So, first impressions of the new snapshot (23w31a), Mojang made a code style guide.
#1.20.2
wise forge
lethal bane
wait, a style guide?
stray scroll
WHAT?!
lethal bane
also someone get maintainers in here <@&1128776937410670663>
stray scroll
all the king's horses and all the king's <@&1128777120068423680>s
lethal bane
seems like I can do that too 
fallow sundial
everyone can do that
feral moat
What do you mean by made a code style guide…
lethal bane
I can't ping the snapshot people
fallow sundial
@wise forge clarify or 
wise forge
Most of their codebase has essentially been reformatted into a particular style
arctic sphinx
Probably looked at the changes in Snowblower?
wise forge
Either that, or vanillagradle has done something different this time
fallow sundial
how would you know?...
wise forge
Which is annoying
Which, you know, probably is the case
Because I'm just tired today...ugh
arctic sphinx
arctic sphinx
Either that, or vanillagradle has done something different this time
Unless it's actual code changes, then yes, it's likely VG (or a change in whatever tool your using)
wise forge
Welp, time to regen everything again
nova sundial
so uhh, when did neo plan to do the "hard break"? Was it 1.20.2?
arctic sphinx
afaik Yes
As of 1.20.2 (or really it was 47.1.3 since Forge had BCs) we're no longer guarenteeing compat
lethal bane
when mojang breaks most mods (i.e. now 1.20.2 confirmed)
muted hemlock
I remember somebody floating a plan to treat the first 1.20.x version with feature previews as the start of the "unstable" period, and the last 1.21.x version without feature previews as the stable version / LTS candidate
thorn tiger
Either that, or vanillagradle has done something different this time
VG uses VineFlower (previously QuiltFlower) as the decompiler now, might be due to that?
wise forge
Probably, though currently this is telling me I should sleep before writing the primer
fallow sundial
sleep is good
tacit vale
@wise forge I updated IntelliJ and in reader mode it now adds fake spaces to get to 4 spaces per indent
You can turn this off to get back to three spaces by clicking Reader Mode in the top right
fallow sundial
psure VF is 4 spaces
inland mesa
3 spaces shudder
inland mesa
I remember somebody floating a plan to treat the first 1.20.x version with featu...
yup that has been the most sane idea so far. 1.20.1 stays as LTS candidate to replace 1.19.2, 1.20.2 becomes "latest"
(if 1.20.2 ends up with annoying changes or upcoming features)
tacit vale
Have y'all been discussing the contents of the snapshot??? It's crazy!
inland mesa
it has some very interesting changes wrt protocol
at least on paper
haven't seen the code
tacit vale
Agreed
wise forge
<@207184927236554752> I updated IntelliJ and in reader mode it now adds fake spa...
Ugh, thanks
lean warren
You can turn this off to get back to three spaces by clicking Reader Mode in the...
You can disable that without leaving reader mode
wise forge
Well, I'll write up the primer later tonight once I get a bit of my head back. I'll help with the private stuff later
lean warren
You can disable that without leaving reader mode
mossy bobcat
I do love that new feature
lethal bane
so that's why reader mode broke 
mossy bobcat
It's great. No more ugly weird spaces in generated sources
lethal bane
now I have a different problem 
tacit vale
It's great. No more ugly weird spaces in generated sources
what do you mean, imo it causes more ugly weird spaces
cuz the s paces it adds are literally fake/invisible/can't be copied
mossy bobcat
what do you mean, imo it causes more ugly weird spaces
I mean, it means I don't have to read the generated sources with 3 spaces anymore, so I won't complain
lethal bane
so that's why reader mode broke <:kek:1134098707869073408>
alright, the main problem is that it doesn't add the line breaks as I have defined them
so it looks very broken
see #squirrels-🦊 message
formal plume
wow
even just reading the changes took a while
there are some really interesting changes here
dull bison
even the discord post is long
formal plume
the stuff they are doing with commands is really interesting but not sure how many people actually use it
been a while since i saw vidoes of command creations
I already have a usage idea for the overlay stuff if we manage to expose it for mods
it could be use to toggle optional content
mossy bobcat
Oh yeah, definitely. Currently you have to register your own pack source thing for that. This would be an amazing system for that
formal plume
(my mod would have a setup where it is technically possible to completly break the game if enought time is spent.......(as in switch to creative). so those features could be in an overlay and disabled on servers)
lethal bane
🍞❓
muted hemlock
afaict the main intention of overlay packs is optional compat for specific minecraft versions
e.g. you make a pack that works on 1.19.2 and 1.20.1 and you have an overlay pack that uses worldgen stuff from 1.20.1, in 1.20.1 the overlay pack is automatically used, in 1.19.2 it's ignored
if we could expand that it might be something we could use with iconditions, having overlay packs that load if and only if a specified mod is loaded
lethal bane
you could also make a pack that works on all future version by just making it only out of overlays
muted hemlock
I already do that though
I haven't updated my pack.mcmeta from 6 in years
minecraft doesn't actually care
lethal bane
yeah, it's mainly for users
formal plume
afaict the main intention of overlay packs is optional compat for specific minec...
Yeah on a second loot you are right
tranquil salmon
fallow sundial
1.20.2
tranquil salmon
if (!($$6 instanceof FlowingFluid flowingfluid)) {
return false;
} else {
FlowingFluid flowingfluid;
Block $$7;
boolean $$8;
boolean flag2;
label82: {
blockstate = p_150717_.m_8055_(p_150718_);
$$7 = blockstate.m_60734_();
$$8 = blockstate.m_60722_(this.f_40687_);
if (!blockstate.m_60795_() && !$$8) {
label80: {
VF do be something 
muted hemlock
oh good more fluid changes
tranquil salmon
```java
if (!($$6 instanceof FlowingFluid flowingfluid)) {
return...
That is actually with ForgeFlower, currently waiting for VineFlower to release 1.9.2
fallow sundial
huh
lean warren
ForgeFlower is having a field day with this snapshot. A bunch of new records are blowing up completely, one of them being GuiSpriteScaling.Stretch: https://media.discordapp.net/attachments/104299991589199872/1136319971618717767/image.png?width=688&height=691
lean warren
tranquil salmon
I think this is meant to be a choice statement
nova sundial
let's ask fry /s
kindred fractal
vineflowern't moment
fallow sundial
flowern't'n't
tranquil salmon
https://github.com/neoforged/NeoForm/blob/1.20.2-dev/versions/snapshot/1.20/23w31a/patches/shared/net/minecraft/network/protocol/game/ClientboundChunkBatchStartPacket.java.patch#L25
When the thing creating the diff wants to leave in the indented close bracket to close the class instead of the unindented one even though I only left in the unindented one
tacit vale
@tranquil salmon we should have a flag in FART (now ART) to restore the acc record flag!
try enabling that
and make sure you're using all the neoforge libs that have been moved over if possible
I know at a minimum ART
tranquil salmon
https://github.com/neoforged/AutoRenamingTool/blob/main/src/main/java/net/minecraftforge/fart/internal/RecordFixer.java
Seems like it only readds components
tacit vale
Fix it then u should have commit access
if u don't I'll give it to you
if u feel like it u can also ping me to give it a sanity check
tranquil salmon
Fix it then u should have commit access
if (isRecord && (access & Opcodes.ACC_RECORD) == 0) {
this.madeChange = true;
access |= Opcodes.ACC_RECORD;
}
4 line fix
fallow sundial
push then
tacit vale
```java
if (isRecord && (access & Opcodes.ACC_RECORD) == 0) {
...
you still messed it up smh
not the actual fix but the part you didn't include, the jopt change in Main, is wrong
I guess I didn't articulate it which is my bad, but I wanted a commit link on a fork somewhere or a full diff but this is fine it's just minor but probably good to fix still
tranquil salmon
the configuration start packet calls clearClientLevel
which does this:
this.gameRenderer.resetData();
this.gameMode = null;
this.narrator.clear();
this.updateScreenAndTick(p_294558_);
this.gui.onDisconnected();
this.downloadedPackSource.clearServerPack();
this.level = null;
this.updateLevelInEngines((ClientLevel)null);
this.player = null;
SkullBlockEntity.clear();
As in: you added the string documentation change to the wrong jopt parameter
wise forge
Oh hey, don't you love chunk classes? Well, too bad! They're all called sections now
fallow sundial
lovely
wise forge
Also, gui sprites have been flattened into individual textures. The actual big texture is now stitched using a texture atlas
tacit vale
yeah that I knew from the changelog
muted hemlock
wait, chunk sections were the 16^3 things
what are the tall things called now
SectionStacks?
frail oriole
They use a texture atlas to generate Gui screens now?
tacit vale
Jenga time
tacit vale
They use a texture atlas to generate Gui screens now?
the misc gui assets yeah I haven't checked the code but I assume the backgrounds of like container screens is still the same and not stitched
all the mega asset files got split up
amazing for pack devs
frail oriole
Ok cool
It sounds like it might be possible to use this to assemble custom sized player and chest screens
tacit vale
why does it sound like that to you?
wise forge
Image scaling in guis that determines stretching, tiling, and nine_sliced is now data driven in the .png.mcmeta
tacit vale
as far as I'm aware this is just regular atlas sticking like how the block atlas work and you don't actually merge textures with this
tranquil salmon
Using a \ then a new line doesn't allow you to insert a new line character into a command
sturdy phoenix
Has BootstapContext been fixed
lean warren
Nope
true pivot
interesting note I discovered while testing modernfix 1.20.1 on this snapshot: the handleUpdateTags packet listener method doesn't seem to fire anymore when joining a singleplayer world
not sure if this is just a misapplied mixin, or if tag syncing has changed
lean warren
handleUpdateTag() is a Forge addition, vanilla calls BlockEntity#load()
true pivot
`handleUpdateTag()` is a Forge addition, vanilla calls `BlockEntity#load()`
I'm referring to the method on ClientPacketListener (1.20.1 name), it exists in vanilla as well (sorry for ping)
lean warren
Ah, yeah, that changed a bit. Both sides now have a "common" packet listener (i.e. ClientCommonPacketListenerImpl), a "configuration" packet listener (i.e. ClientConfigurationPacketListenerImpl) and a "play" packet listener (i.e. ClientPacketListener) and the latter two extend the former. If you want to react to the packet in both config and play phase, then you need to target the common listener and not the specific one
true pivot
thanks mojang
wise forge
oh good more fluid changes
Seems like the fluid changes are just to catch some edge cases with how fluid logic early exits sometimes
wise forge
Ok, here's a primer: https://gist.github.com/ChampionAsh5357/933bdc1028e42e85313e7b58c0a70d39
tawny iron
splitting the sprites is interesting
and very nice
yayy... the absolute bare minimum expectation of a not-shit mobeffect system
winged hull
1.20.2 is already out?
sturdy phoenix
is the chunk -> section rename maybe the beginnings of a 3d chunk loading optimization?
tranquil salmon
new debug command only available when IS_RUNNING_IN_IDE is set to true: debugconfig
It basically just let's you put a player in or out of configuring mode
valid variable names for macros can only contain the characters [a-zA-Z0-9_]
mossy bobcat
Those function macros are cool and I'm definitely going to abuse them a bit
tranquil salmon
function that parses a string to an int
$return $(num)
the DataFixTypes thing on SavedData is concerning, it's not clear how to get around that without just passing a random vanilla fixer there
tawny iron
configuring mode?
tranquil salmon
tawny iron
what's it for
true pivot
datapack registry stuff
quiet talon
does this mean biomes reload now 
tacit vale
without looking at the code it looks like reconfiguring stage is what proxies have been dreaming for
tranquil salmon
without looking at the code it looks like reconfiguring stage is what proxies ha...
Seems like it
tacit vale
Ok, here's a primer: https://gist.github.com/ChampionAsh5357/933bdc1028e42e85313...
"atkes"
no idea what that word is but you can do a ctrl f to find it in ur primer
and "addition integer" I assume should be additional integer
"Calle when ..."
quiet talon
Calle 
stray scroll
I'm guessing "atkes in" should be "accepts"
so it would be "#readPackInfo also accepts an additional integer to get the directory overlays..."
Better Calle Ash
tacit vale
btw ash this is great work
I have no idea how u have the motivation to doc all this stuff
this is gonna be a hellish port
I think it's best to wait for a few more things to get into 1.20.1 pipeline before I/someone else attempts it
esp. fixing server install + probably deferredholder
also perhaps waiting on vineflower to be ready? idk tho that would be even more work to port to...
stray scroll
this reminds me, I think I have yet to fix that parameter annotations bug in AutoRenamingTool which I discovered before the Forge kerfuffle
wise forge
I have no idea how u have the motivation to doc all this stuff
I’m a documenter by trade, it’s literally what I’m pushing my doctoral dissertation to be on
dull bison
I challenge you to document DFU
fallow sundial
professional docusaurator
wise forge
I challenge you to document DFU
I did a while back, pretty sure it got nuked with my watered laptop though
dull bison
proof or it never happened
wise forge
Fair enough, I guess I’ll just have to do it again
bold steppe
Doesn't the overlay stuff kind of solves the loot table and advancement issue we had the other day when discussing conditions?
kindred fractal
Well it only toggles based on resource pack version atm
wise forge
Ok fixed the typos...probably
lunar summit
net.minecraft.client.GuiGraphics#blitNineSliced, #blitRepeating, and #slices have been renamed and privated.
FFS Mojang, I just got done documenting GuiGraphics
wise forge
They're data driven now :p
dull bison
They're data driven now :p
tf?
shadow widget
you define how they stretch/slice/repeat in mcmeta, now
muted hemlock
ewww
what if I want to nineslice a texture in more than one way
#getTotalChunks -> #getTotalSections
does that actually return the number of cubic sections instead of the number of stacked chunks now
normal wigeon
I love how I've been busy implementing actual Ninepatch support in images
Oh well, that means easier to implement Ninepatch support
wise forge
> #getTotalChunks -> #getTotalSections
does that actually return the number of c...
I don't think so yet. I believe the size is still set to a chunk
muted hemlock
SavedData$Factory
SavedData instances are now constructed through a $Factory record. This takes in the supplied constructor to construct the data when none is available, the function deserializer to construct the data from a tag, and the DataFixTypes for fixing the data when an error occurs. This currently cannot be null.Both DimensionDataStorage#computeIfAbsent and #get take in a SavedData$Factory.
bruhhhh
half my SaveDatas dont' even serialize anything, I just use them to stick keep transient data around
guess we'll need to patch that in any case
unless making a noop DataFixTypes is easy
lethal bane
DataFixTypes
and
cannot be null
mossy bobcat
unless making a noop DataFixTypes is easy
Shouldn't be too hard I'd think. I'd need to poke it a bit though
lethal bane
I mean, I have no problem providing a datafixer if I knew how
I'd like to use them even
muted hemlock
oh DataFixTypes is an emum
err there's already a DataFixTypes for savedata in 1.20.1
DataFixTypes.SSAVED_DATA
what would happen if you gave that to a nonvanilla savedata factory
lethal bane
muted hemlock
I don't have 1.20.2 yet
lethal bane
fair point
kindred fractal
there's not even a fabric yet, doubt there would be a neofroge so early 😛
- I'd say getting neoforge to work is likely higher prio than neofroge
true pivot
DataFixTypes.SSAVED_DATA
it no longer exists in 1.20.2 and got split up into a bunch of vanilla saved data fix types
muted hemlock
oh no
lean warren
what if I want to nineslice a texture in more than one way
Then you don't provide an mcmeta at all, build the nine-slice parameters in code as needed and call the relevant method yourself with that data. The new metadata stuff on GUI textures is surprisingly easy to work around from what I have gathered from looking at the code
feral moat
At least for now
sweet herald
I would like an ELI5 on 1.20.2
- is it a breaking update in terms of mod support?
- what exactly has changed to make compat break?
- (likely very stupid question) what took so long between 1.20.1 release and the first 1.20.2 snapshot?
shadow widget
for the third one, summer vacations in Sweden
sweet herald
Bork
formal plume
What do you mean by the first?
If you mean what will change code side for mod makers. Than apart from the massive rewrites that mojang is doing there are many ideas that are being considered for improving NeoForge.
For those ideas look around in this ( #1105595318197825557 ) forum channel
sweet herald
I want to know the full context for both, but more so the former
formal plume
We don't have a fix plan yet. As we don't know what and how soon Mojang will change. Like if they release the next version in two weeks than nothing major breaking will be merged in time.
sweet herald
I am happy to see the changes on the snapshot, despite them not affecting me all that much, and I am also happy to see Mojang update progress pick up again, as there are a number of modmakers that are waiting for a 1.xx.2
tawny iron
why are they waiting on a .2?
formal plume
We are still a volunteer based project so what gets done will depend on who is available and how long.
So giving time estimates is really hard. And because of that giving change info is hard as well
But again currently the best way to see what change ideas are being brought up is to read the discussions in #1105595318197825557
Some of the notable ones are (as far as I see):
#1135328752658829312
#1131130393144336385
#1136448404495532133
sweet herald
why are they waiting on a .2?
Can't blame them, but my guess is that they are holding out on updating mods until it gets to a extended-support-likely sub-version, which tends to be a 2
tawny iron
or it could be a .1
formal plume
But these are just the NeoForge things. We have no idea what Mojang will cook up this time
formal plume
or it could be a .1
Or .12
sweet herald
But these are just the NeoForge things. We have no idea what Mojang will cook up...
Fish
Inb4 the next snapshot actually updates something about the sealife
formal plume
Nah people are guessing desert cause of a Dev twitter post that had some stuff in the background
sweet herald
Apple cake? Waffles? Kanelbulle?
formal plume
That would be dessert 🎂
muted hemlock
Can't blame them, but my guess is that they are holding out on updating mods unt...
I think it's more likely that 1.20.1 will be the "stable" version for 1.20.x and 1.20.2 starts the "unstable" period
analogous to 1.19.2 and 1.19.3
1.20.2 is going to break every mod that does gui stuff I think
sweet herald
Another reason that I can guess is that they are waiting specifically for NeoForge
tawny iron
1.20.2 is going to break every mod that does gui stuff I think
we got more gui changes?
what's the sting this time
muted hemlock
we got more gui changes?
remember back in like 1.7 or whenever it was when mojang broke up the block atlas into individual 16x16 png files
quiet talon
thats only for the misc widgets
muted hemlock
mojang's doing this to the gui widgets
(just the widgets, not the big background pictures)
quiet talon
it won't break that much unless you were relying on drawing the vanilla texture file
muted hemlock
all my mods with guis draw the vanilla texture file
quiet talon
you're custom drawing vanilla widgets?
muted hemlock
yeah, like the arrow here is the vanilla furnace arrow widget cropped and copy-pasted via creative blitting https://camo.githubusercontent.com/cf527f3409ab831a69bff13a0f65e9361914513f3c2b2695075d729a6c3b0ea7/68747470733a2f2f6375626963696e746572706f6c6174696f6e2e6e65742f696d616765732f6d6f72657265642f676174656372616674696e675f6775692e706e67
muted hemlock
the background texture is the villager trade screen
fallow sundial
screem
muted hemlock
why make my own art when vanilla ships with perfectly good art
lean warren
Unless I'm completely blind, the empty arrow is not on the GUI atlas and is instead baked into the individual backgrounds. The full arrow is on the GUI atlas
fallow sundial
just make all of your blocks crafting tables
muted hemlock
oh right nvm I'm probably blitting from the furnace gui lol
lean warren
The GUI atlas also contains a bunch of duplicates
tacit vale
The GUI atlas also contains a bunch of duplicates <:screm:980923567833219093>
Do you have an image?
lean warren
I can give you the full atlas if you want
lethal bane
did you already port the mod?
lean warren
Nope, but vanilla has an export feature since 1.19.3 or 1.19.4
lethal bane
👀 it does?
tacit vale
Where?
lean warren
Yes, press F3 + S ingame and all atlases and dynamic textures (i.e. maps) get exported to screenshots/debug/
tacit vale
Oh cool
lean warren
lethal bane
which one?
lean warren
Close to top left
lethal bane
ah, that one
lean warren
The stoplight ones are Realms expiry indicators
lean warren
I did exactly once 😄
lean warren
In a separate texture. The GUI atlas currently only contains widgets
bold steppe
Hm thought the point of having them in an atlas was less texture swapping
lean warren
I'm fairly certain that for the texture switches to matter in a UI context, they would first have to stop uploading after every little thing. A nine-sliced texture currently does at least 9 individual uploads to the GPU
bold steppe
Yay. To think about it now they swap even more as before they were mostly all on same texture per each screen
Guess performance in a gui isn't much of a concern
lean warren
My guess would be that their goal is to not need an individual background texture for most GUIs and instead draw a nine-slice of the "demo background" texture. The issue with putting GUI backgrounds on the atlas is that the size of the atlas could very quickly explode
bold steppe
I wonder if it would still compare to the modded block atlas tho
I mean mods can add just so many blocks that that thing gets huge
lean warren
Mods also add a ton of GUIs and their size is usually significantly larger than a single block texture
tawny iron
this is a pretty good change imo
allows for people to customize individual sprites without conflicting
and even opening up those same sprites to animations
bold steppe
Yes nice snapshot. Just wish villagers would have been merged more. Zombie villager change was definitely needed but librarian one just seems like a shift of meta..
muted hemlock
Mods also add a ton of GUIs and their size is usually significantly larger than ...
just do what I do and blit different parts of the vanilla guis to make new guis
winged hull
exactly
dull bison
just do what I do and blit different parts of the vanilla guis to make new guis ...
for vanilla like guis this will probably be the way
tawny iron
it won't even change afaik?
it's just now you reference icons individually instead of the whole sheet
lunar summit
Why are there 5 arrow progression buttons with red X's over them 
Crap forgot to turn off ping, sorry XFact >.>
dull bison
Why are there 5 arrow progression buttons with red X's over them <:Thonk:6653222...
because they still need to dedupe the gui textures
bold steppe
Isn't it because they are diff textures so packs can change them individually?
wise forge
^ That's what my thought was
vague otter
is that the horse inventory grid?
lean warren
Possible, no idea
sweet herald
So are you guys satisfied with how the directory system is working?
I am not complaining, but I have a few ideas/thoughts I would like to toss out
muted hemlock
which directory system
sweet herald
- Is it viable for library mods to be in its own directory, or at least a sub-directory of
/mods/?
stray scroll
i'm guessing the pack overlay system?
ah, other thing
should we wait for all the ideas to be tossed out first
sweet herald
- I am not sure as to how tightly things need to be bound to the instance data directory. For instance, shader packs can benefit from being in a "universal" directory, and have the thing in the instance directory be a reference file
stray scroll
1) Is it viable for library mods to be in its own directory, or at least a sub-d...
iirc, any kind of changing in where mods are loaded from, as in anything other than the direct children files of the mods folder, has been disfavored in the past
one of the reasons is that most players use a mod/modpack management tool, like CurseForge, and so this wouldn't have effect for them, and would possibly complicate modpack managers who now have to consider this new folder as well (like how modpack managers have to consider the coremods folder from bygone eras)
stray scroll
2) I am not sure as to how tightly things need to be bound to the instance data ...
as far as I understand, we follow on the idea of the Minecraft launcher on installations -- each installation is separate and self-contained, with nothing being shared between installations
(the assets, libraries, versions, and other similar info are handled by the launcher, as part of the global .minecraft to reduce downloading everything anew, but even then the launcher can be configured via command-line iirc to hold separate assets/libraries/etcs. in another directory)
sweet herald
iirc, any kind of changing in where mods are loaded from, as in anything other t...
Although the distinction of coremods was comically pointless, I feel that the distinction between libraries and core/content mods is quite clear cut
Library mods are designed to help out other mods, and do not do much of anything on their own
mossy bobcat
Yeah, no, it's not that clear cut
Take DynAssetGen, a mod I maintain. it gives tools that allow other mods to generate assets or data at runtime based on existing assets/data - this turns out to be pretty useful for all sorts of stuff. In this regard, it is a library mod. And indeed, if you just installed the mod, you'd notice no difference. However - DynAssetGen offers features that could act as standalone content too, depending on how you look at it - it gives tools that resource packs (or end users, or modpack makers, etc.) could use to generate their own stuff at runtime, without needing to write a full mod. It's sorta like, I dunno, CraftTweaker or something in that respect, which definitely isn't a "library mod".
This is one of the more benign examples I can think of
I've got other mods, such as JsonWrangler, that tow this line even more. Telling what should or shouldn't be a "library mod" is anything but obvious
But yeah - any sort of distinction you try to make like that will run into countless weird edge cases. Best to just keep everything in the mods directory - you avoid pain for modders, users, and modpack makers alike that way
quiet talon
iirc, any kind of changing in where mods are loaded from, as in anything other t...
looks at ServerPackLocator 👀
wise forge
Agh, too early in the morning
mossy bobcat
Game no longer uses numeric values when storing mob effects to world (so, for example,
4becomesminecraft:mining_fatigue)
wise forge
Oh hey, no more packet workaround for forge then
mossy bobcat
Mob effect instances now all seem to have text IDs
At least from the article
Haven't poked the code or anything yet obviously
muted hemlock
> Game no longer uses numeric values when storing mob effects to world (so, for ...
we can 
the forgeregistry
formal plume
this one looks smaller than the last one
wonder how long this snapshot period is gonna be
muted hemlock
I'm guessing they'll snake more fields in the upcoming weeks
datafixerupper go brrrrrrr
formal plume
yeah i wonder how much internal changes this snap has
frigid mountain
still quite a bit
formal plume
i'm kind hoping they do a bigger UI rewrite... more usable widgets and stuff
lethal bane
Gameplay packets are now packed into bigger TCP-packets to reduce overhead from TCP-headers, significantly reducing network usage
does that mean that the max payload size is now bigger than 1MB?
kindred fractal
not necessarily
it just means they merge small packets into a larger packet
or it could just mean that they increased the MTU idk
lethal bane
well, if someone could look at the new value of ClientboundCustomPayloadPacket#MAX_PAYLOAD_SIZE then we'd know
true pivot
<@&1067092163520909374> new snapshot dropped
is there a way for me to get the snapshot alarm role?
true pivot
thanks!
daring ginkgo
I was also gonna ask that 😅
inland mesa
\o/ Game no longer uses numeric values when storing mob effects to world (so, for example, 4 becomes minecraft:mining_fatigue)
one less use of numeric IDs
muted hemlock
Was that the last one?
I thought biomes were but I checked and they use a per-section palette by string id
true pivot
biomes got changed a while back (1.18 iirc)
kindred fractal
last registry that fabric persisted to disk... https://github.com/FabricMC/fabric/blob/1.20.1/fabric-registry-sync-v0/src/main/java/net/fabricmc/fabric/impl/registry/sync/FabricRegistryInit.java#L43
arctic sphinx
Yes, and done
Apparently that role does more than I thought
I'll leave it to the team to poke
inland mesa
damn if there's nothing else saving IDs to disk that'd be one less feature we have to preserve in #1131130393144336385
naive ridge
It was the last one. The method to register with fixed id was removed from vanilla
The Flattening is over!
true pivot
so IDs are finally a runtime thing only now?
stray scroll
bust out the champagne (or a non-alcoholic equivalent)!
gilded halo
(how long has it been since the flattening started? 1.13 released in 2018 apparently, damn)
muted hemlock
to be fair I also just implemented a feature people have been asking me for since 2018
(pipe filters that can filter more than one item)
rare yacht
inland mesa
it's 2023, 2018 is 5 years ago
the issue is that 2018 doesn't feel like that long, not that it isn't ;P
pine ember
I think registry id persistence can be removed now 👍
inland mesa
at this point the only reason to keep forge registries is registry replacement?
inland mesa
cos DeferredRegister + vanilla can deal with everything else
pine ember
Still need to sync the raw ids.
inland mesa
I have mixed feelings about replacement. I think it's a good feature on paper, but there's so few places where it can be used without bringing huge downsides along...
stray scroll
theoretically promising, practically useless
inland mesa
yeah but that can be done by patching / extending the vanilla registries
stray scroll
just like many ideas!
gilded halo
maybe this is me primarily being a fabric modder but I don't really see the use-case for registry replacement? modifying vanilla classes with mixin, etc, seems less invasive than fully replacing them
inland mesa
I'm still of the opinion that mixins should be a last resort, and not the main tool your reach for
if something can realistically be done without them
but with registry replacements, at best, you get to a "last man wins" situation
where if two mods try to replace, only one can succeed
wise forge
Looks like a lot of fun things this update
stray scroll
the theoretical of it is that instead of coremodding into a e.g. block class to modify its behavior, you replace the block with your own version (extending the original block) which does your own stuff
essentially, the key points are no coremodding, and easier to do a future hot-reload since the only thing needed is to fire the registry events
jagged agate
gilded halo
oh right
ngl I forget that hot-reloading is a goal of (neo)forge
jagged agate
i rather have mixin conflict and get an error report than two registry replacing with one silently overridden and no one can debug why one mod isnt working
stray scroll
I'm not sure if it's a goal anymore
gilded halo
it just seems infeasible to me, when mods are fundamentally just a bunch of arbitrary code
fallow sundial
at this point the only reason to keep forge registries is registry replacement?
replacement is going yeet alongside forge regs
muted hemlock
I'm still of the opinion that mixins should be a last resort, and not the main t...
the difference of opinion here being that registry replacement is even more of a last resort than mixins?
inland mesa
I stopped using registry replacements when they started brekaing worldgen
so can't argue
muted hemlock
I guess if people really want to registry replace vanilla stuff they can mixin into the register calls in Blocks
inland mesa
Blocks are like the worst thing to replace though XD
any other registry replacement is less worse than blocks
wise forge
So, main things around is mob effects and loot tables/criterions transitions to codecs: https://gist.github.com/ChampionAsh5357/933bdc1028e42e85313e7b58c0a70d39/revisions#diff-485198454aa7c9ab60942b4f45e19b22d6f3a9c40c5d118c3f5212b1ba14f9b2
inland mesa
nods
muted hemlock
are loot tables still a reloadlistener?
mossy bobcat
I have mixed feelings about replacement. I think it's a good feature on paper, b...
You can do the same thing with mixins and it's less likely to break other mods, so... I think we're probably fine, which is good, because registry replacement causes weirdness
mossy bobcat
So, main things around is mob effects and loot tables/criterions transitions to ...
Loot tables are codec stuff now? Sweet!
wise forge
are loot tables still a reloadlistener?
I believe so
jagged agate
vague otter
> Game no longer uses numeric values when storing mob effects to world (so, for ...
everywhere? [suspicious stew, packets, etc]?
jagged agate
everywhere? [suspicious stew, packets, etc]?
vague otter
huh, i don't think stews could even have multiple effects before
rich void
huh, i don't think stews could even have multiple effects before
Well that is pretty cool
rich void
How do u know they are cleaning it?
muted hemlock
How do u know they are cleaning it?
because we read the snapshot changelog
rich void
I read it too partially
muted hemlock
also some people have decompiled the snapshot and found things like, the big switch statement in MobEffect is replaced by effect subclasses overriding stuff
rich void
Oh that is better
dull bison
also some people have decompiled the snapshot and found things like, the big swi...
they are finally making use of inheritance 🥳
rich void
What's that?
sweet sundial
Oh lovely!
rich void
(I am a newbie when it comes to coding, I must say)
sweet sundial
In the past it was all hardcode in a base class, now each class can have a unique action.
quiet talon
after 20 versions that shitcode is finally gone?
rich void
Nice
inland mesa
What's that?
to provide a slightly more complete explanation: in Object-Oriented programming languages like Java, you define a "class", which is a description of the structure of an object. you can then use this class to construct objects of that type. but for what matters here, a class can inherit from another class, which means the class builds on the parent class, reusing parts it needs and adding/replacing things that must be different.
this means that you can have a skeleton of a class (what they call an abstract class), and have multiple subclasses which each "complete" that parent class with the logic they need.
if you have done any modding as your profile suggests, you probably have seen extends on classes, and @Override on methods
tranquil salmon
https://github.com/neoforged/NeoForm/actions/runs/5814455748/job/15764396860
7 javac compile errors and 3 ecj compile errors
Somehow javac thinks the parameter for inventoryTrigger might be ItemPredicate.Builder[]
lean warren
As far as I can tell from looking at that in Snowblower output, it seems to fail to infer the generic type of the toArray() call. If I add an explicit ItemPredicate generic to that call, then it shuts up
tacit vale
> Game no longer uses numeric values when storing mob effects to world (so, for ...
Wow that is great for the registry overhaul
That was the last vanilla holdover that stores int ids to disk!!!
This means we can entirely remove the serialization of int ids to world save
Mods that rely on it can get fucked
tacit vale
I thought biomes were but I checked and they use a per-section palette by string...
This could've been my fault. I incorrectly read the pallete container code when biomes only use it for networking. And I talked about this in the registry rework thread but already corrected it
tacit vale
last registry that fabric persisted to disk... <https://github.com/FabricMC/fabr...
That checks out. Glad my investigation is correct.
Ok so after reading the backlog everyone came to the same conclusion ages ago.
xD
tranquil salmon
They changed some instances of !var.isPresent() to var.isEmpty()
vague otter
Mods that rely on it can get fucked
i've already got some ideas for more efficiently storing Chisels & Bits data
now i just need to figure out who the C&B dev is
lean warren
Orion
vague otter
i've got a really clever idea half sketched out in my head for how to serialize an octree
prime quest
Serialised octree: the most important meal data structure of the day
bold steppe
but with registry replacements, at best, you get to a "last man wins" situation
I like registry replacement but I don't like you can't add blockstates to s block you replace. Kind of defeats the purpose
quiet talon
I'm not sure if it's a goal anymore <:kek:1134098707869073408>
its a goal of vanilla :p
inland mesa
I like registry replacement but I don't like you can't add blockstates to s bloc...
the blockstate restriction was added because vanilla doesn't sync blockstate ID maps, so if you replace or add new state properties to a block, the IDs become out of sync. in hindsight, we should have probably made it so vanilla is rejected if the blockstate ID map contains anything non-vanilla in it
bold steppe
the blockstate restriction was added because vanilla doesn't sync blockstate ID ...
So does this also apply to the mixin way?
inland mesa
the desync? yes
if you mixin a block to have more blockstates, forge can't detect that something has changed (it doesn't have a list of the original states, it only looks at registry replacement before / after) so it won't error, but then the state id map will be different and a vanilla client won't know about it
fallow sundial
that's why you use a library to add more states
bold steppe
the desync? yes
By mixin I mean replace the new Block line in the registry with your own one
inland mesa
same thing
mixin into SomeBlock vs replace the registration line is the same result: forge doesn't know it has happened
forge can only know if you used registry replacement
kindred fractal
Registry replacement is going away forever. Time to say goodbye
bold steppe
mixin into SomeBlock vs replace the registration line is the same result: forge ...
Ah. That's weird tho I had it in a forge fabric mod for almost 2 years and never noticed blockstate desync
bold steppe
forge can only know if you used registry replacement
But if you use it you can't add new states which was the whole point of the replacement
kindred fractal
not really, the point of the replacement is to replace the block with a custom subclass instead of using mixin to achieve the desired behavior changes
bold steppe
Right but how do you add new states and why does it seems that my mixin code manages to add states without issues?
inland mesa
Ah. That's weird tho I had it in a forge fabric mod for almost 2 years and never...
remember this only applies when the other side is VANILLA
this isn't about vanilla blocks but a vanillaclient connecting to forge server or forge client connecting to vanilla server
bold steppe
Ah right makes sense as to why it would only work with mods that are required on clients and servers
Anyways I'm just saying that just like you can register a new block (which makes vanilla connection not possible afaik) would be nice if you could add states with reg override with same effect
kindred fractal
again, registry replacement is going away
jagged agate
Ah so guess mixin is only way then?
Use Statement library if you’re adding new blockstates to vanilla blocks. They handled all the edge cases that you won’t think of. I’ve been burned waaaaay to many times by other people thinking it’s an easy mixin to add new states to vanilla blocks and then forget to handle all the extra edge cases that come from it.
In past, it breaks all datapacks using blockstate codec as the codec used to require people to define every single property for a block so when one adds a new property… everyone else dies
Another issue is people extending the class for their own block now gets that property. And they code isn’t made to handle that new property. I used to extend VineBlock. Then someone unsafely mixin added waterlogging to VineBlock. My code proceeded to crash because it couldn’t handle my block being waterloggable.
So many bugs and issues by people thinking they will just do it themselves when mixin properties to blocks lol. Just use Statement library pls
As you can see, I am very passionate about that as I broke or crashed due to other people’s block property mixins over at LEAST 10 times
Bonus for one person screwing up their block property mixin to lanterns and caused all lanterns spawned in structures to be waterlogged always
bold steppe
Use Statement library if you’re adding new blockstates to vanilla blocks. They h...
Actually can you elaborate? As I've said I used that stuff for 2 years and didn't get any known issue that I remember. To clarify what I'm doing is replace the new Block line with new MyBlock. Of course adding shit to the state definition of an existing class is extremely unsafe and I've been messed so many times by people randomly adding waterlogged stuff to it for example
jagged agate
i already elaborated
registry replacing with a new block might be better in the regard to adding new property not breaking people extending the original block but you run into issue like giga said that it has to be on both sides
bold steppe
Yes but I mean since the codec issue isn't a thing anymore and your second example doesn't apply to this
jagged agate
I'm talking in general about adding new properties to blocks
bold steppe
I'm trying to see where other of these edge cases are
jagged agate
registry replacing might be safer for adding new property for a block on both sides but I still don't trust it. I trust no one with this.
rich void
to provide a slightly more complete explanation: in Object-Oriented programming ...
Interesting! And yes I have seen those two, but I am very much still learning Java and getting to know everything.
jagged agate
If I had to add a new property to an existing block, what I would personally do is make a new block class extending the old one and have the new property in the new class. Register the new block under my own modid. Then when conditions are met, swap the existing block with my block in the world itself and swap back when needed. Issue is new block won’t be in tags that people specified the old block in directly. That or statement
fallow sundial
another issue with swapping the block is that once the mod is removed, so is the block
also you'd want to update all references to the block (loot tables, recipes, worldgen etc)
jagged agate
Statement it is.
Now imagine what happens once Mojang makes blockstates be data driven lol
fallow sundial
soon enough heh
lethal bane
weren't there plans for blockstates in blockstates at some point?
shadow widget
what does that mean?
lethal bane
for stuff like advanced waterlogging, just with any fluid
gilded halo
having a proper layer system for true fluidlogging without huge amounts of blockstates would be great
lethal bane
I don't remember what the exact plan was, I just remember that someone talked about that like 1.5 years ago or something
rich void
Now imagine what happens once Mojang makes blockstates be data driven lol
Probably in 23w33a xD
vague otter
honestly that might end up with just letting things crash if things don't contain blockstates that they're supposed to
there are structure processors that can crash if non-stairs are in the stairs block tag e.g.
lethal bane
inb4 "No more Blockstate crashes" mod in a similar vein to the "No experimental settings" mods
jagged agate
Imagine all blockstates become tags. You tag something and it gets that blockstate. Cursed as hell and limiting. Which means Mojang will do it
lethal bane
hmm yes, minecraft:age_7
arctic sphinx
<@&1067092163520909374> https://www.minecraft.net/en-us/article/minecraft-snapshot-23w33a
rich void
how do I get that role?
arctic sphinx
Currently? You can't
I'm not doing this again
At least until Maty or Sci confirm it is now solely for notification
arctic sphinx
Apparently that role does more than I thought <:screm:980923567833219093>
I'll l...
.
shadow widget
big attack reach changes 
nova sundial
reaction roles, anyone?
kindred fractal
Mob attack reach changes
@quiet talon
rich void
Currently? You can't
I'm not doing this again <:harold:932605815406342174>
At le...
ooooo ok I'll wait
dull bison
they changed some F3 keybinds
arctic sphinx
Yep, and Narrator key can finally be rebound
quiet talon
> Mob attack reach changes
<@171447165128802316> <:kek:1134098707869073408>
Probably doesn't impact much, the player attributes were never hooked up for mobs
kindred fractal
Ahhh ok
rich void
interesting change: the chorus fruit now INSTANTLY puts you on the ground when falling from ANY height
MLG water buckets are useless now after going to the end
drowsy condor
interesting change: the chorus fruit now INSTANTLY puts you on the ground when f...
they always did?
rich void
they didnt reset fall damage
muted hemlock
i think it was added in 1.9
to be fair I never played 1.9 either
I don't think I've ever seen a chorus fruit
rich void
to be fair I never played 1.9 either
u already modded the game by then?
muted hemlock
not really
stopped playing for a while after 1.4 and started playing/modding again around 1.13
lethal bane
reaction roles, anyone?
looks at #squirrels-🦊 message
but apparently the role was removed
wise forge
Nothing really interesting this time, just mainly refactoring of the reporting system: https://gist.github.com/ChampionAsh5357/933bdc1028e42e85313e7b58c0a70d39/revisions#diff-485198454aa7c9ab60942b4f45e19b22d6f3a9c40c5d118c3f5212b1ba14f9b2
tacit vale
@tranquil salmon so if I wanted to setup a forgedev 1.20.2 dev workspace, do I need to do anything special or will the 1.9.2 vineflower sorta work or what?
I am curious to know how much effort it would be to convert the vineflower release for the latest snapshot to a forgeflower release; I haven't really looked at how much effort it took to go from forgeflower -> vineflower
tranquil salmon
Here is a version of neoform for 23w31a that uses forgeflower: https://maven.neoforged.net/#/releases/net/neoforged/neoform/23w31a-20230821.183055
Also forgedev should just work with the version that uses vineflower although the current version of neogradle hardcodes de.oceanlabs.mcp in some places
dull bison
will NG not be replaced by dedicated stuff in buildsrc?
tacit vale
sounds incredibly brittle
there's already a lot in the build src, probably too much
having a dedicated plugin means we can improve toolchain and apply that to more than just one branch at a time
No idea how NGNext works though
dull bison
NGNext has nuked all the forgedev only stuff
tacit vale
@mental carbon so what's the plan then??
mental carbon
I am working on a ngnext platform plugin
It is currently an includeBuild on FoxLauncher
So that I can effectively develop stuff
But yeah ngnext will have a platform plugin we can just apply
mental carbon
That is fine
I don't thing NGNext and the installer rework in Den will reach readyness for 1.20.2
And they are also not directly needed
tacit vale
Den?
mental carbon
It is the name of the new installer platform
mental carbon
Yes I started, but have not committed anything the repo is empty.
No I did no such thing.
My main goal is to make the installer modular and CLI runnable
The main concept is that you can host installer modules anywhere
By default the installer would look for ours, but if given command line options (or by selecting the relevant options in the UI) you can add other module sources as well 😄
tacit vale
but it's java right
mental carbon
Yes it is java
tranquil salmon
I don't thing NGNext and the installer rework in Den will reach readyness for 1....
Then NG needs to be modified to not hardcode de.oceanlabs.mcp:mcp_config
mental carbon
Yeah
It can be done easily
But at the moment we should figure out what decompiler to use
mental carbon
Well people have been putting work in VineFlower
tacit vale
Then NG needs to be modified to not hardcode `de.oceanlabs.mcp:mcp_config`
Thanks for the reminder
mental carbon
And pushing us to use it
tacit vale
And pushing us to use it
I don't think so? Who
In my eyes we need to wait a bit longer for VF
mental carbon
Was there not significant effort put in to make NeoForm compatible with it? Or was that just curiosity?
Yeah for sure VF is not stable enough
tacit vale
nobody was pushing us but that was coeh
tacit vale
and I fully intend for that to get used when VF is ready
tacit vale
But right now VF is not ready which is fine
tacit vale
But it needs to produce much better code then ForgeFlower
no it doesn't
mental carbon
?
tacit vale
It just needs to not break and provide advantages
mental carbon
Well that is true
tacit vale
Line wrapping is nice when it works. Inline lambdas
has some hiccups with generics afaik
mental carbon
is still waiting for CG
tacit vale
I don't remember if Coeh every explained if VF was a net positive or negative in terms of purely # of patch lines added
if I had to guess tho probably added more lines than it saved
kindred fractal
producing better code in the sense of less compile errors is an advantage for the toolchain
producing better code in general is nicer for modders
tranquil salmon
Readable commands (atleast when using mojmap)
kindred fractal
is that VF or FF?
tacit vale
VF
tranquil salmon
VineFlower
kindred fractal
it's missing some generics though? ah no that's just weird editor colors 😄
tacit vale
I don't remember if Coeh every explained if VF was a net positive or negative in...
Do you remember off the top of your head coehlrich or did you not calculate this
tranquil salmon
I think VineFlower requires more patch files
stray scroll
something something generics
tacit vale
rip
stray scroll
generics is the enemy of the Java decompiler
stray scroll
ah yes, someone please remind me how many patch hunks we have for fixing generics issues alone
kindred fractal
most of the patches are for generics, I'm pretty sure
tranquil salmon
Here's the commit for FF to VF: https://github.com/neoforged/NeoForm/commit/d8f1b8fbbb659d2aa3ff339dafc2cedea126f70a
tacit vale
+600 lines hmm
stray scroll
eliminating false positives, it would need to look for - lines without a </> and corresponding + lines with a </>
tacit vale
there's few enough patches you could just manually verify
rich void
It is the name of the new installer platform
oh as in a fox's den? clever!
mental carbon
Yes
rich void
nice
stray scroll
<@&1067092163520909374> https://www.minecraft.net/en-us/article/minecraft-snapshot-23w35a
formal plume
there is almost nothing in it from an outside perspective
tacit vale
Lot of dep updates
Reminder: Update neoforge build.gradle ext block to update all versions of vanilla deps that are not overriden
tranquil salmon
Judging by the change log I don't think any changes to patches for neoform are going to be needed
wise forge
Dang, can’t ping snapshot
tacit vale
?
wise forge
Nope, I see it
tranquil salmon
rich void
what's this?
tranquil salmon
23w33a to 23w35a diff is about 21k lines
muted hemlock
recipes got refactored?
kindred fractal
just their loading and a bit of their usage
wise forge
Quick look over everything tells the me the major changes are just some refactoring of advancements, triggers, and recipes. Mainly recordizing and codec stuff
tranquil salmon
Recipes don't store their own resource locations anymore
kindred fractal
ah that explains why there is RecipeHolder everywhere now
tranquil salmon
Seems like it's the same case for advancements
muted hemlock
what about loot tabes
fallow sundial
lean warren
They went to 17.0.8 according to the blog post
rich void
Recipes don't store their own resource locations anymore
is that better than what it used to be (in your opinion)?
kindred fractal
does it matter? ¯_(ツ)_/¯
kindred fractal
custom ingredient API will need an overhaul thanks to codec fun
mental carbon
Probably something with dispatch and a serialization registry
kindred fractal
yeah
rich void
does it matter? ¯\_(ツ)_/¯
you tell me
quiet talon
I think I preferred items being aware of their registry name but codecs don't support that and they're going for consistency
rich void
what are codecs?
muted hemlock
!fcw Codecs
kindred hatchBOT
!fcw Codecs
There's an article at https://forge.gemwire.uk/wiki/Codecs from the Forge Community Wiki that may be of interest to you.
muted hemlock
#modder-1․20․1-support too
inland mesa
I find it interesting that there's so many people who are active modders making mods for versions 1.14 and up, that still haven't come across Codecs despite being all over Minecraft
shadow widget
I guess it might have finally changed in this update(no idea), but lots of the stuff relevant to most modders, like recipes for example, doesn't use codecs
mossy bobcat
I think I preferred items being aware of their registry name but codecs don't su...
I mean, it's a single IdentityHashMap lookup basically to get the location from the item nowadays, using the registry, so I think it's fine
quiet talon
it will be interesting to see what the recipe manager looks like now if its using holders
Because it has to still be reloadable, so its not using the primary registry system
muted hemlock
RecipeHolder
quiet talon
is there uh, a separate one for each subtype
RecipeHolder, AdvancementHolder, LootTableHolder
that would be haunted
quiet talon
Can't be, not registry backed
I mean I guess it could be, technically
but unlikely
it sounds like it would be similar to https://github.com/Shadows-of-Fire/Placebo/blob/1.20/src/main/java/dev/shadowsoffire/placebo/reload/DynamicHolder.java
dim sonnet
If we're lucky we might get 21
tawny iron
doesn't look likely if they're only just updating to 17.0.8 now
dim sonnet
Yeah, more likely to get 21 with 1.21
tawny iron
poetic
true pivot
gotta make up for shipping java 16 with 1.17, and java 17 with 1.18
tawny iron
lol
kindred fractal
RecipeHolder, AdvancementHolder, LootTableHolder
Yes for the first two
They're just records with the id and the recipe
fallow sundial
ah yes, instead of fixing the dp reg system to support reloading let's make "Holders"
MOJAAAANG
jagged agate
No loottable holder? Rip. I wish I could easily look up loot table names from loot table object
kindred fractal
ah yes, instead of fixing the dp reg system to support reloading let's make "Hol...
That's completely unrelated though
We're talking about a trivial record with two fields, idk why you'd want to bring up registries
wise forge
@lean warren have time to take over primer writing? I’m not sure what my computer is doing, but it isn’t what it’s supposed now
lean warren
Bad timing, I'm stuck in the middle of something else right now :/
wise forge
Hmm, @kindred fractal who’s the person who does the primer for fabric? Can’t think of anyone else willing
tranquil salmon
private <C extends Container, T extends Recipe<C>> Map<ResourceLocation, RecipeHolder<T>> m_44054_(RecipeType<T> p_44055_) {
return (Map<ResourceLocation, RecipeHolder<T>>)this.f_44007_.getOrDefault(p_44055_, Collections.emptyMap());
}
Cannot cast from Map<ResourceLocation,RecipeHolder<?>> to Map<ResourceLocation,RecipeHolder<T>>
What is wrong with this cast?
inland mesa
<?> is not <T>
so far as java is concerned, they are literally different types
is f_44007_ a map? what type is it?
tranquil salmon
Map<RecipeType<?>, Map<ResourceLocation, RecipeHolder<?>>>
inland mesa
yeah that cannot be cast to T directly
gotta add a temporary
or intermediate cast, ```java
private <C extends Container, T extends Recipe<C>> Map<ResourceLocation, RecipeHolder<T>> m_44054_(RecipeType<T> p_44055_) {
return (Map<ResourceLocation, RecipeHolder<T>>)(Map)this.f_44007_.getOrDefault(p_44055_, Collections.emptyMap());
}
and it will complain about raw types / unsafe
but that's fine
tranquil salmon
It should be getting the value of the map with a type of Map<ResourceLocation, RecipeHolder<?>> then casting that
quiet talon
Honestly I wish java had associative generics like that
inland mesa
yeah
quiet talon
so that you could have Map<T, Thing<T>> and the key->value generics had to match, but not the overall generics
inland mesa
yeah, but I have no idea how you'd declare such constraints
quiet talon
Yeah, you would need a new generic syntax
Map<associative T, Value<T>>
but that looks haunted™️
inland mesa
yeah the way C# does constraints is a bit more welcoming to stuff like that
but anyhow
inland mesa
It should be getting the value of the map with a type of `Map<ResourceLocation, ...
the issue is javac does not let you cast from incompatible types
and Map<RL, RH<?>> is not covariant/contravariant in any way
quiet talon
Do you even need the explicit unchecked cast there?
I think you can just return the rawtype
inland mesa
you mean like ```java
private <C extends Container, T extends Recipe<C>> Map<ResourceLocation, RecipeHolder<T>> m_44054_(RecipeType<T> p_44055_) {
return (Map)this.f_44007_.getOrDefault(p_44055_, Collections.emptyMap());
}
? yeah that should be the sameserializers for Recipes now use codecs
advancements don't use codecs though
AdvancementProgress does
tacit vale
It should be getting the value of the map with a type of `Map<ResourceLocation, ...
Afaik it's bc the nested generics
if it was list<?> vs list<T> it would obviously be different
But it's List<A<?>> to List<T> which is not the same
*when you boil it down to a single generic as an example
tranquil salmon
RecipeProvider.buildRecipes now takes a RecipeOutput instead of Consumer<FinishedRecipe>
Along with void accept(FinishedRecipe) RecipeOutput also has Advancement.Builder advancement()
kindred fractal
Hmm, <@298082405745426433> who’s the person who does the primer for fabric? Can’...
It's apple502j
vague otter
Yeah, you would need a new generic syntax
`Map<associative T, Value<T>>`
<T> Map<T, Value<T>> 😛
honestly i was half joking but i'm having trouble coming up with a better way to do it in C# either, you've fundamentally got to differentiate between "type variables internal to this declarator" vs "type variables that belong to the class/method"
inland mesa
the issue is that it can't be a constraint of the class, it has to be a constraint of the entry, which makes it weird
vague otter
actually this has the problem that it makes it ambiguous to try to declare a method returning such a map
inland mesa
Map<KeyType<:T>, ValueType<:T>> suppose you had a syntax like this
:T wouldn't be a specific type, but a placeholder. anywhere the K,V parameters are used in the class, the :T would become itself a generic parameter.
like ```java
class Map<K,V> {
public V get(K k);
public void put(K k, V v)
}
would be seen for type checking purposes as ```java
class Map<KeyType<:T>, ValueType<:T>> {
public <T> ValueType<T> get(KeyType<T> k);
public <T> void put(KeyType<T> k, ValueType<T> v);
}
I don't think this can be done without these "meta-generics"
obviously you couldn't use T* in practice, specially in C# :P
let me change it to something else for clarity (used :T )
mossy bobcat
Higher order types! Those things Java should really have but doesn't
This is actually not that different to the issue that pops up with lambdas at times, and is why functional interfaces with method type parameters can't be used for lambdas - the inferred types of the lambda parameters in practice would look a lot like that
Or, basically, "type parameters at random places" is another way to think about it, because that's fundamentally what you'd have to support.
fallow sundial
We're talking about a trivial record with two fields, idk why you'd want to brin...
because said record contains a RL and an object, and you read them from a json and throw them in a map, it's basically a dp registry 😛
kindred fractal
There's no tag though
wise forge
It's apple502j
Thanks
wise forge
Cool, got apple’s permissions to links their primers for the time being
tacit vale
Cool, got apple’s permissions to links their primers for the time being
Not even snowblower output is possible to import on your pc? It must be really broken
fallow sundial
aren't apple's primers yarn
fallow sundial
those typos aren't in forge, we don't have our own mappings anymore, they're in mojang's naming
tropic oak
those typos aren't in forge, we don't have our own mappings anymore, they're in ...
Aye. 
stray scroll
all hail Bootstap
wise forge
Not even snowblower output is possible to import on your pc? It must be really b...
Yep it’s great
wise forge
aren't apple's primers yarn
Pretty sure, but that’s how it is since no one else wants or have time to do it
kindred fractal
Ring the alarm
inland mesa
oh shit pre1
RIP function additions
This data pack version removes the recently introduced execute if function and return run functionality. Flaws with those commands (see bugs MC-264595, MC-264699 and MC-264710) require some substantial changes to fix, which we do not want to make close to a release.
arctic sphinx
fallow sundial
The search will only match the beginning of any word in the item's name
For example, searching for "tor" will still show Torch and Redstone Torch but not Daylight Detector anymore
kindred fractal
Make creative search even worse (impossible challenge)
nova sundial
so who's making the micromod to reverse this change?
fallow sundial
Make creative search even worse (impossible challenge)
that's not creative search, that's recipe search
tacit vale
welp
we never started the snapshots
I don't have enough time to help start now either
tacit vale
lol
@tranquil salmon It turns out both of your bug reports caused them to revert an entire feature
Good job, I guess?
I never imagined the issues would be so extensive they would revert the entire thing
kindred fractal
I can only help in 2 weeks 
dull bison
I can only help in 2 weeks <:kekw:995580032070193163>
don't port finish the prs
I can help with porting if someone shows me how
stray scroll
@tacit vale write a quick outline of what needs to be done during porting
tacit vale
<@277525077669445632> write a quick outline of what needs to be done during port...
it's not in my brain readily available like that
And the process is different cuz neoforge now
no more private maven
Ok let me break down the absolute basics:
-
Checkout latest of whatever branch before porting
-
Uncomment the /projects/forge in gitignore, then commit ALL java files (it is now illegal to distribute this on a public GitHub)
-
Commit the files
-
Update mc version and NeoForm version -- this is harder cuz we are using mcpconfig right now, we need to fix NeoGradle, too
-
Run setup -PUPDATING=true, commit
-
Verify that it did things and then start fixing compile errors -- COMMIT OFTEN AND PUSH OFTEN
-
We need a private git repo again if it doesn't already exist. People need permission to it in order to help
I have a more complete document somewhere else
But it's in discord
not somewhere I remember
rich void
people working on the port ignore this:
mending books in ancient cities awesome 10/10
stray scroll
Ok let me break down the absolute basics:
- Checkout latest of whatever branch b...
@mental carbon @fallow sundial gradlemeisters, we'll need someone to go poke at NG for switching to NeoForm
(excluded Shrimp since he seems to be busy )
winged hull
Ok let me break down the absolute basics:
- Checkout latest of whatever branch b...
why is it illegal to distrivute on a public github?
stray scroll
why is it illegal to distrivute on a public github?
because it'll contain the MC source
mental carbon
<@304594425286361088> <@561254664750891009> gradlemeisters, we'll need someone t...
Sure I will check in a minute
stray scroll
because it'll contain the MC source
and as you know, that's illegal
mental carbon
Currently running a unit test sweet
stray scroll
(presuming you meant "suite")
mental carbon
Yeah
Aequivaleo color handling for cliques
It is a fiddle piece of code
Since technically in the recipe graph they are not cliques
But in the real world, those "recipes" produce a clique
Cause from every single color of wool you can get to any other wool with a single craft
So it is tricky to detect and implement properly
stray scroll
@tacit vale do we commit the rejects? where do they live? are they covered by the /projects/forge uncomment?
mental carbon
Normally yet we commit them
And I think so -.-
But they might be in a completely seperate rejects folder in the root
And got my unit test to pass
It helps if you actually pass in the right values......
WTF marc
stray scroll
okay
mental carbon
UserDev is actually pretty close
I solved several issues this week
And Schurli is taking care of the mixin support
So that we can support that natively
stray scroll
okay, good
mental carbon
Once that completes
And I get some people to actually test it
And figure out if we need a better DSL / API in certain places we can do a limited Beta run
dull bison
And Schurli is taking care of the mixin support
mixin works in dev (haven't been able to test prod yet)
tacit vale
<@277525077669445632> do we commit the rejects? where do they live? are they cov...
Oh yes that must be explicitly done
It is not covered afaik
stray scroll
noted down on my whiteboard
mental carbon
Damn
I had hoped my tests would fix the corner case
It seems to work in my unit tests
But in production it stills craps the bed
mental carbon
@stray scroll @tacit vale @fallow sundial Any of you know a good name for the private repo? NeoForge-Private seems boring
nova sundial
SecretNeoForge
mental carbon
How about Kits? (As in a young fox?)
winged hull
SecretFox
muted hemlock
neoforge-private-super-secret-donotshare
nova sundial
do_not_share_or_you_will_be_fired
winged hull
if_you_share_this_you_will_be_suffocated_by_the_ever_present_bolb
mental carbon
Pushing to Kits for now
All maintainers have access
I will clone it then
And apply the update to FGNext
Before starting the migration
But I think I need to wait for @tranquil salmon to do his magic
Okey pushed
Updated the gitignore and then pushing the current state
Before starting the update
Okey, update is on hold untill I can get a new NeoForm update
With which I can test my FG Changes
I will do that tomorrow morning
Right now I had a long day and I need sleep
mossy bobcat
Just have to say - I love that the processes and workflows for doing these updates are all being shared in public channels! I feel like I understand the update process more from just reading what's been in this channel than I did before from everything else I saw
kindred fractal
Remember to add a webhook to #neoforge-private (idk if something's been pushed yet)
tranquil salmon
But I think I need to wait for <@239615312352247809> to do his magic
https://github.com/neoforged/NeoForm/actions/runs/6089580528
This should publish to the maven if testjdks is successfull
kindred fractal
So, vineflower will be used?
tacit vale
<@607058472709652501> <@277525077669445632> <@561254664750891009> Any of you kno...
Personally I like boring names if they have easy acronyms
arctic sphinx
NFP is too close to NFT for my liking
muted hemlock
non-fungible poultry?
arctic sphinx
Kits a good name given the context
mental carbon
The repo of for kids called Kits, after young foxes
We can teenage it later
Rename*
nova sundial
Personally I like boring names if they have easy acronyms
fart -_-
tacit vale
that's not a boring name
that's a stupid name
like can we just give things reasonable names pls thanks. Kits is fine idk I'm not good at it
fallow sundial
good thing it's now ART
tacit vale
I just don't think we should give everything a unique name cuz it gets hard to remember. like why Kits vs like neoforge-staging, -internal, etc?
bc other than the name it's literally the exact same thing as neoforge but our custom version
kindred fractal
Kits is a garbage name btw
Stop trying to be clever, and make a usable product instead, IMO 😛
(NeoForge-Private is good)
tranquil salmon
quiet talon
yeah kits is a horribly irrelevant name
fallow sundial
also reminder to update the login packets as discussed
dim sonnet
yeah kits is a horribly irrelevant name
Meanwhile Ducker...
Which has nothing to do with docker or testing btw
mental carbon
Okey porting has started
mental carbon
also reminder to update the login packets as discussed <:thinkies:98092438733311...
FG Is patched
Loading NeoForm with the new data
And running setup in update mode
mossy bobcat
Meanwhile Ducker... <:harold:932605815406342174>
Honestly I don't mind the silly names. It's a good reminder that this is a modding project for a block game and not everything has to be serious. As long as there's repository descriptions and whatnot, of course...
mental carbon
I explicitly did not choose NeoForge-Private or anything in that direction
For the simple reason that; IMHO I think that name is too long, and can cause confusion.
BTW any Maintainer now has access to the private repo and can help porting
No more gatekeeping bullshit
mental carbon
Meanwhile Ducker... <:harold:932605815406342174>
That name is actually derived from Duck Typing, which is the core component which mixin derives its logic from
108 Reject files detected
dull bison
BTW any Maintainer now has access to the private repo and can help porting
what about volunteers?
mental carbon
what about volunteers?
That is sadly not up to me
dull bison
who do I have to ?
tranquil salmon
So we're doing 1.20.2-pre1 and vineflower at the same time?
mental carbon
Yeah
It is currently only 108 rejects
Which is relatively easily doable
I do however have a problem
Some dependency jars are missing in my IDE
Like JOML
Yet others are there.
So I am bit confused at the moment
A double refreshed fixed it
stray scroll
who do I have to <:stabolb:837125030479134730> ?
not yet™️
fallow sundial
tbh I'd say this is a very special scenario since we're chronically lacking porters lol
mental carbon
Shit
ItemPredicates became a record
Looks like we need to refactor that ItemPredicate core
stray scroll
tbh I'd say this is a very special scenario since we're chronically lacking port...
hmmm
mental carbon
Okey I will leave the ItemPredicate rejects patch currently in place
fallow sundial
don't think people even use custom predicates
mental carbon
don't think people even use custom predicates
I do
So I know people will
It is technically patchable
tranquil salmon
You didn't publish your custom NeoGradle version
quiet talon
what is ItemPredicate for again?
mental carbon
To check for overrides
tranquil salmon
- Plugin Repositories (could not resolve plugin artifact 'net.neoforged.gradle.patcher:net.neoforged.gradle.patcher.gradle.plugin:[6.2.0-features-neoform_defaulted, 6.4)')
mental carbon
`- Plugin Repositories (could not resolve plugin artifact 'net.neoforged.gradle....
I know
Still need to push the branch
And talk to shrimp how we will publish that
fallow sundial
uhh item predicate is for advancements?
quiet talon
oh, yeah, that being a record will be annoying
mental carbon
uhh item predicate is for advancements?
That too
tranquil salmon
What about a webhook for #neoforge-private ?
tranquil salmon
After fixing some brackets I brought the compile errors down from about 5000 to about 1000
kindred fractal
For the simple reason that; IMHO I think that name is too long, and can cause co...
Call it NeoForge-Porting then maybe
The length doesn't matter - what matters is the clarity
nova sundial
NeoForge-Confidential
stray scroll
i'd tend to agree with Tech here on this, since the private repo is less a separate project (and therefore deserving of a new name) but more an extension of NeoForge
the thought did pass in my mind to name the snapshot mod in NF as Kits instead
tranquil salmon
Shouldn't github actions be disabled in the kits repo?
kindred fractal
Please stop calling it kits
I don't even know what that word means as a non-native speaker
fallow sundial
Please stop calling it kits
just don't mistype the k as t
daring ginkgo
Please stop calling it kits
to me it sounds like you have a private repo full of kids 🙈
stray scroll
i was reminded about the fact of runtime mappings, and plans™️ to change to Mojmap
are we targeting that alongside NG-Next?
mental carbon
NG-Next currently does implement obfuscation
Because it supports an arbitrary runtime naming scheme
Whether runtime mappiongs are mojmap or not is not really related to ir
winged hull
to me it sounds like you have a private repo full of kids 🙈
it's full of tits—I mean kits
mental carbon
In practice the name is not perfect
And we should find a better one
That said
I seem to be stuck with migration
I can't use parchment because it breaks a a large amount of our patching system at the moment
And I can't use official, cause there is some bug with the versioning i need to figure out later today
So
I will get back to you guys in a couple of hours, hopefully with a fixed version of NeoGradle that makes neoform work
winged hull
And we should find a better one
whatever name you guys choose don't choose one that's so easy to make fun of xD
dull bison
whatever name you guys choose don't choose one that's so easy to make fun of xD
this is still the same community that made FART so... probably not
winged hull
wait... what's FART?
dim sonnet
Forge Auto Renaming Tool
winged hull
I feel like F.U.C.K would be a good acronym
inland mesa
the immature humor was 100% intentional: a reminder that forge has never been an enterprise product
but fuck is too in your face
anyhow that was forge Neo doesn't have to follow
the private repo I'd call something like Greenhouse
an enclosed area to help things grow
mental carbon
That is a good idea
dull bison
the private repo I'd call something like Greenhouse
the current scheme is either funny or fox (this is not enugh fox)
kindred fractal
the private repo I'd call something like Greenhouse
Surely that will be confusing for someone not familiar with the project?
dull bison
Surely that will be confusing for someone not familiar with the project?
it's a private repo so everyone who can see it should know about it
kindred fractal
Talking about the webhook
tacit vale
And talk to shrimp how we will publish that
Hi, what's up
inland mesa
Surely that will be confusing for someone not familiar with the project?
how so? If I go to a project and see that the updates channel has a Greenhouse repository which I can't see, I would just naturally assume "oh that's where they are doing the porting work"
IMO it's far more intuitive than "Neo-Private" which sounds like we are scheming from the shadows? it doesn't convey that it's for porting
jagged agate
Where do you make the connection that greenhouse = porting work?
Ain’t no way in hell am I going to know that if I wasn’t seeing this prior convo lol
Neoforge-Porting is clear imo
NeoPorting if you want shorter
tacit vale
NP is not P
:>
idc what the name is but I don't really like Kits and it wasn't like voted on or anything. Not that it matters since it's internal only but still
and if we gave everything a fox name we would run out of them quick
I just think especially when it's the exact same product but a different release cycle
there's no point to rename it
daring ginkgo
Private Porting (pp)
inland mesa
Where do you make the connection that greenhouse = porting work?
Greenhouse is where plants grow. It exists as a shelter from the elements. So in a project like this, what else can it be other than unfinished work? And what is the only thing where unfinished work probably needs to be kept private?
inland mesa
Neoforge-Porting is clear imo
Oh yeah sure, but it's less fun IMO.
inland mesa
🤷♂️