. This does not change our posted Minimum Requirements
Developer’s Note: We do not expect the update to OpenGL version to affect many people, if any at all. Our posted Minimum Requirements has been OpenGL 4.0 since 2017, and OpenGL 3.3 turned 15 years old recently.

[vernal sinew]

New version detected: 25w37a.

[runic crypt]

How old is MC?

16y

So 3.3 seems fine

They prob could get away with using newer version tbf

[orchid adder]

TLS enabled by default, but also requires a keystore? is the keystore autogenerated

Debug features

• Some debug features that were previously removed during compilation are now preserved and can be used by players

[runic crypt]

I wonder what those are

[inner hearth]

I mean the server management protocol is disabled by default, so I'm guessing it means more settings to set when you enable it

[orchid adder]

fair point

[runic crypt]

Why dont they just use OpenGL 4.0 as the required min?

Who would even be affected?

[shrewd dirge]

Moved the result slot of gui/container/villager.png up by one pixel

[runic crypt]

Ahh

[mellow briar]

Fabric mixin is unlikely to switch since FLoader is supposed to remain backwards compatible forever

[orchid adder]

that's part of the reason why they're slow to change that stuff

[runic crypt]

But does that mean... They support 3.3?

[onyx plinth]

Datapack version 69

[orchid adder]

nice

[thorn sable]

https://github.com/neoforged/Snowman/commit/6185955ea9df2576c4b70b05602abdb80c855d8d

[orchid adder]

-# it's raining snowmen, hallelujah

[ashen vault]

Java 25 is coming out in a week, I think. Were there any rumors about 1.21.9 on Java 25? 🧐

[vernal sinew]

Minecraft Snapshot 25w37a

A new week, and a new snapshot. This one includes another big batch of bug fixes, copper golem polishing and our beloved friend can now open doors.Happy mining!ChangesOpenGL 3.3 is now required to play the game, up from 3.2. This does not change our posted...

Sep 9, 2025, 1:59 PM

[onyx plinth]

They added some debug flags for modders, for things like brains.

[onyx plinth]

It's not even on Java 24.

[ember lintel]

time for neomixin?

[ashen vault]

java 25 is lts

24 is no

[mellow briar]

I wish but it's such a tricky framework that I'm not sure we have the resources to maintain it

[orchid adder]

outside of serious production use, I am quite curious how Mixin would look like using Class-File API rather than ASM

[runic crypt]

24 isn't major

It's minor version if I recall correctly

They don't use minor versions

Last time was Java 6 and 7

So chances are they might bump it

[plain zenith]

https://github.com/neoforged/NeoForm/actions/runs/17585221774/job/49951699949#step:9:239
renameClient somehow failed

[runic crypt]

Or they will on 1.22

[orchid adder]

I mean, it isn't unprecedented that they'll bump to a non-LTS version of Java

[thorn sable]

so many debug flags in SharedConstants

[onyx plinth]

Was Java 21 an immediate bump?

[orchid adder]

after all, they did update from Java 8 to Java 16 (non-LTS) first, then moved to Java 17 (LTS) a couple versions later

[runic crypt]

Well J24 doesn't have what J25 has xD

But yes they have a couple times

[ashen vault]

Java 21 -> 25 has quite a lot of new features, so

[runic crypt]

Ya

So I'm sure they'll do the leap

[orchid adder]

we shall have to wait and see

[runic crypt]

Yep

[orchid adder]

maybe this cycle, or maybe the next

or soon™

-# c'mon Valhalla

[runic crypt]

In the meantime we can all enjoy a cup of Java

[onyx plinth]

Ok whatever I'm stupid then I'll just stick to breaking open KAPowerRender.

[mellow briar]

Hmmm, ART bug maybe?

[orchid adder]

bugs? in our software?

[plain zenith]

Nevermind I made a mistake with my regex for maven artifacts for the update libraries step

[orchid adder]

crisis of faith averted /j

[keen cloak]

Keybind categories are now a record holding an RL and their index in a list is used for sorting. Crisis averted 😄

[runic crypt]

With the debug stuff now included in releases/snapshots for us

I wonder what else they have they remove before shipping it to the public?

[bitter kelp]

tests i guess

[runic crypt]

Huh

[orchid adder]

Mojang heard our prayers

send virtual hugs to your nearest Mojangsta today!

[ember lintel]

We have been heard

[sturdy wave]

🙇

[ember lintel]

shouldn't we be bowing to you guys?

[plain zenith]

sed -i -E "s/net.neoforged:AutoRenamingTool:[^\"]*/net.neoforged:AutoRenamingTool:$AUTORENAMINGTOOL_VERSION/g" this was the command I used after I changed it because of jetbrains annotations versioning

[sturdy wave]

I guess thanks for being vocal about it

[runic crypt]

Now do Rarity 😄

Make Rarity a Registry 😄

[ember lintel]

another enum has been recordified

[sturdy wave]

If we need to touch it, maybe. Key categories in this case were fixed because we broke their extensibility in the same snapshot cycle, so it was worth to unbreak it again. Also a couple dozen people immediately yelled at us so there's that 😄

[cyan ether]

hadn't realised yelling was so effective

[cyan ether]

that'll probably need to be concurrent'd

[mellow briar]

You mean event'd

[ember lintel]

https://tenor.com/view/the-road-to-el-dorado-both-both-is-good-gif-8304204

[keen cloak]

We already have an event for keymapping registration, just add category registration to that

[plain zenith]

the classes in the internalapi package have been split into interfaces and Impl classes

[ember lintel]

someone at mojang has recently learned a few clean design patterns and wants to use them

[fringe summit]

.... "clean design patterns" is the opposite of Interface + single Impl 😄

If you gotta do it, you gotta do it. If I had to guess, probably a gradle project split

[ember lintel]

that is literally the service pattern, make an interface that's used everywhere and an impl that handles the stuff

[sturdy wave]

It's situational :)

[ember lintel]

noted:
"yelling at mojang to change something works"

[cyan ether]

I was under the assumption that it's an interned identifier and then I looked at the code

yeah uh, event I guess

that's one way to "register" stuff

why is that method public

who needs to be stabbed for that

[ember lintel]

so fabric can use it and explode

[cyan ether]

it won't explode

[mental saddle]

Now we just gotta get their attention on versioning?

[cyan ether]

in either case we'll have to have the event store the key mappings in a temporary list, sort and register all of them at the end

[keen cloak]

Why bother sorting them?

[ember lintel]

event default sorting should be fine (mod load order for @SE and registration order for listeners)

[cyan ether]

some consistency between launches as the order of the mods in the list (and thereby the order the event is fired between buses) isn't guaranteed nor consistent

I would expect players to.. expect that if they shut down the game and launch it back up the category that was the 8th isn't now the last

[keen cloak]

Fair enough

[runic crypt]

Sounds great 👍

[ember lintel]

I think it's time to start a hackmd for those ideas so they don't get forgotten until they can get implemented on the porting branch

[surreal wind]

OpenGL 3.3 is now required to play the game, up from 3.2. This does not change our posted Minimum Requirements

Developer’s Note: We do not expect the update to OpenGL version to affect many people, if any at all. Our posted Minimum Requirements has been OpenGL 4.0 since 2017, and OpenGL 3.3 turned 15 years old recently.

I guess that makes a lot of people very happy

The hitbox is now 0.49 blocks
copper golems could now cross through vertical slab holes if vanilla had them

[ember lintel]

[ashen bison]

What the hell is the context for this tf 😭

[ember lintel]

no context just funny method

[plain zenith]

Also there is an unused MinecraftApi.builder() method

for the options for the debug flags it is everything starting with DEBUG including debugGenerateSquareTerrainWithoutNoise and COMMAND_STACK_TRACES in SharedConstants

which is 78 boolean properties

[surreal patrol]

So many changes to document...

[plain zenith]

the debug flag PANORAMA_SCREENSHOT allows you to take a PanoramixScreenshot by pressing ctrl+f2

weighedsoundevents = new WeighedSoundEvents(resourcelocation, "FOR THE DEBUG!");

[plain zenith]

and 2 int properties for simulated connection lag

[fading dew]

Oh, that’s neat.

[orchid adder]

-# this won't get confusing /j

[winter bear]

that's not for any of the dev only commands neo reenables anyways, since that specifically says "removed during compilation", right?

[orchid adder]

yea

[fringe summit]

Create issues on GitHub not some MD file that will be forgotten

[winter bear]

ok, so, what is it

what does snowman say?

[ember lintel]

We never used issues for this as it would pollute the actual issues and feature requests for current versions, since I joined the porting team we used md files or a pinned message here to keep a record of todos that are not in code. Let's not change things for changes sake.

[fringe summit]

We're only going to add new APIs once it is on the public repo

[mellow briar]

in the past we have added APIs on the porting branches as well

however, it is preferable to add APIs by PRs on the public repo, simply to ensure that they are of sufficient quality

at the very least, make sure that you get someone else to review it

(e.g. XFact)

[ember lintel]

For big new apis this makes sense, but for small things like a single event it is a hinderance since it makes no sense making a public branch before pre1

[mellow briar]

it makes a lot of sense to make a public branch

see also: https://neoforged.net/news/neoforge-snapshots/

it will allow the release to be of higher quality

[ember lintel]

No because then you loose all history on the kits branch

[runic crypt]

huh

[mellow briar]

That's hardly a concern. Getting something in the hands of people who can give us feedback is much more valuable

[ember lintel]

No?! Having history makes porting in many cases way easier. And for me outweighs the little feedback we would get from an earlier porting branch.

[mellow briar]

We have published snapshot versions in the past, and should do it again in the future. I invite you to re-read the blog post. If you want this to change feel free to start a wider team discussion.

As far as I am concerned, the port would already have been made public if I had been contributing to this round. There are a lot of rendering changes that it would be good to double-check

[cursive nimbus]

people in the background begin chanting Main-tai-ner Vote

[ember lintel]

There is no mention when it should happen!

[mellow briar]

In reality, only the initial phase of a port needs to be private, as we work with the raw Minecraft sources to fix the patches that could not be applied. Once we are done with porting these patch rejects and fixing compilation issues, we will now publish the code for snapshot ports on the main NeoForge repository, on a separate branch.

[zenith bolt]

AFAIK The current process is that it is private untill porting is done and the version is working.

Which means that that point is clearly defined

AKA All our tests are passing

[ember lintel]

I have made use of the history in this round multiple times so squashing and recreating each time would make this way slower and more tedious

[zenith bolt]

At least those that did not require an API rework

@ember lintel Are you at the point where the tests are passing?

[ember lintel]

Well each snapshot breaks something

[zenith bolt]

That is fine

[ember lintel]

Which is why I said pre1 is the best point to make it public

[zenith bolt]

But when it is running for a given branch, that is the point where somebody is allowed to fork from your working branch, squash and publish

You are not required to depend on the changes made in that public branch

As it is not the main branch for the source version

Any changes made coming from discussions on that snapshot branch you can ignore

As there is no guarantee that the assumptions made for those changes on that snapshot hold the next week

They form solely a suggestion as to how to implement the actuall API when we go to a release for the full new version

And because porting is a lot easier on unsquashed history

[ember lintel]

That's not how tech wants it, tech said we should use the create kits branch action from the public branch each time and add squashed commits on top of the already published ones

[zenith bolt]

That is not the going policy

And if he wants to change that

He needs to trigger a maintainer vote

[mellow briar]

I did not say it should be done each time, I said it should be the preferred approach once a snapshot is ready

we have done it many times in the past

[zenith bolt]

Actually we have not...

The amount of snapshot builds we have published is a tiny fraction of the amount of ports we have done

[mellow briar]

sure, we have not done it every time

[zenith bolt]

We have done it rarely

[mellow briar]

it depends on availability, state of the port, etc

[zenith bolt]

From a cursory glance on the maven, there is maybe a handfull of snapshots where we even have release a pre-release version

Let alone followed this procedure

[mellow briar]

they never got published on the maven

[zenith bolt]

Even worse then

I will check github just to be sure

[mellow briar]

we simply pushed the port to a public branch and let PR publishing handle it

[keen cloak]

That's very intentional. They are published through the PR publishing system so they don't pollute the primary maven repo

[zenith bolt]

There are exactly 5 porting branches

For snapshots

None of this year

[mellow briar]

we used the port/1.21.x branches for snapshots as well

I was there and I did it myself, I can guarantee you that we did

[zenith bolt]

Okey, there is no way to track that, nor its efficiency.

I don't doubt that it happened on some branches

And I don't doubt that it works better for some people

Or some releases

[keen cloak]

Uh, what?? 1.21.6 was done this way: https://github.com/neoforged/NeoForge/pull/2297

[zenith bolt]

That is the RC

Not the current situation

Where we are explicitly talking about Snapshot builds

For RC; that is more reasonable

Because the stability is significantly higher

[mellow briar]

it got force-pushed, you can't really tell from the PR commits

[zenith bolt]

To be an objective bystander here: Sorry but without history I can't review the benefits or downsides of this approach

I can see the benefits for RCS

But not for snapshots

[mellow briar]

then honestly: don't say anything if you don't know the history

[zenith bolt]

Sorry?

As Steering Council it is my job to point out that you are trying to modify standing policy, whether you believe it or not

It is not your place or job to tell me when to stay out of something as a maintainer

Let alone in such wording

[mellow briar]

it seems that you are not aware of the standing policy yourself

[zenith bolt]

I double checked and tripple checked my values, there is no obvious effective history that this is standing policy.

You applied these procedures on your own, with your own decisions and recognicance.

That does not make them a procedure.

You are free to follow these procedures and ideas when you do the porting

[cyan ether]

that's not mandatory, we've pushed ports to the public with failing tests and with rejects still in the repo

[ember lintel]

just because you didn't follow standing policy once or twice doesn't make that the new policy

[mellow briar]

I am pretty sure that XFact, shartte and myself have all pushed snapshot builds to the public; and the blog post makes it crystal clear that this was the intention, in case you are looking for a public record

[zenith bolt]

As schurli is currently doing the porting, he is free to follow the standing systems, guidelines and policies in place before you

[fringe summit]

yup we have

public as soon as feasible and we don't expect we have to redo the entire port over

[zenith bolt]

Again whether you did push or did not push, does not matter, we have a standing policy that we make it public when feasible and when we do not expect to do the port over

Exactly shartte

And when that is

Is up to the porter in question

[fringe summit]

Orion please. Standing guidelines you are pulling out of thin air. Point to the governance if you want to call it guidelines. Otherwise don't misrepresent "tradition"

[zenith bolt]

If you want ot change that policy

[mellow briar]

it's up to whoever wants to push to the public really

[zenith bolt]

It is up to the porter in question, because you are making there work significantly harder

[mellow briar]

that seems wildly exaggerated

[fringe summit]

"porter in question". Is that title assigned by vote 😄

Or are you also inventing that on the spot

[keen cloak]

The porter??? There is no single porter

[zenith bolt]

No, but they are currently doing the work, maybe porter(s), is theb etter statement, but right now, it is a single person as far as I can see

Or did mutliple people contribute already to the branch

?

[fringe summit]

You really haven't ported in a while 😅

It's a free for all

[keen cloak]

Yes, I've been doing most of the client-side crap

[zenith bolt]

Okey, then the question becomes simple

Based on our standing "Traditions"

[ember lintel]

well technically the porters for this version are you and me

[zenith bolt]

Do all porters involved, or at least a majority of them, feel that the branch is ready, and that there are no significant changes to our API surface to beexpected?

[fringe summit]

Sorry Orion, you can ask whatever you want. But stop acting like there's a procedure you're following here and not just freestyling.

[zenith bolt]

There are also comments by tech and sharte

[keen cloak]

That question is contradictory with the whole point of going public which is to allow making API adjustments in public instead of hiding them in the squashed primary porting commit

[ember lintel]

only XFact, coehlrich and I have committed to this port

[zenith bolt]

Going public has to be balanced against being decently portable and maintainable, that is why we don't just go public right away. That has not changed, not since the MCF days.

Sure we want to go public as fast as possible, and with NF fork and the work we have done, we are able to do it faster.

But the process of porting, and managing the go public of a porting branch has not altered a bit. Just when we can and should do it has changed.

[zenith bolt]

Then decide between you two, when you think you are ready.

Wait if you want, or don't.

But that is up to you

[mellow briar]

what? there is no mandate for anything to be decided

[fringe summit]

Orion, there is no capture the flag system for porting. Don't make stuff up. Anyone can contribute at any point.

[mellow briar]

your role as SC is to trigger whole team votes when needed; nothing more

[ember lintel]

since there are still quite significant changes each snapshot I would vote against going public yet

[fringe summit]

The only reason Kits even exists is because of the convenience of expanded sources

[fringe summit]

And that is the actual reason that has to be weighed

Not our APIs

The patch reject work is the only factor for Kits

[ember lintel]

he is still also a maintainer and a moderator

[zenith bolt]

Exactly

I have just as much say in this as anybody else

The only thing I can do, is trigger a maintainer vote, or better said, point out that you need to trigger a maintainer vote when I think you missed the mark and did not do so when it was purtinent of doing so.

And I can put that hat on and off of my head when I deam it necessary and in the middle of a discussion.

[mellow briar]

the point is that you are making up rules that do not match the actual practice of what we have been doing in the 1+ last year

which is... not great 😕

[fringe summit]

And try to do so from a point of authority, which you don't have.

Seriously. Stop.

[ember lintel]

he never said rule tho

[fringe summit]

Policy

[ember lintel]

policy doesn't have to be formal tho

[fringe summit]

I mean.... Dictionary definition:

a set of ideas or a plan of what to do in particular situations that has been agreed to officially by a group of people, a business organization, a government, or a political party:

That sounds formal.

[ember lintel]

but it doesn't say written, it just has to be agreed...
but enough of that topic, the changes in the snapshots are still way too big for going public

[surreal wind]

IMO, and this is just my opinion not an indication of any policy, written or not

if we have something that can do round-trip patching, and can build successfully and can be set up in userdev successfully, then there's no reason not to publish, since publishing comes to the advantage of modders that want to start coding, and isdone in a clearly separate place so end users are not confused into thinking they should be playing in those versions.

however from my limited understanding we usually only have one porting branch per version? that is, we go internal -> porting branch -> release. sorry if this isn't the case,

if we published, that would mean we have a porting branch that, potentially, ends up having to go back to internal dev for porting to newer snapshots, which means the newer snapshot ends up being developed from a non-internal state as if it was the first snapshot of a series, which feels weird in a way? maybe just me, though.

[fringe summit]

No we'd just do a "normal" update cycle on the porting branch for each subsequent snapshot / pre / rc and then release

Which means someone checks it out, bumps the neoform, runs with the updating flag and fixes the rejects (if any)

And there's a certain tradeoff there in that if large changes are still expected that you can't do in a single sitting, so to speak

[surreal wind]

yeah doing two update cycles in the same .. update cycle, feels weird to me

not wrong or anything, just weird

[fringe summit]

¯_(ツ)_/¯

It depends on the size of what changes

[ember lintel]

Do we already use unpick?

[fringe summit]

JST supports it now, but I don't think it has been wired up to the point of working, @cyan ether ?

[ember lintel]

I would like to add that in this version if possible

[fringe summit]

It would be VERY nice

I think that's also something for post-squash though

and requires buildSrc changes

[ember lintel]

it will fuck up nearly all our patches tho so won't be easy

[fringe summit]

I have some ideas, but we'll see what maty thinks

[zenith bolt]

It will yeah, but I don't think that it should be done in private

I think it is a good idea to do that in public

because it does not really need to be private

[keen cloak]

Earth left a bunch of comments on the JST unpick PR so it still needs some adjustments

[fringe summit]

Yeah I saw, those seemed to be mostly for correctness of the generated code, as far as I could tell. (I haven't gone deep on Unpick yet)

[keen cloak]

Yeah, there's no benefit to it, it's just busywork, it doesn't have any actual functional impact on the patches, that can wait until later

[fringe summit]

I'd like to have one commit that essentially updates our patches from one format to the other if we can manage that

[zenith bolt]

IMHO That is also the argument to do it in private

[fringe summit]

(And keep that commit)

[zenith bolt]

I don't think it effects the public at all

Nor does it effect the private porting

[fringe summit]

If it's private it'd be squashed in with a port

[zenith bolt]

So in reality it likely does not matter

[fringe summit]

And that's bad

[zenith bolt]

I don't understand why that is bad?

[fringe summit]

You could not easily tell if a patch change was made due to the port or a port mistake

or whether it is a mistake in the one-time unpick port of our patches

• you'd squash in the required buildSrc changes alongside that as well

Unpick generally should be completely independent of a version port from a technical standpoint

[zenith bolt]

Yeah but that argument could be helt for any of the changes made during porting, including the initial port. So yeah I think that is a valid point, it simply don't see that as bad, because we have the history on kits and we can check.

I think it should be done in public

Because it does not need to be private

But that might mean that we need to do it late

Which might delay the release

Granted I doubt this would even need to be done during the port

As it should not be breaking

Right?

[keen cloak]

Yeah, there should be no way for unpick to break anything (assuming of course the unpick applicator isn't broken)

[zenith bolt]

Okey so this is kind of besides the point then

Because we can do it quite literally when ever

It would be nice to do it at a release

[fringe summit]

yeah kinda, the idea is that neoform compilation should turn it back into what it was before it was unpicked (since unpick is just "restoring" the pre-compilation state)

[zenith bolt]

Yep

[cyan ether]

my intent was to have unpick as part of userdev and provided by parchment. if we do it as part of neodev we end up in the same issue as with parchment where we version-lock people into the unpick definition except it's not really reversible unlike parchment

[fringe summit]

Sad.

[mellow briar]

In any case I would consider going through Earth's comments a blocker for any unpick support 😄

Otherwise there's no incentive to ever look at them hehe

[fringe summit]

Sure, yeah

[surreal patrol]

Primer update probably won't come until the weekend

Too much debug stuff to document

And all the fun particle changes

And whatever else is past the network package

[keen cloak]

@ember lintel please revert your texture atlas fix, I already commited that (not sure why git let you just push that) and the event would fire twice now

[ember lintel]

you put it in the wrong place tho

[keen cloak]

No

[ember lintel]

it should be outside of the else block

[keen cloak]

It makes no difference whether it's inside or outside the else block, the corresponding if throws so it's only reachable from the else block

[fervent maple]

https://youtu.be/Bu1O2wH0BCg

[thorn sable]

https://www.youtube.com/watch?v=Bu1O2wH0BCg

[thorn sable]

you sniped me

[crimson ridge]

😮 Debug functions are no longer stripped? Today is a good snapshot day

[keen cloak]

I've re-applied the ServerPlayerGameMode reject as-is for now but I'm not 100% sure it's sufficient for the intended purpose anymore due to this change:

[mellow briar]

did the texture atlas get fixed?

not clear from #neoforge-private 😛

[keen cloak]

Yes, Schurli pushed his commit away

[mellow briar]

ahah ok

the force-push diff showed no difference

[ember lintel]

I'm fixing the compile errors rn

I have no idea how to fix the error in NeoForgeRenderTypes

[keen cloak]

Nuke

[ember lintel]

so just remove TRANSLUCENT_ON_PARTICLES_TARGET(() -> getTranslucentParticlesTarget(TextureAtlas.LOCATION_BLOCKS))

[keen cloak]

Yes

[ember lintel]

so now someone needs to fix the creative inventory

[surreal wind]

did they remove the mandatory "NPC" subtext?

[keen cloak]

what happened to the creative inv?

[winter bear]

aiui, no but it disappears earlier

As in at a shorter distance

[winter bear]

I'd still like one example of one of those functions and I don't really want to set up snowblower myself

[ember lintel]

wait... shrimp lives

[keen cloak]

Fake network latency, AI and pathfinding debug renderers, wireframe rendering for chunk geometry, etc.

[ember lintel]

I can't take items out of it

[keen cloak]

Seems to apply to all inventories

[thorn sable]

pshh inventories are for storing stuff anyways, who needs to take stuff out of them

[thorn sable]

if this is cause unsure how to use snowblower, the readme was recently updated to include basic usages
https://github.com/neoforged/snowblower?tab=readme-ov-file#how-to-use
if its more due to snowblower being slow, thats mostly the case when generating many mc versions
and the process time can be trimmed down by limiting which versions snowblower will process
-# https://github.com/neoforged/snowblower?tab=readme-ov-file#processing-specific-versions

-# totally not been waiting for some excuse to drop updated readme link in here

[winter bear]

oh, I saw a video about that from someone claiming that there's enough in the code to recreate that
and them showing recreation of some of the things

[winter bear]

no, it's just because I don't wanna go through it atm

I've done it before and it's fine, just don't wanna put in the effort atm

[thorn sable]

yeah mods have been recreating these removed debug renderers for quite a while now
https://modrinth.com/mod/debugutils
but i suppose these kinds of mods are ded now that vanilla is providing them

[keen cloak]

Fixed

I fucked up re-applying a reject in the first or second snapshot we ported to

[ember lintel]

second, in the first one I played a bit in creative

[winter bear]

yeah, I took that video with a grain of salt tho, since I've never heard of them before that

I could have looked into the source code, but I didn't wanna bother

[gritty swan]

Meanwhile, back on snapshot changes...

I'm def gonna crack ServerEntity.Synchronizer open with an AT immediately. Gimme movement and updates from server kthx

[thorn sable]

gander live entity rendering

[gritty swan]

yes

[thorn sable]

need that compact machine renderer

[gritty swan]

yes, yes I do

I wanna watch people running around in tiny boxes

[thorn sable]

me too

[gritty swan]

Like ants

[thorn sable]

• shrink mod compat, shrink me down and i walk into my machine

[gritty swan]

Imagine if I actually added that. A portal item that enables when you have Shrink installed

[thorn sable]

be even crazier if could interact with machine insides from overworld
but that feels... not impossible but yeah

but anywwhos back to the snapshot... debug flags go brrrr

[gritty swan]

DEBUG_ONLY_GENERATE_HALF_THE_WORLD

Also GameRules.RULE_SPAWNER_BLOCKS_ENABLED

[winter bear]

like checkerboard pattern with void chunks?

[gritty swan]

I think it's more vertical

[thorn sable]

yeah i would imagine deepslate and lower is yeeted when thats on or something

[winter bear]

😔 checkerboard would be funny

[gritty swan]

Go install Mystcraft or RF Tools DIms

./s

[winter bear]

yeah, rf tools dimensions is fun

once made a world filled with awakened draconium

then a rf tools builder + crafting setup for infinite awakened draconium ingots

[keen cloak]

It's a vertical cut, not a horizontal one. When enabled, it only generates terrain where Z<0

[shrewd dirge]

super useful for worldgen https://media.discordapp.net/attachments/232644586638540811/614180779848892427/unknown.png?ex=68c1d2ec&is=68c0816c&hm=08a5ad9e0c1097c348e1f0550d2b039c6f5e691e29de57b7bcb330459f3f76f0&=&format=webp&quality=lossless

[fierce badger]

what

[runic crypt]

Whoa

[crimson ridge]

barely

[ember lintel]

holy crap the debug renderers are powerful https://www.youtube.com/watch?v=KDxVj8iMgag

(just don't turn all on at the same time)

[meager marsh]

weren't they available in code already, just had no menu to toggle them?

[surreal patrol]

No, the debug constants were stripped out before this version

There were some lingering references though

[meager marsh]

https://modrinth.com/mod/debug-renderers

okey that's what i remember

[warm badge]

I’m guessing cause before they just returned false, so the jvm removed the code pieces?

[ember lintel]

I think before they used static flags or a preprocessor to set them, now they use jvm attributes

[keen cloak]

Not the JVM, the compiler did

[warm badge]

oh yea, I forgot if it was the compiler or jvm

[fierce badger]

would it be possible for neo to make the debug easier to enable, like toggling it dynamically or not having to enter the flag

cuz then u have to reload gradle and all that when u wanna change it

[runic crypt]

Probably could just make it a Gradle thing you can flip on/off

[winter bear]

or add a way to regen run configs without a full gradle reload like we used to have

[surreal patrol]

The latter seems more reasonable unless we have a bulk way of patching static constant fields before mod loading, at least at a glance

[broken nest]

not rc yet?

[runic crypt]

Nope

[modest pasture]

There is even no pre releases

[ember lintel]

Judging by how many big changes they are still pushing out I'd be surprised if we see a pre1 before MC Live

[surreal patrol]

@thorn sable Have your primer: https://github.com/neoforged/.github/pull/29/commits/a0533a25368bd00de077b8d506035db70497dadb

Basically, the debug stuff, some changes to the particle submissions, and of course the key mapping categories

[fringe summit]

If I may offer a suggestion: I'd not list removed things if there's no explanation of an alternative since there's little benefit

If it's something a compiler would immediately flag as an error

[surreal patrol]

I mean, I document everything mainly because I usually have to go back and retroactively update where things changed, and there are many times between snapshots where things are removed, then not removed, or between a state of flux as some systems get updated

[fringe summit]

Yes, I just think the info something has been removed (in case of methods/classes) is not useful in isolation. Different story if we know and document a replacement.
The compiler will tell you quite happily that something is gone very quickly 😄

If someone just wanted an overview of what was yanked (in a technical sense) that is something that can be generated from the mappings (half true given the number of unobfuscated symbols)

[surreal patrol]

Sure. To be frank, I don't think most of the additions, changes, and removals in the classes list I document are useful in isolation

You could probably get by just removing the lists, though I still do it as I believe it has some value, even if it's not much

[surreal wind]

I think it is good that there's removals because people can go crazy wondering if it was forgotten in the changes list

[surreal patrol]

The only thing I don't specifically document are private or package-private methods unless I believe they are important to actually making the feature work

[surreal wind]

having them explicitly is just .. "ah fuck I guess I can't do it this way anymore"

[fringe summit]

what?

you compile your mod

the compiler tells you

[surreal wind]

yes

it tells you that it was removed

but it's goodto know if it was just removed

or removed with alternatives

[fringe summit]

Yes, which is why you list alternatives if there are any 😛

[surreal patrol]

I try, though there are probably quite a few times where I just miss the replacements or document the add and removal stuff separately

[fringe summit]

Even if it is listed as removed. If the replacement is non-obvious or more involved, that doesn't mean there isn't one.

But hey whatever, I said my piece 😄

[wide vine]

Removals is decent, especially if porting multiple versions at once

So then you don’t have to keep searching the other primers as you know it is gone

[gloomy sable]

not having removals means theres a lot more abiguity in the meaning of 'this change isnt in the primer', whereas having removals (even without alternatives) removes that ambiguity

[ember lintel]

It is frog day

[runic crypt]

🐸

Ribbit

https://tenor.com/view/frog-ribbit-smiling-animated-gif-12261731

[mental saddle]

Been meaning to ask, why is it called Frog day?

[inland yew]

[surreal wind]

I don't know the history but mojang devs have, for a very long time, used pictures of frogs on their twitter accounts to indicate that a snapshot is building and will be released soon

[mental saddle]

Huh. Ok thanks

[modest pasture]

Because initially snapshots were on Wednesdays and there was a meme called "It's Wednesday my dudes"

[surreal wind]

does the meme predate the mc snapshots? I have only seen it in their snapshot messages

[modest pasture]

No, but it got popular in community at the time

Here is a version of that meme from that time

I think this frog on the picture is from Quark if I recall correctly

It was somewhere 1.15-1.17 snapshots era

[mental saddle]

Oh that makes sense. I’ve definitely seen this one before without the Minecraft context.

[surreal wind]

https://knowyourmeme.com/memes/it-is-wednesday-my-dudes

says it appeared in 2014 on tumblr

smh this page still hasn't accepted that Vine is dead

[tiny wing]

it used to be even on a friday, very long ago

back then MS had not bought mojang yet

https://web.archive.org/web/20130126020104/http://www.mojang.com/2013/01/minecraft-snapshot-13w04a/

and on a Thursday apparently

[vernal sinew]

New version detected: 1.21.9-pre1.

[cyan ether]

@edgy garden

damn it

<@&1067092163520909374>

[tiny wing]

damn

[ashen bison]

Lol got outta school in time

[cyan ether]

sigh, 1.21.9

screams internally

[tiny wing]

https://tenor.com/view/and-so-it-begins-again-moira-rose-moira-catherine-ohara-schitts-creek-gif-20290450

[thorn sable]

👀

[smoky jackal]

I'm pretty sure the memes came from wintergatan's marble machine who also had wednesday updates at one point

[ashen bison]

Wait is it 21.9 pr?

[broken nest]

aw

is it time to rebuild my mod for 1.21.9

[ashen bison]

I love this 🔥😭

[eager dome]

wait can players have texture pack skins now?

[ashen bison]

Thought they did since 21.1

[vagrant tree]

https://www.minecraft.net/en-us/article/minecraft-1-21-9-pre-release-1

[thorn sable]

1.21.9 - Pre Release 1

• Primer: https://github.com/ChampionAsh5357/neoforged-github/blob/update/219-or-22/primers/README.md
• Article: https://www.minecraft.net/en-us/article/minecraft-1-21-9-pre-release-1
• Changelog: https://misode.github.io/versions/?id=1.21.9-pre1
• Notion: https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2704f070f888808abae6c8586904736a&pm=c
• SnowMan: https://github.com/neoforged/Snowman/commit/d2a70ff1cc5b80b0e6c94a702212661610694757

SlicedLimes Videos:

• Main: https://www.youtube.com/watch?v=RjbXy0vek50
• Pack: https://www.youtube.com/watch?v=N7LPEbhNCtk

[vagrant tree]

No more notification namespace

Soon enough they'll make it so you can remove the overworld with datapacks at this point

[vernal sinew]

Minecraft 1.21.9 Pre-Release 1

The copper age is approaching and it's time for this drop's first pre-release. Today's release includes more copper golem improvements, some changes to the mannequins and several bug fixes. Pre-releases don't follow the regular Tuesday release schedule, so...

Sep 16, 2025, 2:01 PM

[vagrant tree]

(currently if you try to do that, the server will refuse to start)

[surreal patrol]

Yep, knew it

[shrewd dirge]

oh great, that's one more mod I don't have to maintain

err

The setworldspawn command is no longer limited to the overworld

not yeeting the overworld

Server commands now run from the respawn dimension by default instead of always the overworld
did they ever fix weather commands

[surreal wind]

oh shit pre-1

neat.

so we have proper statues now

would be even better if pose allowed an array of angles though

[thorn sable]

The setworldspawn command is no longer limited to the overworld

• This means players can respawn e.g. in the Nether when they have no bed or respawn anchor
  @gritty swan dont mind me just gonna go make my compact machine a jail

[sturdy wave]

There is a funny interaction with this where setting the spawn in the end will get you stuck there :)

[thorn sable]

oh mojangsta mentioned the end, end update confirmed /s /j

[ember lintel]

oh right because the portal respawnes you... fun

json defined portals when?

[winter bear]

The Data Pack version is now 87.1

[shrewd dirge]

you can do this with command blocks

[ember lintel]

not if you want tp delay like with nether portals

[shrewd dirge]

you can do this with command blocks and a funny redstone machine encased in bedrock

[fringe summit]

ugh fuck now we're under time pressure again

[smoky jackal]

https://www.curseforge.com/minecraft/modpacks/compact-claustrophobia

[ember lintel]

a simple json like

{
  "type": "minecraft:standing",
  "source_dimension": "minecraft:overworld",
  "destination_dimension": "minecraft:the_nether",
  "frame_material": "minecraft:obsidian"
}

[smoky jackal]

wouldn't that need json defined dimension first to be really interesting (unless they are already a thing ?)

[ember lintel]

dimensions are json defined

huh

[ashen bison]

sure ig 😅

[ember lintel]

ohno the builders for the rpc stuff directly use a Registry.registerForHolder call

[fading dew]

You ought to be able to do it easily enough without that; just increment a scoreboard entry/data value for the entity when in the portal, reset to zero when out of the portal, and teleport when the value reaches a certain number. I think I've done something similar back in my datapack/command days.

[ember lintel]

ewww, anything that requires a scoreboard to do some functionality that isn't score related is an ugly hack and should not be the intended use

[keen cloak]

They both have a separate build method. Deprecate the register methods, AT the build method and you're golden

[ember lintel]

yea probably the only way

[orchid adder]

has Mojang made the customary tree refactors this drop yet

[keen cloak]

Also, it looks they the restriction of holders having to be typed to a given registry's root type (i.e. the reason why DeferredHolder has two generics) might be getting lifted

[ember lintel]

probably keeping them for the next major bump

[ember lintel]

that's only the registration side, I think the bigger problems were the usages where it expected Holder<T> instead of Holder<? extends T>

[winter bear]

When Water Drop?

[orchid adder]

close enough, welcome back Update Aquatic /j

[jagged sable]

Update Aquatic 2: Electric Boogaloo

[keen cloak]

Welp, this is going to be interesting. Basically everything in level rendering is two-stage now (extract first, render later), including block breaking, block outline, sky and weather. This means that RenderHighlightEvent.Block and our DimensionSpecialEffects extensions are getting bonked pretty hard

[gritty swan]

Wait you can set a different dim as world spawn now? Yeeee, that makes my compact world spawn feature possible again

[gritty swan]

oh. Jail room upgrades were already planned lol

I've talked about them a couple of times

[sleek shore]

how can a feature not be possible when you are free to mod however you want ?

damn didn't see the ping thingy 😔

[winter bear]

possible, but more work than it's worth

[lavish herald]

What about https://github.com/neoforged/NeoForge/blob/1.21.x/src/client/java/net/neoforged/neoforge/client/event/RenderLevelStageEvent.java?

[keen cloak]

That should be fine for now, at least in the last snapshots I could integrate it in a functional way. Might need some changes down the line regardless but it's not broken like the other two I mentioned

[ashen bison]

would be a good time for public snapshot rn

[surreal patrol]

@thorn sable Have the primer: https://github.com/neoforged/.github/pull/29/commits/521e578a82bf001753dda51853c521b4972fb180

Mostly some more cleanup, render states, respawn data, and client asset specification

[thorn sable]

Primer before sliced lime uploads, noice

[surreal patrol]

Well, most of today I spent setting up my new desktop and getting stuck for an hour over a kernel bug

Ignore the fact that I have work

[sturdy wave]

slicedlime hasn't even started working on the video today, so no competition :p

[static dagger]

Are you guys having a company grill party?

[shrewd dirge]

grilled frogs

[fierce badger]

theyre filming

[thorn sable]

Oh that’s good to hear, I can stop auto refreshing his YouTube page xD

[sturdy wave]

Definitely not in this rain at the moment 😬

[sturdy wave]

if you join his Discord server you get all the info

(*when videos are delayed or come out...)

[ember lintel]

The ItemStack reject is a bit weird, why was this patched in the first place?

++++ REJECTED HUNK: 10
@@ -723,7 +754,7 @@
 
     @Nullable
     public <T> T set(DataComponentType<T> p_331064_, @Nullable T p_330775_) {
-        return this.components.set(p_331064_, p_330775_);
+        return (T) this.components.set(p_331064_, p_330775_);
     }
 
     public <T> void copyFrom(DataComponentType<T> p_397950_, DataComponentGetter p_397631_) {
++++ END HUNK

[shrewd dirge]

javac/ecj incompatibility maybe?

[ember lintel]

but why would that patch be in neoforge?

[shrewd dirge]

can't do neodev if the vanilla sources don't compile

[ember lintel]

the compile fixes for that should be in neoform

[cyan ether]

https://github.com/neoforged/Kits/commit/d1a81a3d561d45801ce0674d71b22cd697975679 ask yourself :P

[ember lintel]

oh they changed the generic again

[fading dew]

It's a datapack solution for a mod problem. Of course it's an ugly hack. You could also use data stored on the entity, couldn't you? I can't remember how that works.

[ember lintel]

ok only the renderer package left...

[keen cloak]

I'm looking at that right now

[fringe summit]

Any help needed?