1.20.5/6 Snapshots

[elder swallow]

73 rejects with 80% fuzziness

[slow prism]

60%

[elder swallow]

https://github.com/neoforged/NeoForge/pull/602

wdym by 60%?

default is 50%, 60% feels too low

[slow prism]

love how your PR title is reduce and the comment mentions increasing

[elder swallow]

more quality means less fuzzy

[prisma locust]

why change it here instead of in NG

[elder swallow]

🤷

this is way easier to test and tweak

only 35 rejects with 60%

@wooden tendon do you prefer the fuzziness to be changed in NG or neoforge? - see PR 602

70% gives 55 rejects

the thing is... a single uncaught misaligned patch is so much worse than having to apply a few trivial patches by hand

[prisma locust]

to clarify where are the rejects

24w05b?

[elder swallow]

1.20.4 -> 24w05a

[prisma locust]

what happens with 95% quality

i.e. very strict

[elder swallow]

129 rejects

[prisma locust]

that's not very many

what is the process of resolving a reject? apply by hand?

[elder swallow]

yeah

[summer furnace]

before fuzzy we did all of them by hand :P

[elder swallow]

IMO doing them all by hand is quite reasonable

[summer furnace]

(unless it was always fuzzy and I didn't realize)

[prisma locust]

not for snapshots

I feel like the current fuzzy system would not have the problem you've described if we were actively updating to snapshots, but we lack manpower for that right now

[elder swallow]

it would clearly have that problem

[prisma locust]

the chance of a patch being misapplied in the first place goes down as the difference in the underlying codebase decreases

for instance 24w05a -> 24w05b should be a perfect match for 99.9% of patches

[elder swallow]

this kind of stuff is really hard to catch

[summer furnace]

each version yes, but I would expect that cumulatively, the amount of misaligned patches should be similar?

[elder swallow]

1.20.4 code for reference

[prisma locust]

honestly this just looks like the patches not being created with enough context

like why is that even being applied

[elder swallow]

yes, regardless of fuzziness

[elder swallow]

50% fuzziness is very fuzzy

and yeah that is the patch lol

[prisma locust]

this type of patch should not be applied fuzzily

[elder swallow]

easy to say when seeing the patch

[prisma locust]

there isn't much you could change around it and safely apply it

I feel like that isn't the case for all patches though

[elder swallow]

well, what is the probability of a patch being 1) hard to update and 2) handled correctly by the fuzzy patcher?

i.e. what is the actual difficult work

keep in mind that the patches used to be with SRG names, making applying them very hard

now that the patches are direct source patches it should be trivial to apply them

[prisma locust]

that's true...

[slow prism]

I have a curiosity. what if you do 1% lol

[shut wraith]

I would definitely say preferring missed patches over misaligned patches is probably smart

Especially given that as you've pointed out, it's all Mojmaps now

[elder swallow]

1 reject

[prisma locust]

they're probably all misapplied lol

I guess actually that's really the more important data point

not the quantity of rejects/misapplications

[slow prism]

they're probably injected in random places

[prisma locust]

but rather the number of misapplications that would otherwise be rejected

[elder swallow]

hmmm no there's a problem with the way I am counting the rejects

[shut wraith]

Misapplied patches are unfortunately basically impossible to count except by hand

[elder swallow]

there's 1 reject file and 5 classes for which the target could not be found

random injections yeah...

[restive raft]

ayo, they broke the negative amplifier jump boost effect trick for disabling jumping?

[foggy steeple]

worse, the broken negative mining fatigue to stop block mining without slowing down attack speed

My existence is threaten!

[restive raft]

this is a travesty of justice

[open kindle]

is it time to patch that back in? /s

[fallen igloo]

an interesting experiment would be to do the updating incrementally snapshot by snapshot with 95% quality and compare the number of rejects to doing one direct

[elder swallow]

Incremental update is a pain though

Cherry picking sucks IMO

[kind sable]

So did we get the big new obvious feature?

[clear bane]

hmm yes yes... they now have things working in Vulkan /j

[summer furnace]

no further big features yet. they seem to be polishing the new structure

probably prereleases next week

[shut wraith]

Wait okay here's a funky idea: apply patches with a bunch of different fuzzinesses - say, 50 to 100 by 5s or whatever, I don't know how slow that is. Then, you can get a ranked "how likely is this to be misplaced"

Just based on at what point it started failing

[summer furnace]

that sounds extremely slow

[elder swallow]

That's more or less what I did but not very scientifically

[shut wraith]

I don't know how fast patch application is - I'd assumed it was reasonably fast

But I could be wrong for sure. It's not like there's any recomp/decomp to do at least

[elder swallow]

It's quite fast yes

[summer furnace]

honestly me neither, but I assume fuzzy patching is an order of magnitude slower than non-fuzzy, just from its nature

[shut wraith]

But the idea would be that you'd have a report of, basically, patches ordered in order of misplaced-ness when you're done, then just go down the list checking them

[elder swallow]

Checking each patch is what is annoying

[shut wraith]

And if you made it automated, you'd just run it once, get the list, then check from most to least misplaced

Basically, for every patch you'd have a system that'd start at 100, go down towards 50, stop when/if it matches, and record what you stopped at. Hmm. Gotta see how hard it'd be to get diffpatch to do that

[slow prism]

do it scientifically

[boreal flint]

That is to say, do it the same, but come up with a hypothesis before you start and write down the results after

Problem solved

[summer furnace]

you also need multiple samples, to account for variability

[slow prism]

you also need to test in multiple OSs for line endings

[fallen igloo]

no diffpatch uses \n not the system default

[slow prism]

ahem

[rigid walrus]

Remembers the issue with vineflower where line wrapping was different depending on the OS

[slow prism]

remembers NG using system line separators

which it still does btw

[wooden tendon]

No

That was fixed a long time ago

We give the patcher \n now only

[slow prism]

https://github.com/neoforged/NeoGradle/blob/ecea0a6025c2b6d7468c312373f3933fe2d783a2/platform/src/main/java/net/neoforged/gradle/platform/tasks/BakePatches.java#L20

unless it's explicitly set which it doesn't seem to be

[wooden tendon]

That should not matter as far as I was told

We can trivially fix it

[elder swallow]

pushed a comparison if you want to have some fun: https://github.com/neoforged/Kits/commits/24w05a-patch-experiments/

60% is too loose

65% is too loose too

70% causes an annoying misalign in ChatScreen, so that is too loose too

this kind of misapplication is quite problematic

will probably be caught by the compiler but

(this is 80%)

I think 85% is reasonable

[wooden tendon]

Okey

Then lets use that

PR it into NG7

And I will merge it asap

[elder swallow]

👍

[frail hatch]

So the lower bound on what covers recommended

[elder swallow]

if we get too many misapplications we can always increase it later

you can check https://github.com/neoforged/Kits/commit/f2e5588f242befdccbc8992abadc91d532f30e3e, if you think the misapplications here are problematic we can also do 90%

hmmm there is a record field reordering in CommonListenerCookie

[frail hatch]

I am not complaining, I am just stating covers was right it seems

[elder swallow]

wouldn't hurt to have some double checking

maybe 90% is safer

especially in the cases where we move vanilla logic to CommonHooks

[restive raft]

is https://github.com/neoforged/NeoForge/pull/602 still on the table, or is it superseded by a PR to NG?

[wooden tendon]

Will be moved to NG

[fallen igloo]

NG PR to change the default

[restive raft]

shall I close, or shall I let Tech close it on his own initiative

[wooden tendon]

I added a reject

When ever tech has the time he can then move it himself

[restive raft]

when Tech does that, he should add the background on how the new default was chosen in the PR description (or the commit message)

[elder swallow]

I'll send a PR soon TM, giving it a bit more time to discuss the default value

(otherwise I'll go for 90% if nobody says anything)

[wooden tendon]

Pong me when you have the pr, i will merge it immediately

It is a single value which you tested well

I see no reason to fuck around with that pr

[digital drum]

how do you check the changes from snapshot to snapshot like this?

[spiral radish]

That's just a commit comparing patch fuzzyness over the top of Kits (private NeoForge dev but with source code instead of patches for the initial port to a version, which then get's collated down into a single commit in the main repository)

If you want to actually just compare MC versions then you can use https://github.com/neoforged/snowblower Snowblower, or one of the many other 'mc version compare by git commit' projects (For example https://github.com/WinPlay02/GitCraft)

[digital drum]

how do I run/use snowblower?

[solar zephyr]

https://github.com/neoforged/snowblower/blob/097cba9a960bb14bb95f105d7dda980c4dd010e7/src/main/java/net/neoforged/snowblower/Main.java#L29

[digital drum]

?

[solar zephyr]

¯_(ツ)_/¯

I have never used it, don't ask me for support. haha.
I just looked at the main method, it describes what's all needed to run it

[white frigate]

gitcraft was very easy to setup and run, idk about snowblower

[digital drum]

it's doing every single version

how do I only make it do certain versions

[shut wraith]

You tell it to using the command line args

For gitcraft, read the README

[digital drum]

i'm doing ./gradlew run --args="[--only-version=24w05b --min-version=24w05b --max-version=24w05b --mappings=mojmap]"

[reef parcel]

that doesn't look right you would want it to compare on the --min-version=last snapshot, right?

[digital drum]

./gradlew run --args="[--only-version=24w05b --min-version=1.20.4 --max-version=24w05b --mappings=mojmap]"?

[shut wraith]

Remove the square brackets

I do ./gradlew run --args="--create-version-branches --mappings=mojmap --min-version=1.20.4 --no-external-assets --override-repo-target=minecraft-repo" for my setup

(I also have some custom stuff setup locally, like generating gradle stuff in the generated repo to list dependencies, so it can be imported into IntelliJ)

[digital drum]

how long does it usually take?

[shut wraith]

Erm, depends on how many versions? A few minutes per version? Also depends on your machine

Mine is pretty slow to decomp compared to what some people here use

The log will tell you what it's doing though. First run will take longer as it downloads libraries it doesn't already have and whatnot

[digital drum]

hey, I have ran gitcraft, it generated something, how do I see the differences in the ui like way above?

[shut wraith]

Pick your favorite tool for viewing git diffs. It's just a git repo like anything else

[digital drum]

e.g.?

[digital drum]

like what do you use?

[shut wraith]

gitk? Or GitHub, by uploading it as a private repo? But just use whatever you normally use for git stuff

[digital drum]

ok, got it

[summer furnace]

I wonder if we'll get anything interesting in today's snapshot, or they are ramping down to prereleases

[fallen igloo]

I'm still hoping for brewing recipes and block properties codec

[elder swallow]

If there is a pre release remember to bump NG first

[summer furnace]

no frogs so far but, https://vxtwitter.com/gnembon_mc/status/1755249778449916380

there's a strong implication a snapshot is happening?

[fallen nymph]

https://www.minecraft.net/en-us/article/minecraft-snapshot-24w06a

👀

[summer furnace]

<@&1067092163520909374>

[magic sierra]

Added generic.gravity attribute that controls blocks/tick^2 acceleration downward
haha

[plucky brook]

another (neo)forge attribute obsolete

[fallen igloo]

Hoppers will no longer try to pick up item entities if there is a full block placed above it

[magic sierra]

yeap

[summer furnace]

nice, no more need to put composters on them

[shut wraith]

Beat me to it... finally!

[fallen igloo]

that looks like a revert candidate if that disables picking up items through mud

[plucky brook]

I was gonna say that they already don't but the important part here is probably "try"

[shut wraith]

No need for that silliness with sticking certain things above then any more

[plucky brook]

Mash a hopper minecart in the block

[shut wraith]

Reduced amount of data sent by the server during login by reusing parts of the vanilla datapack
Anyone know what this refers to?

And looks like they're still making the change in how data is cased, with a few more things being all lowercase now

[plucky brook]

Might be that they'll only send datapack stuff if it differs from the vanilla datapack or something

[shut wraith]

Wacky

[plucky brook]

check the code if you want to know

[shut wraith]

I should update my gitcraft repo so I can poke around in it

[fallen igloo]

EVEN LESS HARDCODING FOR EFFECTS

[plucky brook]

hey, tele got his wish

[fallen igloo]

seems like only healing should be left in the hardcoded stuff (we'll see when we have the snowman commit)

https://github.com/neoforged/Snowman/commit/78d6ee9658cc1026afbb49bac4eb8cc4f625a66b

[spiral radish]

I was just formatting that

https://www.minecraft.net/en-us/article/minecraft-snapshot-24w06a
https://mdcfe.dev/mc-changes?ver=24w05b
Snowman: https://github.com/neoforged/Snowman/commit/78d6ee9658cc1026afbb49bac4eb8cc4f625a66b

[fallen igloo]

[spiral radish]

Hoppers will no longer try to pick up item entities if there is a full block placed above it
👀

[fallen igloo]

(this replaces the Lifecycle in registration)

[fallen igloo]

SCROLL UP

[spiral radish]

[fallen igloo]

^

[shut wraith]

I really need to get on my laptop so I can run gitcraft

Cause that looks quite interesting

[plucky brook]

I find it funny how github still hasn't gotten support for record

[fallen igloo]

block pos nbt is now a length 3 int array

[plucky brook]

[fallen igloo]

HWAT is this

[summer furnace]

class to id

what's it used for?

[prisma locust]

Lifecycles are dead apparently

or at least their usage inside MappedRegistry has changed

[summer furnace]

yep was mentioned earlier

[dry stump]

minecraftschurli

[Reference to](#1184607569629675530 message) #1184607569629675530 [➤ ](#1184607569629675530 message)

gigaherz referenced · Feb 7, 2024, 4:17 PM

[frail hatch]

rip all of DFU that references lifecycles

[summer furnace]

they still exist, but more indirectly

[plucky brook]

I've yet to understand what Lifecycle is

[frail hatch]

My guess is it is related to whether to show the experimental warning screen window. Unsure if other uses

[prisma locust]

BlockStates have changed to use an Reference2ObjectArrayMap to store the property->value information

[summer furnace]

Lifecycle so far as I understand, is a way for them to be able to know if the current state of the game contains anything that isn't finalized design

eg, the experimental stuff in the save loading

but also in a more general sense, to show if the current environment is "stable", "beta", or "experimental"

if anything in the game is experimental, the entire game is considered experimental

[prisma locust]

I think this change effectively limits you to ~5 properties per blockstate before performance becomes a concern

[summer furnace]

hmm

but memory stops being a concern?

or is that besides the neighbourhood tables?

[prisma locust]

this is not the neighborhood tables no

[summer furnace]

what did they use before?

[prisma locust]

ImmutableMap

which I believe is a hashmap for large enough quantities of keys

[summer furnace]

what does that use inside?

[reef parcel]

Why the heck would they switch to an array map

[prisma locust]

It likely makes sense for them if they never use more than 5 props per blockstate

[summer furnace]

looks like it's a custom implementation...? this code is a mess

[prisma locust]

looks like the regular one to me

[summer furnace]

wtf is RegularImmutableMap?! the get impl uses a loop?

[prisma locust]

oh you mean guava

[summer furnace]

OH it's just a clunky implementation of a hashtable

with linked list buckets

[digital drum]

You know how I update my gitcraft?

[summer furnace]

this is NOT the place to ask

and no I don't know, I have no clue what gitcraft is

[plucky brook]

ah, it's just a marker

it's just that every time I read Lifecycle I think about rust lifetimes

[digital drum]

Are there any breaking changes?

[summer furnace]

in this week's snapshot?

sure

[spiral radish]

There's almost always going to be breaking changes

[summer furnace]

but they may not affect the average modder much

[spiral radish]

It's just the extent of them

[foggy steeple]

Mojang is watching me lol

[plucky brook]

https://tenor.com/view/mike-wazowski-watchingyou-gif-5352035

[elder swallow]

Did they fix synced entity data?

[frail hatch]

What about registry order for particle types? Given torch blocks. Given iirc they fixed attributes or whatever

[elder swallow]

Another massive snapshot, too much to scroll through the whole thing

[errant lion]

oof

[elder swallow]

1.20.5 is gonna be interesting

[clear bane]

Mojang: finnaly, we can merge those Pr's
we have been staling them for years

[errant lion]

What PR

[elder swallow]

Just like Forge without Lex

[clear bane]

it was a joke

[errant lion]

Ah

[clear bane]

although a lot more Pr's have been merged since the split...

[plucky brook]

I mean, afaik, they do work on github and therefore probably do have PRs

[true socket]

but none years old

[frail hatch]

Internal ones for mojangs repo that we can’t see

[plucky brook]

how would you know

they might

[true socket]

I mean true what do we know

[clear bane]

might be, idk
Mojang works on like... a lot of things
and waits until the time is right

I suppose that some things were a few months to years in the works
like the rendering changes in 1.17 and so on

[frail hatch]

Stream codecs maybe

[clear bane]

I mean... mob votes might have been pre determined somewhat, likely

[plucky brook]

with all the migration, probably

[vast parrot]

Well, what is wrong with those?

[vast parrot]

And those?

[fallen igloo]

it seems like us all

[true jacinth]

I pinged you for it in FC, similar thing to attributes

Some objects (like TorchBlock) hard-ref a particle type, but register before them

The order has to be updated or the usage has to be supplier-ified

idk what tech's referring to specifically with synced entity data though

[errant lion]

Whats a ChaseClient/Server

Im guessing RCON?

[vast parrot]

Oh, this, right. Don't expect it happening in this release

[summer furnace]

I believe that entity data values are very fragile due to them getting numeric indices based on registration order, but I don't know if that's what tech meant

[true jacinth]

There were a lot of weird mixin-application-order issues with that, yeah

[errant lion]

Nope Chase Client/Server not RCON

Dunno wtf it is

DediServer its not either

Anyone know what it is?

[prisma locust]

They are fine for you guys, but the system doesn't make it safe to declare new synced entity data on existing entities if the order isn't consistent on both sides. Something we could of course fix ourselves

[slow prism]

and maybe we should

[summer furnace]

the data value indices can't be synchronized as they exist currently because they have no uniquely identifying name that we can use as the key

[slow prism]

and by we i mean you can pr

[true jacinth]

Well now that attachments aren't so awful (and are synced) its not that necessary

Before that it was the only thing that was able to staple a synced data block to an entity

[summer furnace]

IMO we just should discourage modders from using those values in thirdparty objects (a modded entity implementing defineSyncedData is fine)

[prisma locust]

I have such a long list of things to do already

[slow prism]

attachments are only synced for items

lol

[true jacinth]

sigh

[summer furnace]

I'd propose having a PARALLEL API separate from the data values

[true jacinth]

bad system

[summer furnace]

similar to the vanilla synced data but by name

[slow prism]

with completable futures, right?

[summer furnace]

not that kind of parallel

[true jacinth]

wait whytf are they only synced for items

[summer furnace]

parallel to vanilla

because attachments on items get shoved into the stack when serializing it for packets

no opt-in or opt-out

[slow prism]

yes, attachment sync is full of CFs

[true jacinth]

Well make those synced, problem solved™️

[slow prism]

ok i'll show myself out

[foggy steeple]

It’s not safe for us modders but fine for you. If the serializers are registered by Mojang’s method, forge’s parallel loading can cause the serializers to be out of order on client and server which then causes extremely difficult to diagnose client disconnects. My mod and Ars Nouveau has this issue for so long and took an entire week of big brain modders to figure that one out on what mods are involved and what is happening. Moved to forge’s own registry for entity data serializers which solves that issue

The other issue is people trying to inject add Synced Data Accessor into vanilla entities which then explodes when the synced data accessors fields aren’t classloaded in same order on client as server

[errant lion]

Found out what ChaseClient/Server is

its a system where you have a /lead and /follow and basically allows someone to be in the same spot as you but in different game/world

[plucky brook]

that sounds interesting, but I wonder what's it used for

[errant lion]

Honestly dunno

Be crazy if they made it sync up player actions, if it did that then I could see it being more useful

[foggy steeple]

Multiworld gametests (spitballing)

[errant lion]

Found this in latest snapshot

Dunno what its for

[plucky brook]

Find Usages?

[summer furnace]

is that the first sign of a future datapack items refactor?

[crimson blaze]

That Items class was added a 1.20.3 (When the block codec stuff was first added) and the lead & follow stuff was added when they where redoing world gen

[boreal flint]

this already existed and has for a while tbf

[reef parcel]

I thought we only got the Blocks references back then we got both Blocks and Items?

[errant lion]

I wonder what Mojang has in store for us 🤨

[boreal flint]

java21 yet

[reef parcel]

Pretty please

[boreal flint]

mojang pls

[spiral radish]

Perhaps for recording videos/trailers? Place the leader in a Minecart/water stream which isn’t present in the other world

[summer furnace]

java21 would be nice, but I wonder what the cost is of updating java, like does it mean a certain group of people stop being able to play minecraft?

[quiet condor]

what PC's can't run 21?

[summer furnace]

no idea, that was my question :P

[boreal flint]

this has been in the game for a while...

chaseclient/chaseserver is not new

[summer furnace]

like, I was just watching a video last night about how a recent VSCode update made people panic, because some people were using VSCode Server as the only way to interact with servers, and those servers used an old version of glibc, and vscode uses a newer glibc now

so you never know how a small update fucks people

[spiral radish]

I wasn’t trying to imply that it was, just speculating slightly on the usages

[quiet condor]

oh of course :)

[last badger]

Oh, that's very interesting for FC. What does getValues return now? Map<...> or explicitly R2OAMam<...>? That ImmutableMap has been a massive pain.

[slow prism]

tfw ROAM

[last badger]

That's great news for me/FC!

(ImmutableMap isn't meant to be extended, and even a heavily trimmed version is still a pretty big object at that scale. With Map that's no longer a problem)

[gentle rampart]

Java 21 supports the same platforms as 17, but uses newer build tool versions for them as is usual for a new JDK release. https://wiki.openjdk.org/display/Build/Supported+Build+Platforms
In theory shouldn't have any trouble, in practice however... 🤔

[kind sable]

Hoppers will no longer try to pick up item entities if there is a full block placed above it
👀

[quiet condor]

I think that breaks a bunch of farms

[red beacon]

they couldn't pick up items like that anyway, they just tried

AFAIK

you use hopper minecarts to pick up items above the block above

[quiet condor]

oh then it just improves performance?

[plucky brook]

I assume so

it sounds like a if (blockAbove.isSolid()) return;, but one would have to check the code

[red beacon]

In theory at least, I guess we'll have to wait and see any potential issues that might arise

[summer furnace]

basically up until this snapshot

to avoid hopper lag you had to put some inventory on top

originally was dispensers or droppers

then it became composters

because they are cheaper to make

the presence of the inventory disabled the AABB entity check

with this snapshot, it seems like a solid block also disables it

hopefully only a full shape block only

and it doesn't break all the farms that rely on having soul sand or such

which are not quite full size

[rigid walrus]

hoppers check blockstate.isCollisionShapeFullBlock

Hopper minecarts seem to also have the new check

[boreal flint]

yeah... it'd probably be too costly to do anything other than that

since you'd have to do additional math on the sizing and stuff

or some arbitrary rule like occlusion culling or something

[fallen igloo]

you are the third ... people scroll up

[foggy steeple]

Did you know that hoppers now wont check if solid block above?

[plucky brook]

https://media.discordapp.net/attachments/670545060537958401/1134609441052971119/ablobglarezoombutfast.gif

that sentence is so wrong

[turbid flume]

can someone with the decompiled source look and see if they're doing anything new or clever for the wind charge rocket jump fall damage calculation? I do a fair amount of fall damage cancellation/manipulation

[rigid walrus]

They have a new field named ignoreFallDamageAboveY that is type Double in Player where if the player takes fall damage while the field is set to a value that isn't null the fall damage is instead based on the value of the field

[shut wraith]

That's... a way to do that I suppose. I guess not that strange

[reef parcel]

OptionalDouble cries itself to sleep

[shut wraith]

Is that an actual type that java has? Huh

I mean yeah, that's what I'd do assuming it exists

But... I guess they didn't

[reef parcel]

yep, avoids null and boxing/unboxing

[shut wraith]

Well it's still boxing/unboxing

[reef parcel]

java util has double, long, and int types

[shut wraith]

It's just in the creation of the optional object instead of the creation of the Double object

But allocation/whatever wise it's going to be equivalent to boxing. The lack of null is the real nice bit imo

[reef parcel]

i mean, it's a creation of an object, but no gross castchecks if you're doing double myDouble = this.doubleObj;

[shut wraith]

No, you have to call getOrElse or whatever instead

But I totally agree, that would be more sensible for this use case

[reef parcel]

anything's better than using a nullable primitive wrapper obj :p

[rigid walrus]

they store it as a Double but when reading it gets the value as a float

[shut wraith]

Okay that is cursed

...KnownPack looks fun though. Looks like packs are now namespaced, finally!

Lemme know if you want my feedback on any of the changes that'll have to happen to my system for mod packs when you port it. Shouldn't be too bad luckily

But the switch from a string ID to PackLocationInfo lets you express versioning stuff in the pack, which is pretty cool... but probably doesn't quite play nice with mods that have a string for a version

I'll have to think about it mod built-in packs can make use of that at all

But if they could it could drastically decrease what has to be synced, in theory

Packs also declare how they get positioned a bit smarter

But instead of an ID, packs now have a PackLocationInfo that holds:

• an ID
• a title
• the source for the pack
• optionally, info for finding a known pack - which takes an ID again alongside a namespace, and a string version

So actually I'm wrong, we could make mod data/resource packs use that and avoid syncing unnecessarily

Cause the mod version can actually be stored there

We'd just use something like neoforge:modid:modversion for th KnownPack

Gotta think about what you could do with this system more. I can already think of all sorts of stuff I want to add to DynAssetGen

Unrelated to that system - WTF is ClassTreeIdRegistry

Just, counts the use of a class... ah, something in synched entity data. I still don't quite get it to be honest, but alright

Yep, okay, first thing I plan to focus on with 1.20.5 is getting the mod data packs using this system - avoiding syncing stuff unnecessarily with this would be amazing

And if I'm reading it right might just be as simple as providing something for that Optional<KnownPack>

[rigid walrus]

Also the ignoreFallDamageAboveY is still used even if the player didn't go above that y coordinate

[kind sable]

Hoppers will no longer try to pick up item entities if there is a full block placed above it
👀

[boreal flint]

hey guys did you know that hoppers won't try to pick up items anymore if there's a full block above em

[slow prism]

hey guys did you know that I can ban you all /s /jk

[boreal flint]

can't ban me if I have a full block above me

[rigid walrus]

The new check for hoppers does also affect hopper minecarts

[kind sable]

wait... that might not be intentional... many farm rely on the cart being under the block and picking up items from above

[restive raft]

looks to be unintentional, based on https://bugs.mojang.com/browse/MC-268349's continued existence (and assignment to a Mojangsta)

and the "Very Important" priority

[shut wraith]

Okay poked the KnownPack stuff today with fresh eyes and it still seems cool, so that's a good sign. Gotta think how to abuse this in DynAssetGen... and I'm now very curious how they keep going with this

[plucky brook]

what's the short version of what that system is?

or seems to be

[turbid flume]

Thanks for looking into the wind charge code for me. ignoreFallDamageAboveY isn’t as extensible as I would hope, but probably would be useful for making some of what I do feel more natural. Right now I do a lot of cancelling fall damage entirely

[shut wraith]

Basically: data packs can optionally have a namespace, ID, and version attached. When syncing datapack registry entries, if the datapack entry came from a datapack that the client has (that datapack has one of those ID/versions and it matches), it'll use the one from the client instead of syncing

Now, currently that's just applied to the built in datapacks because only those have a version

[plucky brook]

ok, so I was right, but not fully

[shut wraith]

But it could easily be added to mod datapacks too

[plucky brook]

I'll also have to use that for my mod

[stiff galleon]

ahhh, so instead of flinging all the vanilla biomes to the client, the client just uses the vanilla biomes it already has

[shut wraith]

Yep! And they conveniently made the "version" a string instead of an int or something, so it should be comparatively easy to use for mod built-in packs

Which.... yeah I can imagine that'll be a good thing when blocks get datapacked

[stiff galleon]

okay, so datapacks can be loaded clientside now... can we abuse that for fallback tags?

[shut wraith]

But it's also all optional so stuff from datapacks, or stuff that overrides the vanilla biomes, will still get synced fully

[shut wraith]

not easily, as tags don't use this system yet. Oh, you mean just for loading the actual fallback tag? I mean yeah probably, I assume it's opening a ResourceManager for data clientside

[plucky brook]

because the current system is Not Great™️

[boreal flint]

does this mean that datapack content is cached clientside now and doesn't need to be synced every login ?

that seems like a fairly substantial performance improvement

[shut wraith]

Not... Quite. It's not caching, it just uses local stuff if it has it. Registry entries only right now, and they're the biggest ones probably. Presumably, other stuff could use the system later on but right now doesn't seem to

In non-modded, the only packs this'll do anything for are the built in ones - normal resources, and presumably experimental stuff too if a server has that though I haven't checked

[boreal flint]

lame =/

that has potential

[shut wraith]

It doesn't cache anything, it just reuses local stuff if it's from the same datapack

[shut wraith]

But in a modded environment, neo could definitely (pretty easily) make it work with mod datapacks too

Since those also can easily have an ID and version

[shut wraith]

But to answer your original question: datapack registry entries from vanilla are no longer synced

And in vanilla that's the only difference you'll see currently

[true socket]

aside from that, under the snapshot thread on the Feedback server, a mod confirmed it as an issue

mind you, they aren't devs, but they also probably wouldn't reference it if it was wrong

[magic sierra]

here

https://github.com/neoforged/NeoForge/pull/26 Is this thing still useful in Minecraft 24w05a?

[magic sierra]

related bug report: https://bugs.mojang.com/browse/MC-193656

[elder swallow]

what are they doing to their item transfer API

ahah ok that is actually a cursed bug report

[summer furnace]

I assume it's related to the change they made for full blocks?

[fallen igloo]

yes

[edgy arrow]

Oh course, because those drop as an item, right?

[elder swallow]

yep, item entity

[restive raft]

prime candidate for a game test

[kind sable]

I do wish they would publis the tests they have internally

[fallen igloo]

yea so we can verify that we don't break stuff

[edgy arrow]

i wonder how many tests paper / something that cares less about 1-1 compat breaks

[clear bane]

IIRC they do?
Or on bedrock?
I remember I was able to use their game tests thingy at some point

[elder swallow]

they publish the framework but not the actual tests

[clear bane]

Hmmmm
Then maybe on bedrock?

[frail hatch]

I wonder if they would be willing to setup some sort of endpoint for mod loaders to basically run tests against without having to expose the implementation, just to see what fails (even though I guess that wouldn’t be that helpful without knowing what the failed tests are meant to do)

[elder swallow]

so we upload our own custom version of mc to their servers, they run it, and we get the result back???

[frail hatch]

Something, or they somehow send an encrypted set of instructions that can be ran

I don’t know

Was just thinking out loud

[prisma locust]

they might publish the real tests down the road

[frail hatch]

I was sort of just thinking how azure and some web based things do tests

Has anyone actually asked if they would be willing to do so, like to not have it that users have to download I understand, but as a separate jar/library it would be quite useful

[errant lion]

data driven tests

[shut wraith]

I have those for one of my mods

[errant lion]

nice

[shut wraith]

Best yet, the data comes in the form of an image

[errant lion]

what does the image look like

[shut wraith]

https://github.com/lukebemishprojects/BiomeSquisher/blob/1.20.2/common%2Fsrc%2Ftest%2Fresources%2Ftest%2Flayouts%2Ftemp_continentalness_1.png

[errant lion]

huh

what does the test do exactly?

[shut wraith]

Tests what specific slices of the biome space look like after biomes have been injected into them

[errant lion]

neat

[shut wraith]

The JSON defines where the slice is, the png defines what it should look like

[errant lion]

hm

[restive raft]

neat!

do you also have the created PNG saved for manual comparison, when the test fails?

[shut wraith]

No, but I have a command in game to export those slices in the first place, plus a debugging web server that can be started to browse around through the entire biome space

[clear bane]

So what will be today's world blue yonder they put in the snapshot

[digital drum]

there's a snapshot today?

[naive hound]

usually one a week, so it's expected

[digital drum]

is there a set day of the week?

[naive hound]

I thought it was Tuesdays, but that may have changed

[white frigate]

usual day is wednesday afaik

[fallen igloo]

i was tuesday at some point but now it is wednesday at ~ 16:00 GMT

[clear bane]

back in the day, we also had fridays

[quiet condor]

those were secret Fridays, no?

[clear bane]

no idea
but these were the changelogs from back in the day
they didn't chagne change a lot of the format: https://web.archive.org/web/20140207055427/https://mojang.com/2014/02/minecraft-snapshot-14w06a/

[spiral radish]

They seem to flick between Wednesdays and Thursdays, depending on whether their ready or not

[clear bane]

I have seen mondays as well
just for snapshots
now it seems to be pre releases on monday though

[spiral radish]

PreRelease/RCs do tend to be Mondays

[digital drum]

if nones here to show the diffs, someone just ping me here, I'm so free you wouldn't believe it

[clear bane]

people will use Snowman for that

[digital drum]

how do you use snowman? I use gitcraft just because idk how to use snowman

[clear bane]

2 things

1. pings
2. no idea

[digital drum]

srry

[spiral radish]

Snowman is the automated (private) repo, Snowblower is the tool

We have a TeamCity setup which automates it

[digital drum]

got it, can I use it on my local computer

[fallen igloo]

yes but if you have gitcraft already that is not really needed

it's an alternative

[digital drum]

is there a guide? maybe I might switch one day

[fallen igloo]

if no one wrote one in the last few weeks then probably just the main class code

[clear bane]

or just read the script that automates it

(I guess)

[white frigate]

A lot of mob AI has changed. This should mean nothing, but may actually mean something. Don’t sit on slimes.

back when did they more jokey/secretive changelogs

[digital drum]

today it's going to be 24w06b right?

[quiet condor]

iirc it's week 7 so it'll be 24w07a

it's based on week number and year

[summer furnace]

year 24, week 07, release "a"

[digital drum]

got it

[quiet condor]

yeah that

[digital drum]

btw gigaherz can you check the 1.20.4 parchment PR, afaik, you're moderator there

[summer furnace]

Im a discord moderator there yes, but I don't see myself as a maintainer for Parchment

[digital drum]

ok

[elder swallow]

Raise the alarm

https://www.minecraft.net/en-us/article/minecraft-snapshot-24w07a

[digital drum]

time to build

[restive raft]

<@&1067092163520909374>

[elder swallow]

A new variant of Skeletons that shoot poisonous arrows

[slow prism]

<@&1067092163520909374>

[restive raft]

wee woo wee woo

[prisma locust]

interesting solution to the hopper problem

[restive raft]

block tag

[spiral radish]

https://www.minecraft.net/en-us/article/minecraft-snapshot-24w07a
https://mdcfe.dev/mc-changes?ver=24w07a
https://github.com/neoforged/Snowman/commit/380c7467d37bdc1679f2d423107a71351356eec5

[shut wraith]

Panda sneeze loot table? Funky

[spiral radish]

ERROR: Class net/minecraft/core/RegistrySetBuilder couldn't be processed.
java.lang.NullPointerException: Cannot invoke "org.jetbrains.java.decompiler.struct.gen.VarType.getArrayDim()" because "varType" is null
...
ERROR: Don't know how to handle non independent org.jetbrains.java.decompiler.modules.decompiler.exps.InvocationExprent

[digital drum]

gitcraft better lol

[fallen igloo]

ha I knew it

[spiral radish]

Snowman built, although VineFlower errored

[restive raft]

i'm sure coeh will report it to VF once he's awake

[fallen igloo]

hmmmm

[restive raft]

also, I'm starting to be tempted by poking Snowblower to add a command-line option (in some form) to ignore assets

[spiral radish]

dew it

Or at least something to skip the .nbt

[digital drum]

the registry access creates the registry ops now looks like

[fallen igloo]

please do it github becomes laggy with such big commits

[restive raft]

how to do it best is a question, hmmm

perhaps an exclusion filter

--exclude assets/?

[digital drum]

LIGHT THEME ALERT

[spiral radish]

The tags and other json changes are usually semi relevant

[digital drum]

HolderLookup.Delegate is an interface now

[restive raft]

someone compile a bulleted list of things to exclude -- paths (and/or maybe extensions)

[spiral radish]

I feel like it's mainly just .png and .nbt

[digital drum]

?????

what's a BOGGED

[restive raft]

i think I'm also going to stab the TC config for snowman

[spiral radish]

Spino could you not?

[elder swallow]

The new mob

[restive raft]

because I hate this "Failed to load build settings from VCS"

[summer furnace]

nice, new mob

[elder swallow]

What's the matter?

[digital drum]

sorry for that btw

[spiral radish]

Large flashes of white on my screen suddenly, also the fact that that's quite a bit of code posted in screenshot form

If it's relevant than sure

[restive raft]

no flashbangs in the chat

[digital drum]

srry srry srry

Bogged is a new monster, seems to be a new skeleton that shoots poison arrows?

[summer furnace]

yep that's what the release notes say :P

[digital drum]

BlockPathTypes -> PathType

[elder swallow]

Wait what happened to RegistrySetBuilder

Was it really deleted??

[restive raft]

where was this outputted?

so I can link to it

[digital drum]

nope

[elder swallow]

Looks like snowblower failed to decompile it

[digital drum]

still there

[spiral radish]

https://teamcity.neoforged.net/buildConfiguration/MinecraftForge_PrivateRepositories_Minecraft_Snapshots/2612?buildTab=log&focusLine=695&linesState=22.23.43.50.95&logView=flowAware

[spiral radish]

Yep, I mentioned it above

[restive raft]

👍 i'm going to give the folks at VF a heads up, while coeh's not yet here to investigate it

[digital drum]

new field in GameTest

[restive raft]

@rigid walrus FYI

[stiff galleon]

MC-267732 - /place jigsaw still caps max depth at 7
👀

[elder swallow]

coeh: Decompilation error in RegistrySetBuilder, see above

[digital drum]

the teleport command seems to have changed considerably

LookAt is a functional interface

and a record which holds the teleport position?

[spiral radish]

Interesting isGridAligned added to Hoppers for bypassing the full block check as well as the new tag, the MinecartHopper sets this to false

[kind sable]

Panda based slime farm??

ohh the new kinds of animal cruelty we can do with this

tho it is 1/700 chance for each sneez and you have to get only sneezing pandas...

[royal stone]

Those have always been a thing, just too slow to be practical

[reef parcel]

They dropped slime before, it just wasn't a loot table

[kind sable]

booooo ruining my fun yall

[royal stone]

Unless you're doing a peaceful only run, then they are the only option

[reef parcel]

No fun allowed :p

[kind sable]

Also I feeeeel like Boged now spawn even if the experimental stuff is disabled...

encaseStructure is interesting

[foggy steeple]

I made my pollen block and item make pandas sneeze when it touches them lol

You can have perpetually sneezing pandas of any type

[prisma locust]

bumblezone is clearly a game-breaking mod smh

😛

[quiet condor]

so seeing the past snapshots, a bunch of combat changes, which is awesome

pls mojang update combat

gegy please

[misty crow]

Hell yeah, that's going to be amazing for tests which check light levels for example, which I have a bunch of and which I currently have to encase in a stone box

[fallen igloo]

nope it uses barrier blocks

[cloud raven]

gestures at the general concept of modding

[misty crow]

Worst case I mixin the placement and removal in-dev to use something else instead. Or make a PR to make it configurable ¯_(ツ)_/¯

[shut wraith]

👀 I mean I mixin to vanilla datagen to pull out the vanilla biome parameters during a gametest, so, not that weird

[frail hatch]

Yeah pr to just add an overloaded method that accepts the block to surround with

[misty crow]

It would have to be an additional parameter to the gametest annotation in addition to that since the method is invoked automatically when the flag is set on the annotation

[frail hatch]

Ah

[plucky brook]

Can't you just mark the asset files as generated via .gitattributes, which would still generate them, but not render the diff on GitHub unless you specifically ask it to?

[lean anchor]

I just want 'em to finally fix BlockPos#betweenClosedStream(AABB)

[shut wraith]

Ooh that's a thing? I need to look at that

[open kindle]

aw hell nah, they added the jungle skeletons from MCD to base MC

[plucky brook]

See https://github.com/PlatinPython/VFXGenerator/blob/1.20.1/.gitattributes#L7

[shut wraith]

Huh

Gotta use that for my gitcraft setup then

[plucky brook]

Ends up like this

No clue if other git repo viewers support that, I just know GitHub does

[restive raft]

yeah, XFact recommended that and I implemented in https://github.com/neoforged/snowblower/commit/107d245fbeac2c33374f45615b43fffa7c2e9339

[plucky brook]

Whelp, I got sniped

[restive raft]

BEGONE Consumer<String> logger

I say, BEGONE

[plucky brook]

Oh, you don't need the =true?

[restive raft]

implied

the syntax exists because =false

[plucky brook]

I see, I only ever checked the GitHub docs, and those do use =true while the linguist docs don't

[rigid walrus]

Seems like the issue doesn't affect neoform

[restive raft]

it's been suggested to me that it might be that Snowblower is using an outdated VF

i'll have to poke that

[rigid walrus]

I think it's still using forgeflower

[restive raft]

I screm if so

[restive raft]

someone give me the name of an MC class that I can look at which has very long lines (that are linewrapped by VF)

ah, PlayProtocols I think is a good candidate

er. GameProtocols

[restive raft]

@frail hatch @misty crow @fallen igloo @spiral radish what are we excluding from Snowman? .nbt and .png?

or any particular directories?

[slow prism]

snbt*

[restive raft]

so **.png and **.snbt
anything else?

they're .nbt

[shut wraith]

There's both

Well, no clue about snowman

Does it run the debug datagen to generate snbt files?

[restive raft]

it doesn't run anything

it only decompiles

[restive raft]

okay, I've pushed out the Snowblower update

i'll be regenning Snowman again

i'll rename the current dev branch to dev-pre-2.0.14, for the sake of keeping Ater's links intact until we decide to yeet it

[rigid walrus]

gametests now have an option for skyAcccess

gametest areas are now enclosed in barrier blocks (excluding the roof if skyAccess is enabled)

The polished andesite has been changed to bedrock

[restive raft]

it's generating now

but from the start

i'll keep an eye on it while I can

[restive raft]

..whoops

forgot about that

oh well.

oh yeah, ought to mention it before I forget

I also paused the other build config for the Releases, so we aren't hammering the server with two decompilations in parallel

[rigid walrus]

the Hopper interface has a new method named isGridAligned

hoppers will check for a solid block if isGridAligned returns true

[summer furnace]

I assume they did that so minecarts can pull through solid blocks

[rigid walrus]

BlockPathTypes has been renamed to PathType

[restive raft]

looks at Snowman shit.

should I care enough to poke the indentation to 4 spaces

also, yay now GH is more speedy looking at the commit

FYI, seems there's a bit of trouble with the naming of lambda parameters in the Snowman repo

[maiden prairie]

:waves:

is snowman still on FF and not VF

[restive raft]

I moved it to VF

I might be missing a command-line option

I'll have to check with NeoForm

also, welcome back Shrimp

[maiden prairie]

oh I'm only here in passing

college is a bitch after all ^.^

[restive raft]

amen

[maiden prairie]

and 2 jobs

@.@

[restive raft]

👀

[maiden prairie]

how many did you have to remove because they don't exist from FF -> VF 😆

[restive raft]

I didn't need to remove any, surprisingly enough

just a quick substitution of the dep (and the corresponding string in the Tools class) seemed to work all fine

though I'm still in the middle of understanding Snowblower's layout, and perhaps making some things more simpler if possible

[prisma locust]

hello shrimp 👋

[maiden prairie]

hi

I was scrolling twitter and saw discussion of a new mob and I was like "oh, it's snapshot day"

[restive raft]

and seeing if we can make the configuration better

[maiden prairie]

tbh it did kinda get spaghetti'd pretty quickly from invention...

the config is very shit

I can agree on that point even when forgetting the entirety of how I (mostly) designed it

[restive raft]

yea, I was starting to see that yesterday while poking the code

but I'm sure nothing that a few refactors and redesigns can't handle

[rigid walrus]

If you want 4 spaces you pass -ind= to vineflower as a single argument

[restive raft]

i'll have to remember that when I next poke at Snowblower

the decompiler args are hardcoded, iirc

[maiden prairie]

correct

[white frigate]

the fact it is not -ind=" " feels so wrong, how does it know which space to split on

[restive raft]

I think there's some automatic quoting going on

err, actually it might be because it's executed not through a shell, so the command-line args are passed pretty much directly

[summer furnace]

probably ends up as "-ind= "

[restive raft]

yea

[white frigate]

has to be lol

[wild pine]

https://twitter.com/slicedlime/status/1760309382837891566 - no snapshot today

slicedlime 💙💛 (@slicedlime) on X

Due to some technical issues earlier this week, we are not releasing a snapshot for Minecraft: Java Edition today. The issues have since been resolved and you can expect snapshots for 1.20.5 to resume next week, barring any unforeseen circumstances.

Until then, happy mining!

Twitter · Feb 21, 2024, 2:23 PM

[foggy steeple]

Someone deleted the minecraft GitHub repo

[naive hound]

junior dev: "Hey! i found out you can silence a lot of git issues by just adding --force to your command! "

[plucky brook]

branch protection rules

[naive hound]

Sounds like a workplace that doesn't trust their employees. smh. /s

[fallen igloo]

Let the porting begin (https://github.com/neoforged/actions/actions/runs/7991951323/job/21824349414)

@elder swallow The work meeting is still going I can probably start at ~17:30 CET

@slow prism

what happened here?

[misty crow]

It tries to find an artifact with the snapshot version twice in its name: neoform-24w07a-24w07a-20240214.192032. The actual artifact only contains it once

https://maven.neoforged.net/#/releases/net/neoforged/neoform/24w07a-20240214.192032

[fallen igloo]

does the issue come from NG or the action

[slow prism]

neither, it comes from your inputs

you only give it the timestamp of the version

i'll run it again

[fallen igloo]

tech started the workflow not me

[slow prism]

well then tech gets a stabby

[fallen igloo]

tech

[fallen igloo]

118 reject files

[restive raft]

movement on Kits 👀

[elder swallow]

yes it's my bad

[fallen igloo]

I'm gonna join voice in ~20-30 minutes and start with the porting

[elder swallow]

👍

I'll have to leave in an hour, so gonna start with the simple rejects now

[misty crow]

I'll join in on the fun

[elder swallow]

joining vc now

[prisma locust]

curious why now is a good time to port?

we dont have a prerelease yet

[fallen igloo]

who is gonna take the role for porting lead for this one (I'd do it if nobody else wants)

[prisma locust]

not saying don't or anything, I'm just genuinely curious

[elder swallow]

I find that role a bit weird personally

it's a bit early to decide anyway, depends on people's availabilities close to the releae

[restive raft]

having someone coordinate efforts is a good thing, I think

[slow prism]

no please no more porting leads/dictators, ||i still hate the first version of that||

[restive raft]

having someone who can give an estimation of the work remaining

a coordinator of sorts

usually the person who's invested the most time in porting, or have regularly been porting

unless every porter is equaly capable of that role, we should have someone to at least be acknowledged as being "the most likely to be that"

[slow prism]

anyone can do that

it's a matter of seeing how many rejects are left in the dir and how many compile errors there are

and you can't predict who's going to port "the most" in the first minutes

[restive raft]

it's a volunteer role -- someone says "hey, i'll coordinate porting" and unless anyone objects, they have it

and it's mostly informal

and checking the amount of compile errors does need access to Kits and cloning it locally to check

i'd think their main thing would just be keeping this channel up to date on the state of porting

as other porters come and go, and might not remember to poke here with what they've done

[elder swallow]

in practice it can be a porting dictator

which is what I do not want

[restive raft]

only if we give them that much power

maintainers giveth, maintainers taketh away

[slow prism]

the role of subproject leader has given people the dictator power and still is

[restive raft]

it's not an on-high appointment that vests them with any sort of dictatorial power

it's essentially the phrase "a private in action outranks a lieutenant at rest"

comparing the role I have in mind to subproject leaders is a bit apples-to-oranges

[foggy steeple]

So what happens if two maintainers port the same thing by mistake due to no coordination?

[restive raft]

for one, subproject leaders are explicitly defined in our governance modelm per https://github.com/neoforged/governance/blob/main/core.md#subproject-leads

I have no intention of making a "porting coordinator" role be a subproject lead (or at least, at the moment)

[slow prism]

you don't coordinate who ports what

fixing rejects and compile errors is not really a "linear" thing

[restive raft]

besides, the reason why subproject leaders effectively are dictators, despite our governance model specifically not allowing that, is more because we don't have any defined and practiced ways to hold their decisions accountable

[slow prism]

or don't want to

[restive raft]

that is a topic of discussion for governance, which is an area I've bowed out as I've previously indicated

so direct any concerns to Curle if you think the governance model is insufficient, as she wrote the thing

(and is also out of scope for this thread)

[elder swallow]

not a fan of code ownership myself

[elder swallow]

If you need to drop a small Todo somewhere: https://hackmd.io/S718YKv-TnWrRKB4nzYV_Q

(big stuff please on the org project)

[fallen igloo]

down to 93 reject files

[misty crow]

*81

[slow prism]

schurli can't count

[fallen igloo]

schurli didn't pull since xfact pushed

[slow prism]

schurli deserves a stab pat

[misty crow]

@elder swallow you want to get into VC again? I'm also dabbling around with the networking rejects right now

[slow prism]

don't do a 1am vc

[misty crow]

Why not?

[elder swallow]

happy to join vc in ~30 min

[misty crow]

👍

[slow prism]

although for the 2 of you it's barely past 12

but you get the point

get some sleep

[open kindle]

will be joining as well for emotional support™️

[elder swallow]

@frail hatch new itemstack method that makes me remove all your copy attachments patches 😄

[frail hatch]

how useful

[fallen igloo]

50 reject files left

gonna start in a few

[fallen igloo]

do we still need the patched in Ingredient#fromJson now that everything is codecs?

[elder swallow]

probably not

wanna join vc?

[fallen igloo]

yea

@slow prism have fun re-doing the game event frequency datamap stuff

[slow prism]

it's a dp reg now?

it isn't

so that method shouldn't be a problem

gotta setup a workspace

[fallen igloo]

it is now ResourceKey based

[slow prism]

use is now useItemOn and returns a ItemInteractionResult

[slow prism]

@true jacinth pls fix finalizespawn

kthx

[restive raft]

gogogo

[slow prism]

ok so

NBTUtils#writeBlockPos now writes blocks with an int array with 3 ints

do we want to keep using that method in forced chunks and break old saves

[restive raft]

make a datafixer

[slow prism]

no

i'll just keep our old behaviour

[fallen igloo]

make the old as a fallback for reading

[slow prism]

93 errors left

most are networking

hmm

no

that patch is going away

[fallen igloo]

yay one supplier patch less

[slow prism]

81

[misty crow]

I've got part of those fixed locally

[slow prism]

tech started fixing networking

hmm

ig it's time to change overlays

[fallen igloo]

oh good maty is doing overlays

[slow prism]

ew we need so many ats

[fallen igloo]

wildcard my friend, wildcard

[slow prism]

extendedgui is

[misty crow]

Honestly, comment out ExtendedGui entirely for now so we get something compiling and then we can fix that afterwards, it'll take a bit of finesse to fix that properly. Though, in the end we may be able to yeet ExtendedGui entirely

[slow prism]

wait where did screenHeight and screenWidth go

[slow prism]

probably a good idea

hmm we probably need a chattype for system

let's see what we're left with after commenting out extendedgui

100 errors

i went up

whoops

that's probably the overlays tho

also who volunteers to make a porting guide

neighborChanged -> handleNeighborChanged

Screen#screenWidth and Screen#screenHeight are now guigraphics methods

also removed fluidstack methods in FBB and added a streamcodec

60

BlockEntityRenderBoundsDebugRenderer is losing me

yeah besides that the only things left are networking errors

(and obv fixing overlays)

and since i'm bored i'm really tempted to look into overlays

[elder swallow]

@misty crow I also started working on networkign locally

I wasn't sure what you had done

[misty crow]

Heh, that switch from PoseStack to Matrix4f is going to confuse a lot of people. It should be as simple as making a new PoseStack from the matrix, maybe the event should provide that though

[slow prism]

mhh yes

[summer furnace]

read either, write new

[slow prism]

and how is that done

[summer furnace]

mojang has been wanting to move away from blockpos-as-long for a long time

[slow prism]

i should clarily that writeBlockPos used to write {X, Y, Z}

not long

[summer furnace]

ah

🤷‍♂️ then

[slow prism]

oh god, LayeredDraw is so inextensible

[elder swallow]

it is fine to extend

we just need to add a parameter to the .add method that identifies the lambda

[slow prism]

oh also RegisterGuiOverlaysEvent ordering was based on current registered overlays so if you wanted to regsiter before another mod's overlay you had to make sure that your listener fires after

quality™️

[slow prism]

yeah i know

i'm just trying to think of a design that makes sense

as in whether to expose the LayeredDraw and then make it handle sorting

also

another problem

this method

does making the layereddraw have an id make sense or should that method specify an id

[summer furnace]

uh the original registration system accepted resource locations so you could reference a mod's layer by name... if the event doesn't allow it, the feature was "lost" during the change over

[slow prism]

if the overlay you're depending on isn't registered yet that's -1

it needs topsort

[summer furnace]

that was supposed to be processed after all mods have had a chance to register....

[slow prism]

although topsort is probably overkill

[summer furnace]

I feel like something has been lost since my initial impl :p

granted, I don't know when it was lost

[misty crow]

Should be as simple as this:

PoseStack poseStack = new PoseStack();
poseStack.mulPose(event.getPoseStack());

[slow prism]

should i subclass layereddraw or add to it

[elder swallow]

I'd say add to it

[plucky brook]

So a datafixer 🙃

[slow prism]

hmmmmmmmm

hmmmm

RenderGuiOverlayEvent is also a problem

this needs a hefty amount of patches

[shut wraith]

I can poke the new resource pack features as soon as it's all released and PR something to use those for mods if you all would like

[slow prism]

ugh add(LayeredDraw) is a problem

this is very annoying

we'd need to fire an event with 3 layereddraw instances

[prisma locust]

I don't think they are completely done the PoseStack changes, I wouldn't refactor the event yet

[slow prism]

opinions?

discord

[prisma locust]

did you just yeet an OnlyIn off Layer?

[slow prism]

harold didn't notice that

i can't say i love it

but it would get the job done

@elder swallow thoughts?

[elder swallow]

IMO add without ID should be illegal

but idk what exactly you are doing

I think you should just preserve the event-based registration

[slow prism]

that requires extending, cause mods can use layereddraw themselves

[elder swallow]

ah hmmm

then maybe we extend it yeah

[slow prism]

an event-based registration would also mean that we need to massively patch the vanilla add layer calls

which has never lead to issues

since the layereddraws are recreated every time

I don't like any solution tbh

[misty crow]

What do you mean it's recreated every time? The Gui gets instantiated exactly once

[slow prism]

well I meant it's recreated when a gui is

what I wanted to do was to fire an event with the 3 layereddraws in the ctor

mojang finally makes something sane in their code and manages to screw us. classic

i mean whatever we do it's likely going to be cleaner than the old way of replacing the gui

but still ugly

[misty crow]

Gui is the HUD, it is never recreated

[slow prism]

that's still not what i meant smh

it's not a "static" thing anymore

[fallen igloo]

it's even final in Minecraft

[elder swallow]

jesus

[frail hatch]

Generics!

[foggy steeple]

Java generics were a mistake….

[shut wraith]

Unfortunately, I've seen worse. I've seen generics so bad that IntelliJ thinks they're fine when they're not

Never fun to deal with generics pain

[frail hatch]

Yeah, I’ve written generics like that before as well, as well once wrote generics that both IntelliJ and Javac thought were fine but the code that Javac produced crashed on execution because it was really really not okay or something (I forget the specifics)

[median wasp]

I've caused exceptions in LambdaMetaFactory with code that Javac compiled fine

Though ECJ was smarter and knew it was an error

[frail hatch]

Yeah this wasn’t even lambda meta factory, just generics with & in them. That produced an error type I had never seen before and on googling stack overflow said it was a javac bug that it even compiles

[fallen igloo]

that is easy because the MH invoke methods are special in the JVM

[frail hatch]

ECJ is a bit stringent on generics as well that some things valid in IntelliJ and javac it doesn’t allow

[median wasp]

For my one, all you need to do is create a lambda with multiple interfaces (a normal thing you can do), but then add a SAM abstract class (not interface) as well

[shut wraith]

I present: https://github.com/lukebemish/codecextras/blob/main/src%2Ftest%2Fjava%2Fdev%2Flukebemish%2Fcodecextras%2Ftest%2Fpolymorphic%2FPolymorphicSubClass.java#L16. IntelliJ thinks the type arguments are redundant. They're not

[median wasp]

Javac doesn't care that it's an abstract class because it's still SAM. But LMF on the other hand actually checks that all supertypes are in fact interfaces

[median wasp]

Oh I've had this precise issue

[frail hatch]

I have that suppressed or redundant diamond expressions in quite a few places in mek

[shut wraith]

Most places IntelliJ gives you it it's actually redundant. Just not here

[frail hatch]

I know

I only have it in places where javac fails to compile after IntelliJ said it is fine

[shut wraith]

IntelliJ just doesn't like how cursed my type parameters get there

[median wasp]

I've had a place where IJ gave me a redundant type args warning in Kotlin, but removing the args resulted in the compiler crashing

[shut wraith]

IntelliJ had the most useless generics errors when I was writing test cases for that system...

But granted I think that's my fault for abusing codecs and lambdas to the degree that I did

[misty crow]

@elder swallow networking done, we should be ready to rebase it on the main porting branch and merge it back in

[elder swallow]

yes

ok nice

[true jacinth]

why has maty handed me so many pings

surely finalize cannot be that broken

[foggy steeple]

Ping pong

[fallen igloo]

rejects done

networking done

[misty crow]

Networking is merged into the main porting branch now

[slow prism]

the spawn tag is gone

[true jacinth]

spawn tag?

You mean the MobSpawnType?

[slow prism]

no the compoundtag

[true jacinth]

oh

that's fine

it was useless for like 95% of things anyway

Just remove the param and update the coremod then rerun find targets

[elder swallow]

find targets was never ported to NG btw

[fallen igloo]

we have patches now

[magic sierra]

?

[misty crow]

We don't know yet, we aren't able to compile the whole thing yet

[magic sierra]

k

[fallen igloo]

3 compile errors

2 of them because registerMapping is gone

the third is interaction events

2 compile errors left in ItemStack

@slow prism you made them

[slow prism]

yeah i realised i changed the wrong interactionresult

[fallen igloo]

no but that event is used in 2 places on uses InteractionResult the other ItemInteractionResult

[slow prism]

sigh

you're right

fuck the interaction events

well they need splitting that's for sure

the hard part

how do you name them

[stiff galleon]

is it finally time to refactor all the interaction events

[slow prism]

we should probably redo the interaction events after the port

but we need a temporary non-bad solution

[true jacinth]

Well whatever you do I'll probably have to redo later

so just make them compile

[slow prism]

the compilation errors keep popping up after fixing others

[misty crow]

I have ItemStack and ServerAdvancementManager fixed on my end, I'll push that in a second

[slow prism]

i have them fixed already

push race time?

tho how did you fix the event

[misty crow]

-if (e.isCanceled()) return e.getCancellationResult();
+if (e.isCanceled()) return e.getCancellationResult().result();

[slow prism]

that makes more sense

[misty crow]

I won

[fallen igloo]

maty the testframework has 30 compile errors