26.1 Snapshots

[inner creek]

which is making me slightly hyperactive, but doesn't really remove all of the brainfog

invalidate caches didn't fix it

[plucky hollow]

question you did download java 25 and update to gradle 9.1.0? i had to do that for my builds to import

[inner creek]

OH java 25

that might do it

I knew I was missing something lol

ugh something reset in intellij and the Terminal got set back to powershell

I find powershell extremely awkward to use for my basic level of commandline usage

[twin sun]

Thanks for the info 😄 I haven't even started merging Arch Loom with any of the Loom versions that support unobfuscated Minecraft, so the testing will definitely have to wait a while

[paper cedar]

yeah that should work

[inner creek]

yeah except I used java 21

[paper cedar]

Hm, that's an excellt question if you actually have to (?) Since generating the MC artifacts is based on NeoForm still for our part 😅

[inner creek]

now it worked after switching the toolchain to 25

[paper cedar]

Oh, right... 😄 Oooooooh yes, then Gradle will also complain heavily that it can't resolve a J21 version of NeoForge

[inner creek]

it didn't actually!

it just failed to find, without explanation

[paper cedar]

That is super weird, I remember it doing that for me when I ran an actual task

[inner creek]

maybe gradle version dependant?

[paper cedar]

I mean that paragraph-long infodump about variant selection blah blah yada yada

[inner creek]

was using 8.9 in this folder

[paper cedar]

Hm, that should have had that long-form error already. weird

It's possible that it only shows when you try to run a task like assemble, and not on sync

[inner creek]

LITERALLY the only code change lol

[plucky hollow]

thats jspecify right

[paper cedar]

Yeah, if you don't touch villager trades, there's very few changes (which I suppose is good)

[inner creek]

no the class is defaulting to jsr305 which I had already used in the far past

[plucky hollow]

i spent far too long going over my mods being sure im using those over jetbrains
(rly need something like what neo has to yell at me in moddev)

[twin sun]

Loom still expects mappings in a ton of places (you can provide empty ones, primary meant for legacy versions without mappings, but I expect that'll break a ton of things)

modmuss and contributors did a ton of work addressing that but it'll be quite a pain to integrate into our fork, and I'm not 100% sure if we still want to port the fork to unobf versions in the first place

[inner creek]

actually wait

the jetbrains ones are still in the classpath

that error was bogus from before I set the java25

[paper cedar]

Oh, you mean you'd do a breaking change or just abandon the fork? Or upstream the changes?

[inner creek]

so 0 code changes woo!

sadly, that doesn't mean I can flag the existing build as compatible

because.

I should probably make it upper-unbounded

since we don't know when mods will be likely to be incompatible anymore

[paper cedar]

"anymore" = you never really did 😛

[inner creek]

also that should have been .11

cos I use ResourceLocation Identifier

[inner creek]

in the past it was generally a good rule of thumb that 1.x -> 1.(x+1) was going to have enough breaking changes to not warrant leaving it unbounded

this stopped being true with 1.19.3+

[twin sun]

I don't think it needs any breaking changes, but yeah I'm not sure if we'll abandon the fork for new MC versions or not.

The primary features of AL over the normal Neo toolchain are mapping support, which we'll drop for 26.1 Neo in any case as it's such a burden to maintain especially now, and binpatching which I think might be coming to MDG?

[paper cedar]

Is already in, yes

[inner creek]

ah, a conundrum....

if I build the mod again with 1.21.11, I can flag it both 1.21.11 and 26.1-snapshot-1

but if I build with 26.1, it will be java 25

but there are literally no code changes

so do I re-release 1.21.11 just to update the mods.toml

or do I look to see if I can use java 25 features in this mod, so I can justify it being a new file 🤔

[plucky hollow]

that should be fine id say, launchers should handle the java for you
but you can always set the java version in mods toml
https://docs.neoforged.net/docs/gettingstarted/modfiles#features
althought ive never actually seen anyone do this

[paper cedar]

Don't do that

That is super pointless

[plucky hollow]

no?

[paper cedar]

no one downloads mods that require you to manually bump your java version up for your instance

[inner creek]

well I don't want to force java 25 on 1.21.11 players

[paper cedar]

I'd just not bother with cross-version compat like that

[inner creek]

so if I release the file built with java 25 it will be literally just a java bump vs 1.21.11

but I can build with java 21 and have it flagged on CF ... except it will still just be a new file that contains no changes other than metadata

so it feels like a lose-lose situation :P

what the fuck intellij? what are these indentation changes?!

those are NESTED blocks they GO NESTED in indentation too!

[worthy mesa]

at some point this changed, but seemingly only for some classes?
I have a class where it tries to put switch expression cases inline with the switch, but only for that class

[lusty thistle]

Hey! I just wanted to try the MDK for 26.1 using tthe provided commands in the PR:

mkdir NeoForge-pr2879
cd NeoForge-pr2879
curl -L https://prmaven.neoforged.net/NeoForge/pr2879/net/neoforged/neoforge/26.1.0.0-alpha.1+snapshot-1/mdk-pr2879.zip -o mdk.zip
jar xf mdk.zip
rm mdk.zip || del mdk.zip

Without modification, I got this error message:

Execution failed for task ':prepareDataRun'.

Trying to prepare unknown run: data. Available run types: [client, clientData, serverData, gameTestServer, server, junit]

It was fixed by doing this:

        data {
-           data()
+           clientData()
+           serverData()
            programArguments.addAll '--mod', project.mod_id, '--all', '--output', file('src/generated/resources/').getAbsolutePath(), '--existing', file('src/main/resources/').getAbsolutePath()
        }

But then, I get this message:

NeoForge-pr2879:main: No matching variant of net.neoforged26.1.0.0-alpha.1+snapshot-1 was found. The consumer was configured to find a library for use during compile-time, compatible with Java 21, preferably in the form of class files, preferably optimized for standard JVMs, and its dependencies declared externally but:
Full message: https://pastee.dev/p/QH4oaZEh

At the bottom, it tell me the possible solution to be using Java 25 or newer as Gradle JVM but I did that.

OOF! I now realized that the problem was that I had to change this as well:

- java.toolchain.languageVersion = JavaLanguageVersion.of(21)
+ java.toolchain.languageVersion = JavaLanguageVersion.of(25)

Now everything works.

tl;dr

Provided MDK requires changes, see my provided diffs above. Maybe something to report to your PR publishing bot instead 😄
I think you already know, just ignore my message lol

[steady oxide]

hmm surely there's a way to make the fact that users need to upgrade from j21 to j25 more obvious with an explicit error?

[plucky hollow]

off topic but i like how the net.neoforged:neoforge:26.1.0.0-alpha.1+snapshot-1 turned the artifact id into neo emoji

[paper cedar]

Nope that's the Gradle way

[steady oxide]

I have to imagine that there is some way to re-throw a "nicer" error to the user if this error is detected. (I know that it was on a much older Gradle version and different architecture, but the ForgeGradle days didn't have much issue with explicitly telling the user where they went wrong in terms of Java version.)

[steady oxide]

You could pretty easily add a check for JDK version and error out early right here:

https://github.com/neoforged/ModDevGradle/blob/704957196a120ecc493e0137f2cd2ed24929cca6/src/main/java/net/neoforged/moddevgradle/internal/ModDevArtifactsWorkflow.java#L123

Old check for comparison: https://github.com/SizableShrimp/ForgeGradle/blob/651e43fd7c68068c835548583c5eb924e13d9251/src/userdev/java/net/minecraftforge/gradle/userdev/MinecraftUserRepo.java#L189-L197

I think it's pretty safe to assume that if a user is requesting a Minecraft artifact of a specific Java version, the Java version of their project should be at least as high as that (if it is set). And the error message from this check is a lot nicer to look at than the Gradle vomit that you get otherwise.

Also it's pretty interesting to me that NFRT optionally accepts a Java executable as a CLI argument, but uses its own javac to recompile the game. Why not instead allow passing a path to javac as well and using that as a tool call for recompilation if present? It would save the headache of having to run NFRT with different JDK versions, but only if the version capability reports strictly greater than J21. And using the actual JDK for a given -release flag is a safer bet anyways

[wintry flower]

I believe the problem is that NFRT needs at least java 21 to run

But it uses the javac api directly, not the executable

[wintry flower]

MDG doesn't parse the neoform config at configuration time

So there wouldn't be a way to read the java version from it

[steady oxide]

Yes, why not use the executable?

[wintry flower]

We could use the VesionCapabilities I suppose, but it would require an up-to-date MDG since we'd have to hardcode it there

[steady oxide]

is the version capabilities thing not sufficient (?) That is being used at configuration time from what it looks like

Ah

[wintry flower]

I'm happy to accept a PR for it

There's a solid chance that modders will bump the plugin but forget about the java version

[steady oxide]

In any case, you could likely punt the check to when the neoform config is available too. That would still be before Gradle vomits its trash to the console I would hope

I can look into it later; I think it's going to be quite a common occurrence, so giving people an obvious error is helpful

[steady oxide]

Is there a reason why it's necessary to use the api vs. calling out to the executable?

[wintry flower]

No, that would be too late. We only get the neoform config transitively from the neoforge dep 😄

[wintry flower]

No idea

https://github.com/neoforged/NeoFormRuntime/blob/main/src/main/java/net/neoforged/neoform/runtime/actions/RecompileSourcesActionWithJDK.java this is the code fyi

[steady oxide]

I saw. I guess the conditional logging of errors/warnings would be more annoying if it were an executable call

[wintry flower]

I guess. It's also not a big deal ig

[steady oxide]

Hmm. Perhaps an alternative would be listening/checking for Gradle throwing such an error (is that possible?) to the user at configuration time about mismatched Java on the requested neoforge artifact and then spitting it back out again but as a "nicer" error with more explicit language.

[wintry flower]

We could also tweak the NeoForge publication to skip the java version attribute

In any case I will include this information in the blog post, maybe that will be enough

[steady oxide]

Fair enough. Based on the comment though, it seems like an issue was already run into though with using such a new JDK paired with an old -release version?

[wintry flower]

I guess that something else required java 8 🤔

(for legacy i.e. 1.17 - 1.20.1)

[paper cedar]

We compile directly from ZIP to ZIP

Hence javac is not usable directly, we need a compiler driver shim

There has not been an issue with JDK 21 and -release 8, even for binpatches, BUT for binpatches the bytecode is now slightly different with javac 25 and -release 8

Which is why I did start this branch:

But it's not super high priority

[paper cedar]

decompiler libraries IIRC

As in, passing J9+ JDK as libraries to the old versions of fernflower fails

That's why it requires the path to a J8 JDK, to pass the libs on to ART+fernflower

[jagged ridge]

Can confirm, it happened to me when working on 1.16 NFRT support

Technically speaking Fernflower can probably run on a newer JDK as long as it has rt.jar from Java 8, but there's no point as long as there are Java 8 builds available for the host

[paper cedar]

but yeah, it should use that external JDK to actually run the javac driver, but it needs some much more complicated setup (i.e. that needs to be compiled with java8)

[paper cedar]

I jumped in a bit late. Another issue here is that the language version itself is late-configured, which means you have to do that assertion in an afterEvaluate block. Doable I suppose, but as Tech said, you'd only know once the user had updated the plugin already.

It's somewhat annoying overall, that Gradle has no real path to retrieve some config via HTTP at configuration time, and still be configuration-cache-compliant

[jagged ridge]

at risk of asking an obvious question, can you not just do it manually and manage the download caching yourself

[paper cedar]

It will detect that you're trying to use the network

Since it intercepts disk I/O you do.

[jagged ridge]

Oh wow

[paper cedar]

There are hacks via ValueSource that suppress the Disk I/O checks it does, but overall there's no clear-cut easy way

But they do instrument your entire plugin and redirect all java.io.* calls to their own shim to detect this stuff

and track dependencies to external sources for your plugin/config code

[wintry flower]

I'm not sure that's quite true

It should be compliant but slow

[paper cedar]

You can do a ValueSource, and immediately query it

[jagged ridge]

I guess that is convenient for enabling the cache with legacy code

[paper cedar]

But otherwise it'll detect your cache writes as a hard dependency on files

which would be untrue

[jagged ridge]

Forcing everyone to rewrite their stuff to use Gradle I/O APIs would be a non-starter

[paper cedar]

They would have to expose their configuration resolution code in a configuration-time friendly way

configuration resolution = dependency resolution

i.e. you'd still want it to be offline mode compliant

[steady oxide]

Hmm fair point. Although I guess it would save people for when they bump the version to J29 in ~2 years :P

[paper cedar]

They'll manage. That gradle resolution error is not that bad

😄

If you actually get to it

[hushed canopy]

Which is exactly the right approach here?

If you need something that isn't gradle-inspected I/O, then it needs to go through a VaueSource so that the reuslting value is conf-cached and so that the source is re-run on every invocation

Ideally you'd be able to route the value source around and only un-lazy it at the last possible moment but it's still useful even if you don't because it ensures that conf cache works by isolating the logic used to retrieve the value, and running that in a config-independent context every time

That said, that's awful UX in this case, cause it's basically going to lock you into non-offline-compatible mode unless you're really careful

🤷‍♀️ I mean you can probably make it work, TBH

[still elbow]

is the neoforge versioning seriously going to be just 26.1.0.x for 26.1?

[paper cedar]

... as opposed to?

[indigo venture]

there were other formats discussed but none of them presented significant enough benefits for the team to agree to change neos versioning system, so the existing one of appending a neo build no. to the mc version was kept

[paper cedar]

Yeah it's just the logical continuation of the previous versioning

In other news: I cherry-picked the missing commits from 1.21.x over to port/26.1 and will make another snapshot version release should it build correctly

[indigo venture]

oo nice. how hard is the porting of those with all the param name changes?

[paper cedar]

Well it wasn't a lot of commits and patch changes so it could be worse

I just had to patch up a single line with references to two params 😄

The params are also the easier ones, the local variables also all changed, those would be harder if the patches are actually significant

I am confused. Wasn't it previously possible to create tags in the GitHub web interface?

(Other than creating releases)

Well anyhow.... as a reminder to the other maintainers, making a snapshot release r/n is manual and requires pushing a tag with the right name format (it's just release/<version>):

To github.com:neoforged/NeoForge.git
 * [new tag]             release/26.1.0.0-alpha.2+snapshot-1 -> release/26.1.0.0-alpha.2+snapshot-1

[still elbow]

also i just realised there are now snapshot versions of neo for the general public to use it seems?

[indigo venture]

could you update the message of yours ive pinned with the latest version?

[paper cedar]

Yeah

Seems like 26.1.0.0-alpha.2+snapshot-1 published successfully

[graceful laurel]

VF seems to be choking pretty hard on the changes done for statements-before-super

[paper cedar]

Yes, it does, coeh has been PRing fixes, I believe

We patch some of these out in NeoForm

[hushed canopy]

I guess I'm just confused why it wouldn't be 26.1.x? Aren't the vanilla versions just going to be Year.Release, or have I misunderstood that?

(Basically why the extra version, when will that actually bump?)

[graceful laurel]

Year.Release[.Hotfix]

[paper cedar]

Year.DropNumber.Patch and for the first relase of the drop, they omit the patch

[hushed canopy]

Ahh there's a .patch they omit on the first, got it.

[paper cedar]

we pad it with .0 to have a consistent scheme

in case of the first release, ofc

[hushed canopy]

Yeah okay that checks out to me

Seems the most sensible approach, plus version comparisons will still work right with existing versions, luckily

So all around, could have been a lot worse

[wintry flower]

The cherry-picks didn't appear in #neoforge 🤔

Oh you cherry-picked in the wrong order 😅

[paper cedar]

hm?

[wintry flower]

Compare with 1.21.x

The commits are not in the same order

[paper cedar]

Doesn't really matter though, does it? 😄

[wintry flower]

Hmmm

Nevermind it looks correct

[paper cedar]

Even if, they're different commits

[wintry flower]

Yeah it's not very relevant

[paper cedar]

I still think it's weird there's no default trailer for Git commits pointing to the source of the cherry-pick, but oh well.

They have the PR numbers in the commit description

[opaque basalt]

seems like RenderStateExtensions class's onUpdateEntityRenderState method dosen't use anymore in current neoforge 21.6-snapshot version

[paper cedar]

Hm

[paper cedar]

Thank you for the bug report, that is indeed a porting bug

[opaque basalt]

no problem...!

[jagged ridge]

Is it expected that lambda numbers would shift between Neo 1.21.11 and Neo for 26.1 snapshot 1 if the function looks identical?

[paper cedar]

yes

javac 25

vs. javac 21

I noticed that you can't even compile with -release 8 anymore in javac25 and expect binpatches to apply. it generates different bytecode, so I'd not be surprised if it also generates lambdas differently

[jagged ridge]

Right, forgot they updated the compiler in addition to not obfuscating

[fresh gyro]

@paper cedar I checked the current PR build of the port to 26.1

NG can most likely load it once you update to the new NeoForm to have the fixed arguments in that config file.

[paper cedar]

We can make an updated release soon with the updated NF

[fresh gyro]

That is fine by me

I will wait untill then

Alternatively I can probably just push to that branch

As it should not change anything

Other then that setup thingy

[paper cedar]

You mean the NF update?

Do it

[fresh gyro]

Yeah currently checking some other things in NG7

To validate and then I will push that

Should just be a bump to .3 for NF

[paper cedar]

Uuuuuh, which branch are you looking at, Orion 😄

That's the wrong one:

[fresh gyro]

I misread

This morning .......

I reverted it back

[paper cedar]

You want port/26.1 😄

[fresh gyro]

For some reason I clicked on the target branch

I know

I read that file twice

And still misread it

FFS

[paper cedar]

the 1.21.11_unobfuscated branch is really just there as a "reference" so we can compare 26.1 against a prior version with the same naming

[fresh gyro]

Yeah

I know

There we go

My browser hang twice already today

[fresh gyro]

@paper cedar Is it normal that the PR build did not get a new version number?

[paper cedar]

No that's a bug

[fresh gyro]

Okiedokie

[paper cedar]

I noticed that too. Doesn't matter for the tagged releases, but apparently gitinfo['tag'] returned by gradleutils isn't the current tag

it's just any previous tag

[fresh gyro]

....

Okey

That has me slightly worried

[paper cedar]

Urgh yet another learning

Annotated tag crashes the release build

Lightweight tag doesn't

@fresh gyro 26.1.0.0-alpha.3+snapshot-1

[fresh gyro]

Lets see

Including downloading crap on my slow ass internet

So I would call that good 😄

[untold oyster]

[paper cedar]

git fetch tag "would clobber tag yadayadayada"

[untold oyster]

where did you see that? locally?

[paper cedar]

No, obviously not. Release builds = CI/CD 😄

https://github.com/neoforged/NeoForge/actions/runs/20554837860/job/59037554968

Deleted the tag, made a new lightweight one -> works

[untold oyster]

smh, should've linked that in the first place /j

[paper cedar]

We need to disable fetch tags for builds originating from a tag me thinks

[untold oyster]

I'm confused as to the reasoning behind https://github.com/neoforged/actions/blob/b0f0d3c1f6c6d818a253e9d2e5e7ccc0af31a1de/.github/workflows/gradle-publish.yml#L154-L160

[paper cedar]

If you're running for a tag, no point in fetching commit history

The entire thing is bodged. We set fetch-tags: true for checkout, but that doesnt actually work properly with recent git either

[untold oyster]

but we then have a separate git fetch --tags?

[paper cedar]

Hence the explicit fetch tags right below it

[untold oyster]

[paper cedar]

There's a comment on that, that was added later to fix a regression with git

IMHO that should all be ripped out

And replaced with: https://github.com/neoforged/GradleUtils/commit/12b1a7d63cf3f5a2205105a8ae3ec3e41b7aee35

[untold oyster]

but by the time you committed to it, the bug was fixed in git, no?

[paper cedar]

Comitted to what?

[untold oyster]

err, your changes to the gradle-publish workflow last week

[paper cedar]

No idea, I didn't follow up

The linked issue is still open at least

[untold oyster]

since if we look at the linked issue, the issue seems to be fixed in 2.49.0 according to a comment, and the action logs confirm that the current git version on the runner is 2.52.0

so we can eliminate the git fetch --tags

https://github.com/neoforged/NeoForge/actions/runs/20553517437/job/59034457454#step:2:68 seems to corroborate this

[paper cedar]

Well, as I said -> #1445764870564745228 message

All of that is redundant

fetch-tags and fetch-depth as well

[fresh gyro]

Do we have any example of a Fabric version for this Snapshot already?

Or is that still far away?

[paper cedar]

Uh... Fabric version?

WDYM

[fresh gyro]

Thinking about looking at their stuff to do some comparisons

So I was wondering

With my hat of to you and tech

We were very very fast

With getting a working version

[bold forge]

yes it should

[paper cedar]

I haven't actually checked. Comparing Mixins against our Patches is mostly pointless.

Did you have anything particular in mind?

[fresh gyro]

Mostly just curiosity

Just checking some of their tooling

How they handle the new unobfuscation stuff

[paper cedar]

They did publish an API as usual but they also liberally disable modules (probably not needed for 26.1)

I don't know when they swapped to MojMap but I think it's pretty easy for them (no patches, no installer)

[fresh gyro]

Hmm okey

Well I will do some digging

Lets see if I can find anything interesting

But I doubt it

[obsidian perch]

They did like one or two snapshots after the deobfuscation was announced iirc

[zenith dove]

during the 1.21.11 snapshot phase, yeah

[zenith dove]

we have a new loom plugin that disables remapping entirely

[paper cedar]

You mean a loom update

[zenith dove]

no, a new plugin entirely

loom was split into fabric-loom-remap and fabric-loom, both are on version 1.14

[paper cedar]

Ah interesting

That sounds more like remapping was split off

[fresh gyro]

Okey that is a bit sad, as that means I need to reengineer my stuff but alright then

[zenith dove]

in many ways it's closer to neo now, you can grab a 26.1 template from our template generator or review our new porting docs
https://docs.fabricmc.net/develop/porting/next

[zenith dove]

pretty much, i'd consider fabric-loom-remap closer to MDG legacy

[fresh gyro]

Yeah I guessed as much as I need to completely rebuild the setup for my multi loader system then anyway

[worthy mesa]

modmuss does/did plan a loom 2.0 with bigger changes to come later

[fresh gyro]

Alrighty

[zenith dove]

i think it's closer to did at this point, there's a lot more work there than originally anticipated and the current loom system works fine. we have higher priority work atm like the new javadoc system and the api renames for 26.1

[zenith dove]

idk how helpful it would be, but i'm multiloader/version for 26.1 already through stonecutter. i haven't ported neo to 26.1 yet since my last attempt at that failed
https://github.com/cassiancc/Item-Descriptions/

[fresh gyro]

Will be pretty trivial, as Placitum will do all the heavy lifting

Just need to make sure that it works 😄

[worthy mesa]

autoswitch was also ported with neo if you need a reference, but the changes needed are pretty trivial

[fresh gyro]

But first: NG support for 26.1 😄

Which is nice and ready to go

And a bug fix for 21.11 I think to fix my unit tests

[zenith dove]

that's what i figured, but it wasn't able to find the snapshot build at all, i should give it another look at some point

[fresh gyro]

As that currently errors out it seems if there are no datagenerators

[worthy mesa]

its on maven this time round

[zenith dove]

that's even weirder it didn't work then

[worthy mesa]

you need new mdg, iirc
definately if you had parchment enabled

[zenith dove]

i believe i bumped it, but lemme test again

forgot to bump the java version, works now

[paper cedar]

What failed?

Oh, I think I know what's up

If it was a wall of text from gradle

You need to update your build.gradle to use Java 25

It can't find a version of NF compatible with J21

[zenith dove]

not a very visible error, but yeah it was the java 25 thing

[paper cedar]

You get that errro twice I think, one is from IntelliJ trying to resolve sources and it hides even more details

[worthy mesa]

I saw that recently when I tried to set the compiler option to be release = 21
since I backport to older versions, having that would be nice
also wanted to test if that worked with markdown javadoc

[paper cedar]

In case you have that setting enabled to make it auto-download sources

[worthy mesa]

this thing

tasks.withType(JavaCompile).configureEach {
    options.release = 21
}

[paper cedar]

That's not really the recommended way of doing that anymore anyway

[worthy mesa]

if you mean to use a toolchain, that is not what I want

[paper cedar]

Fucking hell.

I've been chasing a ghost for about 20 minutes.

"Why does this Mixin not apply, not crash, not anything?!"

Well turns out. My proguard config stripped it of all methods.

BTW. Phantoms spawning in a void world is not normal, right?

[worthy mesa]

they spawn in the sky, so I think it is?
I don't think MC checks for void world

[paper cedar]

Hmmm, interesting. I never really noticed that before.

I can't rule out that I temporarily made myself non-creative in this test world. But I was certainly surprised to look up and see 10 phantoms circling while being in Creative Mode 😅

[worthy mesa]

does insomnia even tick in creative?

[paper cedar]

Hard to say. But to validate this I'd need to make a new world and probably sit in it in creative hehe

[inner creek]

it does, afaik

I keep leaving debug runs open in the background while debugging and then return to find phantoms around

[worthy mesa]

I mostly test in survival peaceful, so that's interesting

[paper cedar]

@wintry flower how did you notice this one? 😄 https://github.com/neoforged/NeoForge/commit/c9ef0d0db314655715583cd953cf0c35d9fa4c2e

Could that have led to a crash like this?

java.lang.ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 1025
at it.unimi.dsi.fastutil.longs.LongLinkedOpenHashSet.add(LongLinkedOpenHashSet.java:457) ~[fastutil-8.5.18.jar:?] {}
at TRANSFORMER/[email protected]/net.minecraft.server.level.ChunkMap.setChunkUnsaved(ChunkMap.java:214) ~[minecraft-client-patched-26.1.0.0-alpha.3+snapshot-1.jar:?] {neoforge:access_transformer,neoforge:access_transformer,neoforge:mixin[APP:ae2.mixins.json:chunkloading.ChunkMapMixin from mod ae2]}
at TRANSFORMER/[email protected]/net.minecraft.world.level.chunk.LevelChunk.markUnsaved(LevelChunk.java:182) ~[minecraft-client-patched-26.1.0.0-alpha.3+snapshot-1.jar:?] {}
at TRANSFORMER/[email protected]/net.minecraft.world.level.Level.blockEntityChanged(Level.java:872) ~[minecraft-client-patched-26.1.0.0-alpha.3+snapshot-1.jar:?] {neoforge:access_transformer,neoforge:access_transformer}
at TRANSFORMER/[email protected]/net.minecraft.world.level.block.entity.BlockEntity.setChanged(BlockEntity.java:222) ~[minecraft-client-patched-26.1.0.0-alpha.3+snapshot-1.jar:?] {}
at TRANSFORMER/[email protected]/net.minecraft.world.level.block.entity.BlockEntity.setChanged(BlockEntity.java:217) ~[minecraft-client-patched-26.1.0.0-alpha.3+snapshot-1.jar:?] {}

[wintry flower]

yes most likely

Commoble found it

[tawny spire]

commoble

[Reference to](#neoforge-github message) #neoforge-github [➤ ](#neoforge-github message)patch goof in LevelChunkSection in 26.1 due to deobf changes causes the server to shit itself every 65536 block updates

https://github.com/neoforged/NeoForge/blob/1.21.x/patches/net/minecraft/world/level/chunk/LevelChunkSection.java.patch#L8
https://github.com/neoforged/NeoForge/blob/port/26.1/patches/net/minecraft/world/level/chunk/LevelChunkSection.java.patch#L8

technici4n referenced · Dec 30, 2025, 4:47 AM

[paper cedar]

excellent

CC: @steady heath that might have been the cause of that unexplained modern dynamics crash you had...

[visual basin]

Soaryn found a reproducible test case, I just did the debugging

[hazy wind]

always soaryn... /s

[high sonnet]

Sooooo no snapshot today it looks like

[visual basin]

🐸 is on vacation

[high sonnet]

Or frozen in ice

[plucky hollow]

id say first snapshot back would be 6th

[untold oyster]

next snapshot is in the new year

i'd say the same

[high sonnet]

It would have been funny if they did a snapshot that is the same as the last. Like "sorry we were all on holiday so no changes in this one"

[untold oyster]

too much overhead, probably

[gray frigate]

Or 13th

[untold oyster]

oh yeah, right
I said sometime ago that maybe it might not be the first Tuesday back, since first Tuesday assumes the have work done since last snapshot which they probably don't

someone could go check the historical records to see if they did publish snapshots on the first snapshot day after New Years

[worthy mesa]

25w02a was jan 8
18w01a was jan 3

(just ctrl+f for 01-0 on the launcher meta)

[zenith dove]

they stated in 26.1-snapshot-1 it was the last release of the year

[untold oyster]

so, potentially but I'd say remote chances

hmmm

[wintry flower]

@graceful laurel turns out I am really not convinced by these commas on new lines 😅

it looks quite weird, and I'm not convinced it's of any use when porting

[gray frigate]

That's not commas on new line, comma on new line would be putting it first

[hazy wind]

its almost new years
why snapshot when you can shoot fireworks at your coworkers!
no wait, thats banned in sweden, nvm

[plucky hollow]

i sure hope it is, why would you want to shoot fireworks at anyone?
into the sky sure but at other people?

[hazy wind]

https://tenor.com/view/joke-jk-penguin-laugh-its-was-a-joke-gif-14574338

[untold cypress]

Haskell formatting be like

main = defaultMain [
       bgroup "fib" [ bench "10" $ whnf fib 10
                    , bench "35" $ whnf fib 35
                    , bench "37" $ whnf fib 37
                    ]
                   ]

[gray frigate]

thats what the style guide at work has me do

for all params

so,

void FuncNameYesPascal(
    int A
  , int B
    )
{
    // code
}

[untold cypress]

for reference sci, while that seems to be widely used it's not the only option and I don't use it

[worthy mesa]

it's not terrible format
certainly nice for comments without block comments

[gray frigate]

TBF with this, it does make adding (or removing) an arg a single line diff, except for the first arg

[graceful laurel]

Which, surprise surprise, is exactly the reason why I did that in some patches

[paper cedar]

it doesnt really matter if its a single line diff

the context will break the patch should that line change anyway

[untold cypress]

the way rust solves that is just allowing a trailing ,

[gray frigate]

at work, it does keep the p4 blame for a given argument line correct, which has been helpful

[untold cypress]

so just

fn foo(
  a: u32,
  b: u32,
  c: u32,
) {
  todo!()
}
```is completely valid

[untold oyster]

Perforce?

I only know of it from the Git book

[gray frigate]

yup, perforce

its pretty common in games

it handles large binary blob assets better

things like textures

[untold cypress]

isn't that what lfs is for?

[gray frigate]

the thing lfs doesnt solve is exclusive checkout

w/ p4, checkout can be exclusive, so nobody else is modifying the same file as you

this matters for things that can't (easily) be merged (see textures)

[untold cypress]

ah, file locks

[gray frigate]

fwiw, i have worked on games in both p4 and git (w/ LFS)

one of the other big issues with git is needing a copy of the repo locally

even if sparse, that gets very large very quickly

p4, you only need to store current revision locally

when you have 1TB+ of texture/model data, not having to keep a second copy of everything matters a lot

now dont get me wrong, for code i very much do prefer git

but its not like p4 is without its benefits too

and unified VCS is also a good thing, splitting code and data has its own drawbacks

[untold cypress]

from what I remember you can do similar stuff with git, but sure, use what works for you

[gray frigate]

look, i dont make decisions on what VCS to use

there are ways to do similar things in git, but its not as automatic
the project i worked on that used git LFS, and had huge assets, also had a custom git client to handle it all

all the p4 projects ive been on, its just the standard p4 client

the fact that p4 is just "off the shelf" matters for enterprise

[fresh gyro]

Hey is anybody able to get into a world on the current 26.1 snapshot build of NeoForge?

It gets stuck at this screen

WHich makes me wonder if i have a thread hanging somewhere

My thread dump seems to be entirely normal

[wintry flower]

yes I am able to on alpha 2

[fresh gyro]

I am currently on alpha 3

And it gets stuck here

[wintry flower]

ah no oops this was in neoforge so a very recent (not released yet) version

I guess you'll have to check why it's not connecting

could be a misnamed variable literally anywhere

[fresh gyro]

Yeah

Well the client thinks it is rendering the world.....

Actually never mind

Level is still null

[visual basin]

I had alpha-1 working earlier, lemme try alpha-3

[visual basin]

yeah alpha-3 works for me

[fresh gyro]

Wtf

Why would this not directly work

[wintry flower]

could be a linux/wayland-exclusive problem

[untold oyster]

[fresh gyro]

This seems not correct

[worthy mesa]

do you have prchment?

[fresh gyro]

It is 26.1 so no

[untold oyster]

can't repro the issue on my fresh setup of port/26.1 -- I can create a world and load into it fine

for avoidance of doubt, I am on Linux using Wayland

[untold oyster]

where is this

[untold cypress]

or it's a compositor issue, someone try on different wayland based ones

[untold oyster]

what I have here is packet == delimiterPacket

are you in a mod dev workspace or in a Neo dev workspace?

[fresh gyro]

Yep

It was parchment

FFS

[untold oyster]

smh

[fresh gyro]

Okey

Note: need to figure out a way to enforce it to be disabled

That that does not crash our fucking patcher

Is somehow a wonder....

[hushed canopy]

Oh wait you found another of these issues, hooray!

Looks a heck of a lot like https://github.com/neoforged/JavaSourceTransformer/issues/44

Which I thought maty fixed. But I guess there's some edge case or something?

[empty night]

TMW the codebase has more edges than a icosahedron

[wintry flower]

ah yeah I think shartte ran into this exact bug

we already made a patch to MDG that disables parchment param renaming

[hushed canopy]

Ohh wait is this on unobf versions

[untold oyster]

yes

wait, why am I answering

[hushed canopy]

Ahh I think that must be it, the fix maty had probably tests for parchment/parchment name conflicts but doesn't detect parchment/existing or parchment/local name conflicts

Welp

[untold oyster]

-# I answered on instinct before I considered you may be referring to another thing

[empty night]

-# you wanted to feel included, it's okay

[fresh gyro]

I should probably do that too.

makes a note

[fresh gyro]

What is the version format that CurseForge supports for uploading?

[paper cedar]

What are you using to upload?

[fresh gyro]

CurseForgeGradle, Darkhaxes plugin

[paper cedar]

CurseForge has a weird support-level for snapshots (which is, you can manually download files for them, but the launcher does not support them at all)

26.1-snapshot is their version id for all 26.1 snapshots

[fresh gyro]

Ahh

[paper cedar]

they dont differentiate between individual versions (Modrinth does)

[fresh gyro]

So yeah it probably detects the mc version from my properties

Okey

I think I can fix that manually then

[paper cedar]

Yeah I set the version manually for Modern Dynamics too

[fresh gyro]

Ah yeah I even set it manually

I just need to set it from a different property then

[lucid tree]

Should I design a plugin that publishes to both modrinth and curseforge?

[hard citrus]

I’ve been happy with modmuss’ mod publish plugin

[paper cedar]

don't do more gradle plugins for that

The more appropriate way is ultimately GH actions

[hard citrus]

And whenever something doesn’t work I’ll PR a fix

[paper cedar]

While this fell slightly into disrepair, it's still the best option IMHO: https://github.com/Kira-NT/mc-publish

[lucid tree]

https://github.com/modmuss50/mod-publish-plugin

Okay ima switch to this

[hard citrus]

That’s less portable but true

[paper cedar]

You shouldn't publish from your own machine either way

The issue with gradle plugins is that you cannot retry

and since you're necessarily hitting external systems

[hard citrus]

No I just call the publish task in GH actions

[paper cedar]

if you do it in separate jobs, you will have to ship the artifacts over from job to job

which is very hard to do correctly without gradle rebuilding the jars

[lucid tree]

I publish from my own machine

Whats wrong with that?

[hard citrus]

I’ve made more cursed GH actions stuff ;) but that’s besides the point

[jagged ridge]

Higher chance of jars being compromised

[paper cedar]

not reproducible, if you inadvertently depended on some arcane config on your machine

[lucid tree]

On my own machine?

[paper cedar]

i.e. you 5 years ago put some token in gradle.properties, forgot and your publish only works because of that 😄

[hard citrus]

-# the action runs commands on a server via ssh

[lucid tree]

Well I wont forget that lmao

[paper cedar]

I'd recommend anyone to publish via some form of CI/CD tbh

[lucid tree]

I mean if there was some way to make GH actions run when commit message contains "release" sure

Time in a bottle doesnt need a per commit release

[indigo venture]

git tags

[lucid tree]

I mean I suppose

[paper cedar]

yes, git tags

[lucid tree]

Hrm

Maybe Ill do that

[paper cedar]

that is how this snapshot is being released

we're not doing per-commit for this one

[lucid tree]

I mean I dont want it to run every commit

[hard citrus]

Yeah then use tags

[lucid tree]

Hm alr then

[hard citrus]

GH releases require tags anyways

So if you plan to publish to GH releases too you’ll need tags

[lucid tree]

Then ill do that then, maybe.

For 26.1 release of TIAB

https://modmuss50.github.io/mod-publish-plugin/multi_platform/

Fuck it

Ima switch to this for damn sure

[vapid juniper]

I've used unifiedpublishing plugin in the past, it works well

[lucid tree]

Neat

[alpine minnow]

The tools that used to help with unobfuscation Minecraft’s code what will happen to them?

[fresh gyro]

They stop being maintained

We don't need them anymore

[round spear]

Guessing we still would be willing to merge PRs but like the priority within the team to review said PR isn’t likely to be that high

[paper cedar]

• What went wrong:
  Execution failed for task ':neoforge:createChangelog'.

Cannot invoke "org.eclipse.jgit.lib.AnyObjectId.hashCode()" because "id" is null
Fuck JGit

[untold oyster]

@paper cedar deleted tag?

[paper cedar]

Well. See above.

[untold oyster]

[paper cedar]

Disabled changelog, will try again.

Alright, new version: 26.1.0.0-alpha.4+snapshot-1

[plucky hollow]

its day my dudes, what we all thinking, snapshot today or nah?

[paper cedar]

Ah right, technically they could release one

I don't think there will be one

[lethal onyx]

today is a public holiday in sweden

so if we get one this week, probably not today

[plucky hollow]

oh it is, then yeah maybe no snapshot today then

[inner creek]

I don't expect one this week

[worthy mesa]

https://www.minecraft.net/en-us/article/minecraft-26-1-snapshot-2

[plucky hollow]

<@&1067092163520909374> and so it begins

[alpine minnow]

Oooo

[quartz spade]

New version detected: 26.1-snapshot-2.

[next harness]

yee

[craggy river]

Huh. Craftable nametags?

[hazy wind]

ohhhhhhh

[plucky hollow]

26.1 Snapshot 2

• Primer: https://github.com/ChampionAsh5357/neoforged-github/blob/update/26.1/primers/26.1/index.md
• Article: https://www.minecraft.net/en-us/article/minecraft-26-1-snapshot-2
• Changelog: https://misode.github.io/versions/?id=26.1-snapshot-2
• Notion: https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2e14f070f88880ec8dd7d172e6ba4e58&pm=c
• SnowMan: https://github.com/neoforged/Snowman/commit/c75852363925125bdc02842fa03e93589b9dab82

SlicedLimes Videos:

• Main: https://www.youtube.com/watch?v=ro-fzLZXEfI
• Pack: https://www.youtube.com/watch?v=RzGvQBs-h-0

[craggy river]

They seriously are making this too easy for people these days. First saddles, then name tags?

[shy valve]

I see they're trying to entice me into playing with sound again. clever trick Mojang

[paper cedar]

Ah nice so we can check our update infra with unobf

Aaaaaaaaaaaaaand it has bugs. 😄

https://github.com/neoforged/NeoForm/actions/runs/20786054390/job/59695769292

[inner creek]

oh shit

unexpected snapshot lol

[quartz spade]

Minecraft 26.1 Snapshot 2

We're back from vacation and the lights are back on in the shipping room! In this very first Snapshot of the new year, we're bringing you the first features from our cutest game drop yet! Befriend floofier wolf pups, kittens, piglets, calves, lambs, baby c...

Jan 7, 2026, 3:05 PM

[north berry]

kitties

[inner creek]

nice!

[untold oyster]

just when I woke up

[shy valve]

there's a balance between fun and easy. having to grind through mineshafts or fishing to get nametags is neither hard nor fun. it's just tedious for something people would like to use more often. Saddles are even more justifiable considering it poses a barrier to what is quite possiblly the best early game transportation, but by the time you find one in a chest, you probably don't need horses anymore, so what's the point of mounts. Craftable saddles was a good choice. ||plus i've had these recipes in a personal datapack for years now because it just made sense to me||

[untold oyster]

...huh, MC ran on 2GB by default?

[inner creek]

yes

[untold oyster]

I thought it ran on 4GB

[inner creek]

nah, vanilla runs "fine" on 2gb

or at least did

[shy valve]

oh noe! my toaster can't run MC anymore. gonna have to migrate to the fridge.

[bold forge]

418

[untold oyster]

interesting new rules in the version manifest

arguments.default-user-jvm

[paper cedar]

Oh we definitely don't support those

[inner creek]

LOOK AT THIS OMG

[plucky hollow]

TagKey<Item> METAL_NUGGETS = bind("metal_nuggets");
well that goes against our tag conventions

[paper cedar]

oh fuck thats good

😄

[untold oyster]

literally playable

[plucky hollow]

do we update nuggets/iron and such to be metal_nuggets/iron now?
cc: @runic anchor

[bold forge]

that's too cute

[obsidian perch]

Discord

[obsidian perch]

Chick!

[compact swan]

how many files were added and deleted this snapshot?

[compact swan]

ah pretty good

[inner creek]

I made it a sticker

[compact swan]

hell yeah

[paper cedar]

One bugfix later: https://github.com/neoforged/NeoForm/actions/runs/20786480245/job/59697310460

[inner creek]

doesn't look very good as an emoji sadly

[paper cedar]

Ah, ofcourse i forgot to do what Shrimp had told me, and it thinks the snapshot is a special version

Sadly, I only remember that shrimp told me something about that, not exactly what though.

[snow island]

This is real?

[plucky hollow]

looks like mojang is making x,y,z fields in pos stuff private
alot of the changes are just .x -> .x()

[compact swan]

huh

[paper cedar]

That was not a record before?

[plucky hollow]

also chunk/block pos long packing methods renamed from toLong and fromLong became pack and unpack

[inner creek]

yes look at the new snapshot. also look at this cutie

[plucky hollow]

nope

[compact swan]

awwww

[paper cedar]

Okay so it's less about making x/y private and more about them making it a record 😄

[inner creek]

the other ones are adorable too but the rabbits and chicks are the biggest change

[snow island]

Now I am wondering about theme of this drop

[runic anchor]

Sure

[inner creek]

depends on the other features they might add later

[runic anchor]

Is Mojang adding a food nugget

McDonalds in Mc

[plucky hollow]

chicken nuggets mojang i want

please and thank you

[runic anchor]

I can’t read blog so idk what is happening and I’m at work rn

[inner creek]

kill adult chicken -> you get chicken meat
kill baby chicken -> you get chicken nugget

[tawny spire]

gigaherz

[Reference to](#1445764870564745228 message) #1445764870564745228 [➤ ](#1445764870564745228 message)nice!

Jan 7, 2026, 3:16 PM

[inner creek]

this

the snapshot is small

[shy valve]

https://tenor.com/view/jail-right-to-jail-right-away-parks-and-recreation-parks-and-rec-gif-16177531

Kill baby chicken?!

[runic anchor]

So what does that have to do with metal nuggets

[inner creek]

it happens to be a change in the code 🤷‍♂️

[inner creek]

wrong link, I think

one of the bots has the wrong link

https://www.minecraft.net/en-us/article/minecraft-26-1-snapshot-2

[plucky hollow]

link in pin works fine for me

[inner creek]

they provide no explanation for it

[worthy mesa]

for the nametags

[plucky hollow]

added for the name tag recipe, so that it accepts any nugget

[runic anchor]

It can be viewed as nuggets/metal and Mojang just does metal_nuggets

[plucky hollow]

also am i reading that right? have to obtain a nametag to unlock the nametag recipe?

[runic anchor]

So no need to redo our tag and see metal nugget as a subset of nugget

[worthy mesa]

has paper, or nametag

[inner creek]

ooooh

[plucky hollow]

guess this is done to try and "force" players to explore
and find a nametag via dungeon chests and such first

[inner creek]

doesn't matter though? unless you disable crafting locked recipes

I'm surprised they didn't include string in the recipe

[plucky hollow]

yeah nugget feels weird tbh

its just paper with a string

wheres the nugget fit in

[inner creek]

I guess they see it as a metal ring

[paper cedar]

Not having to look up timestamps for neoform versions is nice

[plucky hollow]

i guess it does kinda look like a metal ring
but ive always seen it as string

inb4 26.1-snapshot-3-hotfix "nametag recipe now uses string"

[runic anchor]

Seems too thick for string

[inner creek]

yeah

but I always imagined

this kind of tag, with string and cardboard

[worthy mesa]

I always saw it as metal for both the ring and tag
since animals are outdoors

[strong wing]

do max librarians still offer them in a trade

[untold cypress]

I only ever used saddles because some mod's crafting recipe required it

[untold cypress]

According to the wiki, yes, at 20 emeralds a piece

[paper cedar]

Could be worse

new ChunkPos(BlockPos) -> ChunkPos.containing(BlockPos)

ChunkPos#asLong() -> ChunkPos()#pack
ChunkPos#toLong() -> ChunkPos()#pack

And predictably: new ChunkPos(long) -> ChunkPos#unpack(long)

[round spear]

Ohno does this mean for mek additions baby mobs I will need to do something more than just have a smaller model . At least the list doesn’t mention baby zombies yet

[plucky hollow]

i demand more cute adorable baby models, so... yes get on that /s

need a baby zombie we cant kill cause he too damn adorable

[round spear]

Aren’t the unlockedBy when separate like that just need any of instead of all of? I think it is just to give you the recipe if you find a name tag and haven’t made any paper or nuggets

[round spear]

Take that up with mojang

[lethal onyx]

kill the babies

[plucky hollow]

heard it here first schurli hates babies

[lethal onyx]

inb4 baby skeletons and creepers

[round spear]

Yes please, then I don’t have to maintain those in mek additions. Also endermen, and strays. (And wither skeletons )

[inner creek]

Baby Bogged

there's more than just strays :P

Drowned already have babies right?

[round spear]

I have those too. I just forgot they added those and wasn’t looking at my list

[round spear]

Yes

As do husks

[inner creek]

Parched?

[round spear]

?

[zenith dove]

those are skeletons so no

[inner creek]

the Parched is a deserd skeleton

desert*

[paper cedar]

https://github.com/neoforged/NeoForge/pull/2911/changes

[round spear]

Oh

[inner creek]

ignore the tool belt, that's my addition :p

[round spear]

I have yet to have 1.21.11 compile

I haven’t looked at vanilla’s changes since 1.21.1 yet to add parity to

[paper cedar]

Most of it is patch context changes, only semantic change I could see is that fall-damage calculations now calculates an "effective fall distance" first, and it's not quite clear to me where that was before (it might have been done in the player subclass, tbh.)

I moved the event to below that calculation and made it use the effective distance

Here: https://github.com/neoforged/NeoForge/pull/2911/changes#diff-660dc5e9536846ae570d834fe28cf91fcfc8229bcaa47c44b5fb0d06e3ca6e3dL385

[round spear]

But yes eventually I will have the parched one as well (once I get mek compiling again and start going through the changelog to see what stuff I need to mirror)

[inner creek]

ah

[round spear]

I am still many many hours from getting mek to be compilable on 1.21.11

And then I get to adjust to transactions….

[plucky hollow]

👀

[paper cedar]

I just made it a PR so I could wait for the CI/CD checks. Didn't want to pollute the port/26.1 history with "fix up xyz" crap if I could avoid it 😄

[plucky hollow]

thats fair was just suprised to see it so quick

[paper cedar]

Well I just hammered it out the snapshot was trivial in terms of porting

Okay, I tagged a version which just finished building, too: 26.1.0.0-alpha.5+snapshot-2

[plucky hollow]

wooo time to update my mods

[paper cedar]

I probably use ChunkPos somewhere, but that shouldnt take long to update

[plucky hollow]

nice not a single line of code changed for me, but thats not hard when your a furniture mod xD

[paper cedar]

Now: add baby furniture.

[plucky hollow]

oh cat trees and shit would be real nice ngl

but alas im not the artist for my mods
and the artist has gotten busy with irl shenanigans recently

[round spear]

Still something you could add to your backlog!

[plucky hollow]

yeah ive been slowly learning block bench

but wont be able to match what my friend did
they some how matched vanillas style while also being their own thing

[paper cedar]

I also updated the earlier pin

[robust pike]

Have a primer: https://github.com/neoforged/.github/pull/39/commits/b9f682d6b4b85a5827df5d2d6b61d82b31ee4b53

Mostly just some entity model splitting and ChunkPos now a record

Back to resting

[steady heath]

https://www.youtube.com/watch?v=ro-fzLZXEfI

[untold oyster]

can't wait for the inevitable plushies

but first, gimme ghastling/happy ghast

[plucky hollow]

i still only see pahi-ghast when seeing the happy ghast

[lethal onyx]

Because I have not seen it posted here: https://www.youtube.com/channel/UC2UyQpCGPw01q62SapQSuyw

[robust pike]

I got a code review on the primer, never expected that

But thanks shnups

[plucky hollow]

i mean not posted since last snapshot but yeah the datapack changes are uploaded to a 2nd channel now
which confused me yesterday as i thought lime was moving the snapshot videos as a whole to a different channel
not just the pack changes
#1445764870564745228 message
#1445764870564745228 message

[inner creek]

he's just separating the snapshot news from the technical (see: resourcepack and datapack) change news

[lethal onyx]

so... what does one need to do to update to a snapshot build using NG?

[fresh gyro]

Just bump the version?

You need to be on 7.1.20 though

[compact swan]

I think they should add baby creepers

[plucky hollow]

so long as they arnt like modded baby creepers

those things suck

[compact swan]

what are those like?

[plucky hollow]

tbh i might be confusing them with a different mob
but i remember a mod adding baby mobs
and they had some of the most annoying effects

@round spear am i thinking of mekanism additions?

[plucky hollow]

https://www.youtube.com/watch?v=RzGvQBs-h-0

[inner creek]

they explode less, but move super fast and explode much faster, so you have very little time to move away

the worst, by far, are baby skeletons

much harder to hit, move annoyingly fast in the way skeletons move sideways and such

and quick-fire arrows at you

[compact swan]

damn that sounds like a lot of pain

[round spear]

Baby creepers in mek additions have half the explosion power of normal creeepers

[round spear]

(That is their only special power)

[shy valve]

i'm pretty sure that feature was why i stopped using mek additions lol

[round spear]

The fact that they only blow up like 5 blocks max when a normal creeper blows up way more than that?

[runic anchor]

when's 26.1 release date

[zenith dove]

march/aprilish probably, they haven't said

[inner creek]

previous spring drops have been late March

[alpine minnow]

Ugh thats too far away😭

[empty night]

Not me over here hoping Mojang finally kills minecraft:mob_griefing and splits it into enderman, creeper, and villager game rules...

I learned the other day that turning off that rule also stops villager breeders and other farms from working, because they stop picking up food and harvesting

[untold cypress]

Yeah, picking up items is griefing

[untold oyster]

-# DIE, GRIEFER

[lucid tree]

So villagers starve to death 💀

[plucky hollow]

oh shit, its 🐸 day this past week zoomed by for me

[rain herald]

"Also Play Hytale" is coming trust

[paper cedar]

Yeah, I'll hold off on that for a week to get the sentiment of ppl actually playing it 😅

[plucky hollow]

the fact the fan base has to be like "trust me its coming" is the whole reason im not intrested in hytale
its had way too many rewrites/restarts and chances at this point and has lost me as any potential player

[rain herald]

I meant the splash text

I should've quote wrapped it

I am also not buying the hype but I think itd be cute to add a splash

[rain herald]

true

[plucky hollow]

i already have a few friends who are like "yooo you should port your mods to hytale"
like jesus at least give the base game some time breath before to bombard it with mods

which is another reason why im not so intrested in it
if people are already like, "the mods gonna be crazy" the base game cant be all that good
and from what ive heard theres no client side modding capabilities? which i feel will limits mods potentials
i mean look at mc plugins vs mods, ones clearly more powerful than the other and thats cause you have client to play with
rather than being server only reskininng existing game elements, hijacking and forcing them into what you want them to do

[bitter gorge]

@plucky hollow it is time

[plucky hollow]

im not seeing any new version yet

[tawny spire]

[Jump to referenced message](#minecraft-updates message) in #minecraft-updates

covers1624 referenced · Jan 13, 2026, 3:01 PM

Minecraft 26.1-snapshot-3

This is a new version. Changes listed from 26.1-snapshot-2.

Links:
client | server

Quoted | Last updated · Jan 13, 2026, 12:57 PM

[quartz spade]

New version detected: 26.1-snapshot-3.

[zenith dove]

hytale is built to be server modded whereas minecraft is not, the server seems to have a lot more control than minecraft's. this is a game built by people who have been server modding for years, they want it to be as good as can be

[plucky hollow]

26.1 Snapshot 3

• Primer: https://github.com/ChampionAsh5357/neoforged-github/blob/update/26.1/primers/26.1/index.md
• Article: https://www.minecraft.net/en-us/article/minecraft-26-1-snapshot-3
• Changelog: https://misode.github.io/versions/?id=26.1-snapshot-3
• Notion: https://apexmodder.notion.site/2524f070f88880648482e3cbdcba566a?v=2524f070f88880b29f60000c52785ecb&p=2e74f070f88880c0a8f6f1de7e62c418&pm=c
• SnowMan: https://github.com/neoforged/Snowman/commit/267ca579ae98928beff4cbc7c3946c3516b3556b

SlicedLimes Videos:

• Main: https://www.youtube.com/watch?v=YLHTLf8RNIs
• Pack: https://www.youtube.com/watch?v=ILnYqYftvX4

||<@&1067092163520909374>||

[next harness]

[inner creek]

woo!

ingame game rule editor!

[inner creek]

oh interesting

[fresh gyro]

So many tags

[inner creek]

they added a time-tracking system

[plucky hollow]

https://github.com/neoforged/Snowman/commit/267ca579ae98928beff4cbc7c3946c3516b3556b

[inner creek]

yay one less patch

[lethal onyx]

The Tripwire texture is now rendered as alpha cutout instead of transparent

Interesting

[visual basin]

does it just use cutout rendertype now instead of having its own special bunny rendertype

[lethal onyx]

Count in smelting is huge

[plucky hollow]

yes

[visual basin]

now you can duplicate ores without mods!

[plucky hollow]

@paper cedar another thing ItemResource may want to inherit?

[inner creek]

[plucky hollow]

ItemResource already extends TypedInstance and DataComponentGetter why not also this

[inner creek]

could extend ItemInstance instead of the other two, I guess

[bitter gorge]

because ItemResource doesn't have a count

[plucky hollow]

wait..

[inner creek]

oh right

[plucky hollow]

did mojang add ItemResource?

[quartz spade]

Minecraft 26.1 Snapshot 3

We're back in action with another Snapshot! In this week's release we're making it possible to access game rules from the in-game pause screen, through the new world options button. On the data pack side of things, we're introducing a new concept of world ...

Jan 13, 2026, 3:01 PM

[inner creek]

no cos it has count

so it's automatically not the same :P

[plucky hollow]

oh yeah the count i missed that

[inner creek]

it's just an immutable itemstack

[plucky hollow]

i like immutable stacks

ooh

[bitter gorge]

its POJO for data/network

[plucky hollow]

SpawnEggItem lost its BY_ID map intresting

[runic anchor]

FUCK YES

[untold cypress]

so what mods have been doing for ages can now be done natively

[runic anchor]

FINALLY

WE HAVE ASCENDED

TIME TO DELETE THE NEO PATCH AND MY CUSTOM RECIPE

[untold cypress]

the first example that comes to mind is powah's uranium

[inner creek]

I knew some people would be very happy as soon as I saw it

I underestimated it

[untold cypress]

Oh well, uranite as it's called iirc

[visual basin]

#supports_vegetation which provides support to the following blocks:
what does that do to croptypes

oh fuck they added a bunch more

[untold cypress]

this time stuff seems interesting

Maybe I'll need to expose that in my mod

[visual basin]

#supports_vegetation
#supports_crops
#supports_stem_crops
#supports_pumpkin_stem
#supports_melon_stem
#supports_sugar_cane
#supports_sugar_cane_adjacently
#supports_cactus
#supports_chorus_plant
#supports_chorus_flower
#supports_nether_sprouts
#supports_azalea
#supports_warped_fungus
#supports_crimson_fungus
#supports_mangrove_propagule
#supports_hanging_mangrove_propagule
#supports_nether_wart
#supports_warped_roots
#supports_crimson_roots
#supports_wither_rose
#supports_cocoa
#supports_lily_pad
#supports_frogspawn
#support_override_cactus_flower
#cannot_support_seagrass
#cannot_support_kelp
#dry_vegetation_may_place_on -> #supports_dry_vegetation
#bamboo_plantable_on -> #supports_bamboo
#small_dripleaf_placeable -> #supports_small_dripleaf
#big_dripleaf_placeable -> #supports_big_dripleaf
#mushroom_grow_block -> #overrides_mushroom_light_requirement
#grows_crops

[plucky hollow]

holders can now be given a component map?

Holder

[visual basin]

[plucky hollow]

registries have a component lookup

intresting

[visual basin]

related to default datacomponents for items maybe

might be necessary for data-driven items later

[plucky hollow]

expanding the component system for any registry type id guess

but mostly items for now?

[raven wharf]

article link from this bot is broken btw

[rain herald]

data driven time is such an interesting addition

[plucky hollow]

yeah thats cause its using the old link format iirc
the pin has the correct link

[rain herald]

I liked timelines now theyre taking it further

very fun

can u now make an in-game day 24 hours with a datapack?

[raven wharf]

yeah I scrolled up to it when I noticed it wasn't working just wanted to state for the bot just incase it wasn't known lol

[plucky hollow]

oh nice we dont need our custom behavior for custom spawn eggs anymore

[rain herald]

kill the bot

[untold cypress]

apex's a better bot

[plucky hollow]

little do yall know ive been a bot the whole time

[untold cypress]

https://tenor.com/view/game-changer-dropout-sam-reich-ive-been-here-the-whole-time-gif-25231055

[plucky hollow]

hey a fellow dropout fan

oh yeah default item components seem to be coming from the registry now

[compact swan]

Sam Reich is so cool

[plucky hollow]

thats AirItem btw which used to look at the Item components for the base name
not it pulls it from the Holder not the Item

wonder how this messes with the modify default component event we have

[untold cypress]

So any registry could technically have default components now?

[plucky hollow]

yep Item.components does not exist anymore

[plucky hollow]

by looks of it thats where we are headed

unsure if thats how it fully works in this snapshot tho

[visual basin]

wonder if the datamap system could piggyback off that

[obsidian perch]

I... don't know actually

The thing is though that the world would have to be running 24/7 or it would get out of sync

[plucky hollow]

maybe even drop datamap in favor of data component in future?

[visual basin]

maybe

how are datacomponents even cheesed into holders though

how are they set, how mutable are they

[untold cypress]

can you only have data maps for registries?

[visual basin]

datamaps are tied to the registry system, yeah

[rain herald]

this is huge

[inner creek]

I wonder if that's the first step toward json-defined data component maps
like our DataMaps but using DataComponents

[rain herald]

wow

mojang w

[compact swan]

what do they mean with this? loot tables can be based on time?

[rain herald]

https://tenor.com/view/incredible-talented-lady-brilliant-gaga-gif-14857187

[inner creek]

yep!

[paper cedar]

yes

[rain herald]

indeed

[untold cypress]

Night only loot

[visual basin]

they already were, they just added a new field for which time you're checking

[rain herald]

where?

who depended on time

[visual basin]

in the minecraft:time_check loot predicate, it says

[compact swan]

interesting.... I wonder what they will do with that

[rain herald]

ohhh

the check already existed

just without the clock

[compact swan]

well what is it used for?

[rain herald]

great question for intellij

[visual basin]

nowhere

mojang doesn't use it themselves

[untold cypress]

Sun crystals during day, moon crystals at night

[runic anchor]

Likely for datapackers

[inner creek]

I assume it was made for packmakers

[visual basin]

theyy just added it as a nice thing for datapacks

[runic anchor]

Content creators

[inner creek]

adventure map stuffs

[runic anchor]

Influencers

YouTubers

Tiktokers

[untold cypress]

Outfluencers

[rain herald]

imagine an entity that only drops certain loot during sunset or sunrise

[untold cypress]

^

[inner creek]

"Sunflowers only drop loot when the sun is out"

[untold cypress]

That's a weather + time check tho

[compact swan]

ooo

[untold cypress]

and idk if there's a weather predicate

[visual basin]

WeatherCheck!

[gray frigate]

If the client is made with the idea of server-side-only mods in mind, it is possible to make server-only mods substantially more powerful than they are in MC

[plucky hollow]

oh i think i see, item properties now has a delayedComponent method which tells the component initializer to include said component as default, items then use the initializer to build the default component map and pass it along to the hold holder via a static initializer list

[gray frigate]

part of the problem with MC is that some things (animated models, guis) are completely client code created/defined

in most games, thats just data, so theres no need to mod the client for those if it can load some additional data from the server

[visual basin]

okay so holder components are immutable and generated on registration

[plucky hollow]

ItemProperties -> delayedHolder -> Item() -> BuiltinRegistries.DATA_COMPONENT_INITIALIZERS
after that im unsure

[plucky hollow]

seems to be that way from my quick skimming around

[visual basin]

which is the opposite of datamaps

[round spear]

Are delayed holders similar to our deferred holder in any way?

[robust pike]

Seems like they killed off interactAt

Merged completely into interact

[plucky hollow]

delayedComponent my bad, its just how mojang is tracking the components on the properties now
before handing them off to the registry, for holders to lookup by registry key

but yeah i feel this might kill the modify default item component event we have

did anyone even use that event?

[gray frigate]

if the client is able to load new data from the server, that opens up a lot of capabilities

[round spear]

Part of the idea of that event is to add components to vanilla stuff not just your own

[obsidian perch]

Check WAIFU

[visual basin]

a quick discord search for that event name suggests people are definitely using it

[plucky hollow]

tbh looking at it more we might be able to andThen the intiializer to inject our event into it

similar to how the finalizer validates the components, we just andThen our event before the validator?

[plucky hollow]

but i could see the event being dropped in future anyways, if components truly do go json defined

at that point just use datapack to add components to vanilla stuff

[visual basin]

eehh

as long as datamaps can be assigned by tag and mojangcomponents can't then datamaps have a niche

[plucky hollow]

we may also want to also swap ItemResource internals to use ItemStackTemplate rather than ItemStack
so its fully immutable, not backed by some mutable stack type

also ive seen a few places in vanilla code that swapped ItemStack to ItemStackTemplate so maybe ItemResource gets a toStackTemplate method too, but also not that desperately needed either

[lethal onyx]

It looks like they are swapping in ItemInstance (the common interface between ItemStack and ItemStackTemplate) in some places too

[plucky hollow]

oh ive seen a few places that took in ItemStackTemplate

but taking in the iface is nice too

i mean its mostly been rendering and network stuff which makes sense that should be immutable
so yeah not as needed for resources to fully support

[paper cedar]

I'll start the update

[wintry flower]

That's probably not a great idea

It's essential that conversions from/to ItemStack remain fast

We should also preserve symmetry between fluids and items

[plucky hollow]

thats actually fair yeah, i was just thinking of removing any possible chance of resources being mutable
having them be backed by a immutable type would make sense

[wintry flower]

Well, is accidental mutation of resources a problem in practice?

I haven't seen it happen so far

[plucky hollow]

yeah ive not seen it myself either but is possible mostly by misusing the is(Predicate<ItemStack>) method since that passes the internal stack directly

[paper cedar]

Hm. our AfterWeather rendering event is emitted before the weather-special-effects renderer is called

Is that a port mistake?

[plucky hollow]

but keeping the resource->stack conversion fast i will agree is a good

[plucky hollow]

sounds like a misplaced patch to me

[paper cedar]

I moved it to after

Ok due to the extensive changes to time-keeping, I think our patch to control the flow of time in levels needs a deeper look.

Oh, the introduction of ItemTemplate makes at least one patch redundant

(In ItemParticleOption)

-            this.itemStack = itemStack;
+            this.itemStack = itemStack.copy(); //Forge: Fix stack updating after the fact causing particle changes.

[plucky hollow]

yeah mojang having immutable stacks gonna be nice, more they get used

[paper cedar]

Someone is going nuts on RegistryDataLoader

patch-wise, the reformats caused by vineflower when variables are renamed is a real pain in the ass.

i.e.:

26.1-snapshot-2:

return CompletableFuture.allOf(loadCompletions).thenApplyAsync(ignored -> {

26.1-snapshot-3:

return CompletableFuture.allOf(loadCompletions)
    .thenApplyAsync(
        ignored -> {

Ah. ServerClockManager is probably where the time logic went.

[paper cedar]

TODOs (will edit):

• restore SleepFinishedTimeEvent
• our custom overloads for daytime customization in Level (this.setDayTime(this.levelData.getDayTime() + advanceDaytime());)
• decide what to do about state.is(BlockTags.SUPPORTS_VEGETATION); where we previously added instanceof FarmBlock
• decide whether to keep state.canBeHydrated(...) in combination with fluidState.is(FluidTags.SUPPORTS_SUGAR_CANE_ADJACENTLY)
• should Item#getMaxStackSize(ItemStack) now be made Item#getMaxStackSize(ItemInstance), since the max stack size query was moved from ItemStack to ItemInstance?

Notes:
FungusBlock -> NetherFungusBlock
FarmBlock -> FarmlandBlock

[paper cedar]

Isn't this a misapplied patch too? I think this lacks an else for the subsequent default block?

[plucky hollow]

https://www.youtube.com/watch?v=YLHTLf8RNIs

[lethal onyx]

more reloadable registries?

[paper cedar]

I think they just refactored it

After noticing it was incredibly large/bloated

[lethal onyx]

we should have a formatter run on the VF output so it is more consistent

[paper cedar]

They pulled out the inner classes ("loading tasks") into top-level classes

[paper cedar]

I am not sure it wouldn't make it worse. Although, yeah. maybe 🤔
It'd probably be a formatter that makes it very verbose, each method parameter on its own line and stuff like that....

Hmhmhmhmhmhm. Maybe we should just do that for generating, storing and applying patches

@wintry flower do you remember why we validate components in component() rather than in validateComponents? Probably because we want "nice" stack-traces ?

There's now a "delayed component" system for components that can only be created once registries exist

So that'll be a bit difficult

[lethal onyx]

we'd need to andThen it to the initializer

[paper cedar]

I think that returns an entire map. I am juggling it into SingleComponentInitializer#asInitializer

That's where it calls #create on the supplier

But at that point you'll not get a stacktrace that points to the culprit, really

Mhmmm. okay but I think that's the point at which we shove that modify event into

That's certainly interesting. It's modifying a built-in registry in the Item constructor

public Item(Item.Properties properties) {
    this.descriptionId = properties.effectiveDescriptionId();
    DataComponentInitializers.Initializer<Item> componentInitializer = properties.finalizeInitializer(
        Component.translatable(this.descriptionId), properties.effectiveModel()
    );
    BuiltInRegistries.DATA_COMPONENT_INITIALIZERS.add(properties.itemIdOrThrow(), componentInitializer);

Oh. Huh. That's not a real registry

[visual basin]

it's just some weird object they stuck in BuiltInRegistries

[paper cedar]

Also uh

I never noticed that check in the Item constructor that we patch out 😄

The one that enforces item subclasses are named [...]Item 😄

[robust pike]

It took me a moment to realize, but the default components are now datagennable, at least at a glance

Or at least on the way to being so

[paper cedar]

I am still trying to wrap my head around that part yet

At first I thought they did this so default components can safely reference registryobjects

[robust pike]

Ah, yeah

That and holder sets

That would make sense since they first reload the full registries, then build the components, and finally handle the rest

[paper cedar]

So which means they re-evaluate these everytime the regs change (?)

yes, seems like it

[robust pike]

That's my understanding

[paper cedar]

return ReloadableServerRegistries.reload(contextLayers, updatedContextTags, resourceManager, backgroundExecutor)
    .thenCompose(
        fullRegistries -> CompletableFuture.<List<DataComponentInitializers.PendingComponents<?>>>supplyAsync(
                () -> BuiltInRegistries.DATA_COMPONENT_INITIALIZERS.build(fullRegistries.lookupWithUpdatedTags()), backgroundExecutor
            )

We can probably still support that event, but it'd interestingly be 1) more generic since it is not item specific, 2) would re-run everytime registries reload

[robust pike]

Yeah, technically components can be on any registry object, provided you can get the holder reference

[paper cedar]

Oh wait it's not an event yet

Okay. Wait. I am confused.

We patch in modifyDefaultComponentsFrom but it's deprecated

When did we add that?

[robust pike]

I wanna say 20.6 or 21?

[paper cedar]

Oh, so we deprecated that and replaced it with the event, I presume?

Well 26.1 is as good a time as any to yeet the method then

[robust pike]

I believe so

I like how the immutable itemstack is now ItemStackTemplate

[paper cedar]

I like even more how both implement ItemInstance:D

10 rejects remain

[shy valve]

sorry, just saw this. it's more that if i could play peaceful and still have access to everything the game offers i would. So anything that adds new threats without adding some other massive value is personally not worth it.

[paper cedar]

Hm. So....

We previously replaced
state.is(Blocks.FARMLAND) with state.getBlock() instanceof net.minecraft.world.level.block.FarmBlock

But Vanilla has now replaced their checks with state.is(BlockTags.SUPPORTS_CROPS)

I suppose that will make the instanceof redundant, yes?

[visual basin]

I hope so!

[shy valve]

yes and no. it now behooves mods to ensure their blocks are in the correct tags since inheritence is no longer assumed.

[visual basin]

right right

[paper cedar]

Yes, it will introduce bugs where mods forget to do so.

[visual basin]

maybe make a dev announcement summarizing the important new tags when 26.1 finishes?

[paper cedar]

(Removing the instanceof that is)

[visual basin]

or at least any instanceof -> tag updates

[paper cedar]

There is another place where the new tag is not 1:1 sufficient to replace our patch, IMHO and that is for fluids.
For sugarcane, we replaced the hardcoded is(WATER) check with state.canBeHydrated(level, pos, fluidState, below.relative(direction))

[visual basin]

(it's not the first time mojang converted instanceof to a tag and it won't be the last)

[shy valve]

probably worth making a note of every time new tags and registries are added so that modders know to update. as we approach data-driven items/blocks these kinds of features are going to become more and more common and thus easier to lose track of or fall behind.

[paper cedar]

Vanilla now does fluidState.is(FluidTags.SUPPORTS_SUGAR_CANE_ADJACENTLY)

But since our patch is actually level-sensitive, I've left it in

[round spear]

(There are config values to disable spawning, you don't even need to use a custom datapack )

[paper cedar]

Rejects that can be addressed right now are addressed I think

Thoughts on ItemStackTemplate getCraftingRemainder(ItemStack itemStack)?

Vanilla now returns an immutable stack (ItemStackTemplate)

Should we generally make our itemstack-sensitive methods in Item now consume ItemInstance?

As the super-interface of both mutable and immutable ItemStacks?

[plucky hollow]

i for one am in favor of using the new ifaces for super of immutable/mutable and resource/value

[paper cedar]

I am also not sure why you would mutate the given stack in a getter like that

[plucky hollow]

yeah the param should rly be TypedInstance or something
to take in either mutable or immutable, prefering immutable ofc

[paper cedar]

Hm?

ItemInstance is that

It's the super-interface of ItemStack/ItemStackTemplate.