mental carbon Sep 10, 2023, 5:45 PM

#

Cause else we need to always check if we have a vanilla recipe outputs

#

That can return that data

#

Which is completely unreasonable

fallow sundial Sep 10, 2023, 5:47 PM

#

we have? what, methods patched onto other vanilla interfaces are nullable (check bootstap context for example)

mental carbon Sep 10, 2023, 5:48 PM

#

fallow sundial we have? what, methods patched onto other vanilla interfaces are nullable (check...

So lets discuss this out then, cause I can see your side

#

From my point of view this is the problem:

#

What do you do when you do not have th relevant provider / advancement serializer / object returned by added method?

#

Cause making it "nullable", or have any other state then not being there, means there is a functional behavioural state that has to be defined

#

But for example with the new condition system, that is not possible

kindred fractal Sep 10, 2023, 5:49 PM

#

Why would the recipe output expose the provider?

mental carbon Sep 10, 2023, 5:49 PM

#

It will literally crash / not work, if you try to serialize / deserialize a recipe or advancement without the lookup provider being present on the recipeoutput in some fashion

#

Because we needed it to serialize conditions and especially holdersets

#

So we patched in a method to provide it

kindred fractal Sep 10, 2023, 5:50 PM

#

Why do you need it to serialize conditions?

mental carbon Sep 10, 2023, 5:50 PM

#

And made the only vanilla implemention implement that patched method and return the provider that it has access to

mental carbon Sep 10, 2023, 5:50 PM

#

kindred fractal Why do you need it to serialize conditions?

Because they can use it...... It is a codec

#

There is simply no guarantee that there is not a condition which uses them

#

Or which uses the registry

kindred fractal Sep 10, 2023, 5:51 PM

#

Ah

mental carbon Sep 10, 2023, 5:52 PM

#

Basically what we did, was make the whole recipe serialization system uses codecs (where possible, mojang does recipe serialization it self still via normal ways, but the new conditions and ingredient system, forge provides now is completely codec based)

#

And that means that, given that recipes already needed it itself during codec based operations (yes not yet during serialization, because they use custom logic), we needed a way to get that into the saving logic of a finished recipe

#

So that the conditions could be serialized and tagged on onto the jsonobject if needed

#

And that means that a lookup needs to be there

#

And as such

#

My question becomes:

#

If we want to keep this binary compatibility such a high priority (a higher priority, then cleaning up the code massively, and even a higher priority then allowing better modding infrastructure). How should we address this?

Might I point out that we yesterdag had this kind of discussion as well, and there it was even decided to crash a vanilla client so that modders could have better infrastructure.

kindred fractal Sep 10, 2023, 5:56 PM

#

I think it's fine for Forge to make these adjustments to vanilla like we already do for the existing file helper etc

fallow sundial Sep 10, 2023, 5:59 PM

#

I just don't see the need for a registry ops

mental carbon Sep 10, 2023, 5:59 PM

#

Because it is needed for the conditional ops

#

I am starting to feel like you are not trusting mine or @dull bison judgement on making these changes

fallow sundial Sep 10, 2023, 6:00 PM

#

mental carbon Because it is needed for the conditional ops

how does that need a registryops? thonk

mental carbon Sep 10, 2023, 6:01 PM

#

Because it needs to be properly able to access registry names, etc

#

If you do a serialization which normally needs a registryops but you do not provide one, then it will serialize either the full object, if it can or throw.

fallow sundial Sep 10, 2023, 6:03 PM

#

condition testing needs the condition context, and condition de serialization shouldn't touch registry contents, that's the opposite of a condition

mental carbon Sep 10, 2023, 6:03 PM

#

Given that we do not know what kind of objects are stored in a ICondition, as discussed yesterday, or for that matter in an Ingredient that can be stored in an ICondition, there is no way to know if it is needed or not.

For all you know, people store HolderSetTypes in conditions, which need them....

#

?!?

#

What are you talking about

#

These are codec based systems now

#

They can do anything a codec can

#

Including touching the registry if they need to.........

jagged agate Sep 10, 2023, 6:04 PM

#

Makes sense to me. Registryops it is

dull bison Sep 10, 2023, 6:04 PM

#

fallow sundial condition testing needs the condition context, and condition de serialization sh...

what about a condition that is based on the existance of a datapack registry object?

mental carbon Sep 10, 2023, 6:04 PM

#

Or a none_registry object like "item_exists"

#

.......

fallow sundial Sep 10, 2023, 6:05 PM

#

a point of conditions is to avoid deserializing when registry contents exist, the condition shouldn't assume existence of an object when deserialized

mental carbon Sep 10, 2023, 6:05 PM

#

Yeah but the condition needs to check whether it even exists.........

#

Yeah you can do it with a name only

#

and for that that works

jagged agate Sep 10, 2023, 6:05 PM

#

This is a recipe system right? I can see how people can do conditions that only makes a recipe exist if an item exist

mental carbon Sep 10, 2023, 6:06 PM

#

That is already implemented

#

But it actually does not use the item

#

It uses the name

#

To prevent the lookup

#

Which is maties point

fallow sundial Sep 10, 2023, 6:06 PM

#

mental carbon Yeah but the condition needs to check whether it even exists.........

yes. and that happens when the condition is checked against, not deserialized, i.e outside of the codec?

mental carbon Sep 10, 2023, 6:06 PM

#

Yet a condition, can easily do something like a holderset lookup, with an empty check ontop

#

The lookup of the holderset

#

Happens in the codec

#

Which requires the registry ops

#

But the evaluation if they are empty happens outside

#

This really feels like, potatoe potatoe discussion

#

We would like to open up the ability for modders to make more advanced conditions and ingredients

#

By allowing them the access to the registryops

#

So that they can make better content, in a codec native way.

#

This will come eventually, hell even in the next one or two minor versions if I squint my eyes and look at the code a bit funny

#

At that point there will be a provider in that place exactly

#

And then there is no way around that

#

So what is the harm to open up that option to our modders now?

dull bison Sep 10, 2023, 6:29 PM

#

maty the condition is checked in the codec now

fallow sundial Sep 10, 2023, 6:30 PM

#

this is a matter of perspective. the way I see it a condition is stateless. it's constructed using static data and based on that later checks for the validity of a context. because otherwise the test method becomes the constructor. why not add an item field to itemexists and let the deserialization of the condition fail because the item doesn't exist

mental carbon Sep 10, 2023, 6:31 PM

#

Actually the condition is now indeed checked within the codec for recipes

#

Or better said

kindred fractal Sep 10, 2023, 6:31 PM

#

Yeah the condition should be checked in the test method, not when deserializing

mental carbon Sep 10, 2023, 6:32 PM

#

https://github.com/neoforged/Kits/blob/97745f59e0a5a0a0891650e52c0fb307cdf3af75/src/main/java/net/minecraftforge/common/crafting/conditions/ConditionalOps.java#L78-L96

#

It is done before the deserialization

#

But only barely

#

It works like a dispatch codec now

#

Where a dispatch codec pulls the type key

#

And then grabs the codec from that to further deserialize the specific type

#

This grabs the context and the conditions

#

And then evaluates them all

#

If it finds them all being good

#

Then it invokes the actuall codec to deserialize the target

#

While if it does not

#

Then it never invokes the underlying codec

#

And just returns DataResult with a partial empty optional

#

Which gets promoted

#

https://gist.github.com/marchermans/a7c8678c27c19d051b15e575a459388d

#

And the gist for people that can not see it

#

The big advantage is that making something conditional, is a single line patch

fallow sundial Sep 10, 2023, 6:35 PM

#

yes, I don't see how that's any relevant to conditions being stateless and being de/serialized with a registryops allowing the constructor to become a test method

mental carbon Sep 10, 2023, 6:35 PM

#

fallow sundial yes, I don't see how that's any relevant to conditions being stateless and being...

The constructor is not a test method.......

#

It never was........

fallow sundial Sep 10, 2023, 6:35 PM

#

not but you're allowing it to

mental carbon Sep 10, 2023, 6:36 PM

#

The only thing it can do now, is get a value from the registry....

#

Any value from the registriy

fallow sundial Sep 10, 2023, 6:36 PM

#

by deserializing with the context that test otherwise gets

mental carbon Sep 10, 2023, 6:36 PM

#

Or any combination of values from a registry

fallow sundial Sep 10, 2023, 6:36 PM

#

test becomes useless

mental carbon Sep 10, 2023, 6:37 PM

#

If you are really that paranoid, or worried about it, we can unwrap the ops and get the bare ops if you want

#

With out any wrapping

#

And run the conditions list codec on the bare root ops

#

That way it won't have registry ops access

fallow sundial Sep 10, 2023, 6:37 PM

#

now you can just throw inside the constructor because you have the context you need and not deal with it. "freedom" that comes at the cost of horrible datagen experience

mental carbon Sep 10, 2023, 6:37 PM

#

What are you calling horrible

fallow sundial Sep 10, 2023, 6:38 PM

#

a constructor throwing

mental carbon Sep 10, 2023, 6:38 PM

#

You are going from: Adding a none default method is bad for hypothetical compat, to a constructor which could potentially if done extremely maliciously that will throw, to horrible datagen

fallow sundial Sep 10, 2023, 6:38 PM

#

that recipeoutput patch is needed because you de/serialize conditions with the context (the registry ops)

mental carbon Sep 10, 2023, 6:39 PM

#

fallow sundial that recipeoutput patch is needed because you de/serialize conditions with the c...

Yes, so? We want a nice standardized system. That is easy to understand and use

#

It is not just conditions which can use this btw

#

Ingredients and other values as well.......

#

Granted it is not

#

Cause they directly use JsonOps

#

At the moment

#

But they trivially could

fallow sundial Sep 10, 2023, 6:41 PM

#

mental carbon Yes, so? We want a nice standardized system. That is easy to understand and use

and I'm arguing that by deserializing with the ops you're allowing the constructor to easily become the test method

mental carbon Sep 10, 2023, 6:41 PM

#

No

#

It is still running through a codec

#

There is no direct way to get the registry contents into a conditions constructor

#

You have to bend the codec, in a horrible bend, to get it, and additionally that does not give you access to the HolderGetter

#

Nothing more

#

Yes you can make it throw

#

But you could already do that

#

By simply going to the minecraft server instance

#

And asking for it there........

#

So I don't see what the problem is

fallow sundial Sep 10, 2023, 6:44 PM

#

yes you can?.. write a custom codec and cast the ops to registryops. and to make matters worse registry codecs easily throw. a byname codec throws element not found instead of properly failing. so it's easy and trivial to write improper *_exists conditions as the registry ops allows you to opaquely get registry contents inside the constructor

mental carbon Sep 10, 2023, 6:44 PM

#

Yeah but that is not an argument

#

With the old code, you could also easily do that

#

It was only by convention that we said not to do it

#

During the call to fromJson on the old serializer, you can easily ask the registry or the server to get the object or throw before the test

#

It was simply by agreed upon convention that we did not allow it

#

Not by any codewise prevention

#

And that is still possible

fallow sundial Sep 10, 2023, 6:45 PM

#

mental carbon With the old code, you could also easily do that

the old code was manual deserialization from json? codecs allow you to do byname and not even realise the mistake because it just.. works?

mental carbon Sep 10, 2023, 6:46 PM

#

fallow sundial the old code was manual deserialization from json? codecs allow you to do byname...

Yes this was all manual

mental carbon Sep 10, 2023, 6:46 PM

#

fallow sundial the old code was manual deserialization from json? codecs allow you to do byname...

I will give you that it is a tiny bit easier, but you at least need a custom codec, and a direct cast

#

It is not like you can just call a random codec with byName and then it works

#

You have to engineer it against the design of the system

#

If we are really that paranoid, then we should stop modding on java in general..........

#

And with that I said my last word

#

If you find it such an attocios chagne

#

Put it up for a vote along the maintainers

fallow sundial Sep 10, 2023, 6:50 PM

#

mental carbon I will give you that it is a tiny bit easier, but you at least need a custom cod...

what? no? registryfilecodec, Biome.CODEC

mental carbon Sep 10, 2023, 6:52 PM

#

You are rightz

#

holdergetter.get(...) will throw

#

That is sad

fallow sundial Sep 10, 2023, 6:53 PM

#

those are idiomatic bynames

mental carbon Sep 10, 2023, 6:53 PM

#

I noted your discontent with the change

#

And I will think about it

quiet talon Sep 10, 2023, 6:57 PM

#

conditions should def not be in the codec, they are irrelevant to the object, they are external entirely

fallow sundial Sep 10, 2023, 6:59 PM

#

from what I can tell it's a throwaway codec to reduce patch surface

#

while it does look like code smell it's fine-ish™️

#

I do find its throwaway nature weird because every recipe now creates a new codec which is not free

mental carbon Sep 10, 2023, 7:04 PM

#

fallow sundial I do find its throwaway nature weird because every recipe now creates a new code...

It is the same as a dispatch codec

#

The advantage of this approach is that any codec, as long as it supports optional results

#

can be made conditionally

fallow sundial Sep 10, 2023, 7:05 PM

#

mental carbon It is the same as a dispatch codec

I'm not talking about the codec itself

#

I'm talking about the fact that you're creating that new conditionalcodec every single time a recipe is decoded

mental carbon Sep 10, 2023, 7:06 PM

#

Yes you are, you are talking about the codec which checks if the conditions are met, and if not cancels the further deserialization

#

Exactly

#

Which is a couple of additional objects......

fallow sundial Sep 10, 2023, 7:07 PM

#

yeah that's my complaint thinkies

#

the codec class itself I don't care about, I care about the https://github.com/neoforged/Kits/blob/1.20.2/src/main/java/net/minecraftforge/common/crafting/conditions/ICondition.java#L26C32-L26C76 ConditionalOps.createConditionalCodec(codec).codec() call every time a recipe is decoded. yes microptimisations blah blah but it is at least a new useless map for each recipe

mental carbon Sep 10, 2023, 7:10 PM

#

It does not create a map

#

It creates some lambdas sure.....

#

But not a map

fallow sundial Sep 10, 2023, 7:11 PM

#

it does.
https://github.com/Mojang/DataFixerUpper/blob/master/src/main/java/com/mojang/serialization/CompressorHolder.java#L10C85-L10C85
https://github.com/Mojang/DataFixerUpper/blob/master/src/main/java/com/mojang/serialization/MapCodec.java#L16

mental carbon Sep 10, 2023, 7:11 PM

#

Ohh right

#

That is a compressor holder....

#

Meh

#

Forgot about that optimization map that they have there

mossy bobcat Sep 10, 2023, 7:12 PM

#

Yeah spamming creation of codecs is never good - if you have to dynamically create new codecs there's probably a design issue of some sort going on

mental carbon Sep 10, 2023, 7:12 PM

#

It is a weighting of functionality

#

We wanted to be able to make every codecable object optional capable via conditions

#

For future expansion

mossy bobcat Sep 10, 2023, 7:13 PM

#

I can't see the code in question obviously - but is there a reason the codec can't just be made once ahead of time?

mental carbon Sep 10, 2023, 7:13 PM

#

As a bonus that results in a tiny patch if we need to add it to vanilla

#

the trade of: it creates a map

fallow sundial Sep 10, 2023, 7:13 PM

#

like this approach works if mojang were to use a dispatch codec for recipe deserialization. but it doesn't, not now. so you can grab the conditions array from the json object manually like mojang does with the type

mossy bobcat Sep 10, 2023, 7:13 PM

#

Yes but I have to assume that most of the created codecs are identical

mental carbon Sep 10, 2023, 7:13 PM

#

mossy bobcat I can't see the code in question obviously - but is there a reason the codec can...

It requires the target codec that in the end will deserialize it

fallow sundial Sep 10, 2023, 7:13 PM

#

and when and if mojang makes it a dispatch codec you can throw this system at it

mossy bobcat Sep 10, 2023, 7:13 PM

#

mental carbon It requires the target codec that in the end will deserialize it

Right but this is being called more than once per any given target codec, right?

#

If I understand the issue right

fallow sundial Sep 10, 2023, 7:14 PM

#

basically, mojang doesn't use a disptach codec

mossy bobcat Sep 10, 2023, 7:15 PM

#

Can you not wrap the original codec once and call it a day?

fallow sundial Sep 10, 2023, 7:15 PM

#

but orion patched it with this which wraps the "dispatch" codec every invocation

mental carbon Sep 10, 2023, 7:15 PM

#

Not for recipes, yet

mossy bobcat Sep 10, 2023, 7:15 PM

#

fallow sundial basically, mojang doesn't use a disptach codec

...the heck that's vanilla code?

fallow sundial Sep 10, 2023, 7:15 PM

#

and I don't see a problem with decoding recipes outside like mojang does with the type

mental carbon Sep 10, 2023, 7:15 PM

#

Yes

mossy bobcat Sep 10, 2023, 7:16 PM

#

I guess that makes sense

fallow sundial Sep 10, 2023, 7:16 PM

#

when mojang makes it a dispatch codec this works

mental carbon Sep 10, 2023, 7:16 PM

#

From the code it is very clear that they are rewriting this as we speak

fallow sundial Sep 10, 2023, 7:16 PM

#

but until then ehh

#

looks bad

mental carbon Sep 10, 2023, 7:16 PM

#

If it is really such an issue

#

I can make it a recipe dispatch codec

#

And then it is a single codec

#

Made once

#

And then never again

mossy bobcat Sep 10, 2023, 7:17 PM

#

mental carbon Yes

Okay, so, I'd argue that the proper way to do this is to wrap the recipe serializers from the registry, so that each serializer gets wrapper once

#

At least until it's a dispatch codec

mental carbon Sep 10, 2023, 7:17 PM

#

The proper simple fix is to make this a dispatch codec

fallow sundial Sep 10, 2023, 7:17 PM

#

mental carbon I can make it a recipe dispatch codec

just do that and make sure it throws the same error

mossy bobcat Sep 10, 2023, 7:18 PM

#

But yeah, honestly, I'd make it a dispatch codec

#

That's probably simplest

mental carbon Sep 10, 2023, 7:18 PM

#

I just realised that it is not that trivial XD

#

Actually

#

never mind

#

i am stupid

#

And overthinking this

fallow sundial Sep 10, 2023, 7:19 PM

#

just use this constructor directly https://github.com/Mojang/DataFixerUpper/blob/master/src/main/java/com/mojang/serialization/codecs/KeyDispatchCodec.java#L45 so you can do a custom dataresult error

#

actually you can use partialdispatch

#

there's an overload with a custom function returning dataresult

mossy bobcat Sep 10, 2023, 7:21 PM

#

As long as it throws a JsonSyntaxException like vanilla it all should be fine

fallow sundial Sep 10, 2023, 7:21 PM

#

yeah that's a problem harold

#

or well

#

ehh

#

I guess not really, as long as you check the error message

#

it's fine™️

mossy bobcat Sep 10, 2023, 7:22 PM

#

Not really that hard. mapError with a custom error prefix is the way to go

fallow sundial Sep 10, 2023, 7:23 PM

#

ah nvm vanilla always throws a jsonsyntaxexception when deserializing recipes

#

so just make sure it has the same error message (using the partialdispatch overload I mentioned)

mossy bobcat Sep 10, 2023, 7:23 PM

#

fallow sundial ah nvm vanilla always throws a jsonsyntaxexception when deserializing recipes

Nope sometimes it throws a JsonParseException

mental carbon Sep 10, 2023, 7:23 PM

#

public static Codec<Recipe<?>> DISPATCH_CODEC = ResourceLocation.CODEC.dispatch(recipe -> BuiltInRegistries.RECIPE_SERIALIZER.getKey(recipe.getSerializer()), resourceLocation -> BuiltInRegistries.RECIPE_SERIALIZER
                .getOptional(resourceLocation)
                .orElseThrow(() -> new JsonSyntaxException("Invalid or unsupported recipe type '%s'".formatted(resourceLocation)))
                .codec());

#

This should do the same

fallow sundial Sep 10, 2023, 7:23 PM

#

oh those aren't the same

#

damn it LUL

mossy bobcat Sep 10, 2023, 7:24 PM

#

mental carbon ```java public static Codec<Recipe<?>> DISPATCH_CODEC = ResourceLocation.CODEC.d...

Yeah that'll work; doing it in the dispatch is the way to go

fallow sundial Sep 10, 2023, 7:24 PM

#

throwing inside the dispatch function is thinkies

#

but if it works

mossy bobcat Sep 10, 2023, 7:24 PM

#

Funky stack trace but very easy to catch the exception

mental carbon Sep 10, 2023, 7:24 PM

#

Yep

#

It is literally the exact same execution logic

#

A bit further down the stack

#

But it works perfectly fine

mossy bobcat Sep 10, 2023, 7:25 PM

#

way further down the stack probably cause it's DFU fuckery but whatever

mental carbon Sep 10, 2023, 7:27 PM

#

fallow sundial Sep 10, 2023, 7:27 PM

#

remove that getasstring call, it's just uneeded overhead

mental carbon Sep 10, 2023, 7:28 PM

#

Oh right XD

#

I am not sure why Mojang did not patch this first thing they did when they had codecs on serializer

fallow sundial Sep 10, 2023, 7:28 PM

#

also did anyone tackle the common packets or networking in general? I can look into it tomorrow if not

mental carbon Sep 10, 2023, 7:30 PM

#

I am on most of that as we speak 😄

#

Some of the packet splitter is remaining

#

And some packets might need to be moved aroun

#

The biggest problem is currently GUI

fallow sundial Sep 10, 2023, 7:32 PM

#

well I wanted to discuss some stuff about networking like why the context is wrapped in a supplier and if we still need to do it

#

and I'm not sure whether the network event as an event is still worth keeping

mental carbon Sep 10, 2023, 7:34 PM

#

Feel free to poke at it

#

We have it al on the open heart surgery right now

#

So if you think you need to change something

#

And you have at least a shred of reasoning to improve the api for modders

#

Then go for it

#

Same way me and schurli did with the codec stuff

fallow sundial Sep 10, 2023, 7:39 PM

#

tries to find the disection tools

dull bison Sep 10, 2023, 7:56 PM

#

fallow sundial *tries to find the disection tools*

the hacksaw works too if you don't find it

rich void Sep 10, 2023, 8:03 PM

#

or a machete

dull bison Sep 10, 2023, 8:04 PM

#

~30 files with errors remaining

networking
gui

fallow sundial Sep 10, 2023, 8:05 PM

#

@mental carbon don't forget to make the codecs final

mental carbon Sep 10, 2023, 8:05 PM

#

Rigt

#

55 errors totla on my end

kindred fractal Sep 10, 2023, 9:21 PM

#

fallow sundial but orion patched it with this which wraps the "dispatch" codec every invocation

Optional dead_tater

#

No but seriously:

I don't understand why the dispatch codec is necessary in any way. Why not check for conditions in the json via the usual static helper?
Which part of the context is provided via the codec and which via the IContext? It sounds very problematic to have two context sources... The dynamic registries can be part of the IContext

mental carbon Sep 10, 2023, 9:29 PM

#

You guys are getting two things twisted....

#

we made it a codec in preparation for when Mojang makes recipes a codec. Every other subsystem in this area of the api is now a codec. Even the recipe itself. Just Mojangs dispatch is not. So it is natural that we create a condition system which works natively with the games own serialization systems, and those are codecs.
We added support for registry based ops (of which the condition ops is one of) to the recipe output to make more powerfull codecs possible. Does this open up the possibility of people making a condition to screw up. Sure, is this a massive problem no.

kindred fractal Sep 10, 2023, 9:34 PM

#

The ability to check for conditions in a json object sounds useful to keep

#

(can use a codec under the hood that is fine)

mental carbon Sep 10, 2023, 9:34 PM

#

Basically this system allows any codec to be prefixed with a condition

#

Regardless of if it is json or not

#

It is entirely driven by codecs

#

You can do conditional mbt if you want

kindred fractal Sep 10, 2023, 9:35 PM

#

Yes but did you keep a nice helper for JsonObject checking?

mental carbon Sep 10, 2023, 9:37 PM

#

We did our best, so yes. The helper takes a codec, ops and a carrier for that ops. So in case of recipe jsons, it takes the newly created dispatch recipe codec, the wrapped Json ops, and the jsonobject for the recipe

#

And it returns an optional

kindred fractal Sep 10, 2023, 9:37 PM

#

What if I have a JsonObject and I want to check it for conditions?

#

As in: "Does it have a condition that prevents me from loading it?"

mental carbon Sep 10, 2023, 9:38 PM

#

    static <T> Optional<T> getConditionally(Codec<T> codec, DynamicOps<JsonElement> ops, JsonElement element) {
        return getWithConditionalCodec(ConditionalOps.createConditionalCodec(codec).codec(), ops, element);
    }
    
    static <T> Optional<T> getWithConditionalCodec(Codec<Optional<T>> codec, DynamicOps<JsonElement> ops, JsonElement element) {
        return Util.getOrThrow(codec.parse(ops, element).promotePartial((m) -> {}), JsonParseException::new);
    }

#

These are the two methods we provide you with

#

And then there is this one:

#

    static boolean conditionsMatched(DynamicOps<JsonElement> ops, JsonElement element) {
        final Codec<Unit> codec = Codec.unit(Unit.INSTANCE);
        return getConditionally(codec, ops, element).isPresent();
    }

#

That just does a pure check

#

Using a unit codec for the payload

kindred fractal Sep 10, 2023, 9:39 PM

#

Ah yeah that's good (naming should be present but details)

#

Why not use conditionsMatched for recipes?

mental carbon Sep 10, 2023, 9:39 PM

#

Because we do not need to

#

We can just get the optional

#

Which does the decode of the condition, then checks the conditions, and then decodes the object in a single clean patch line without needing to fuss around

kindred fractal Sep 10, 2023, 9:40 PM

#

The dispatch codec is unnecessary

mental carbon Sep 10, 2023, 9:41 PM

#

It is, but it cleans up the one last place were there is direct json reading in the recipe manager

kindred fractal Sep 10, 2023, 9:41 PM

#

I don't like it, that's all I have to say. Let mojang clean their own code, it's generally not our job

mental carbon Sep 10, 2023, 9:41 PM

#

IMHO Forge should set an example as to how to do it right

#

And this is a very big eye sore

#

It has been for years

#

And it should have been the first method they patched when they added the registry based dispatch

#

Sadly they did not

kindred fractal Sep 10, 2023, 9:42 PM

#

Surely there's a reason why they didn't do it when refactoring recipes

#

Although I can't see it because it looks so painfully obvious

#

You have a point that showing a good example of a dispatch codec is very helpful

#

So alright then, all good. Sorry I'm tired and wasted 15 min of your time and my time 🙃

mental carbon Sep 10, 2023, 9:44 PM

#

It is okey

#

I mean the point of this today, was to not make it perfect for now

#

You can clearly see in this code that this is an area that is heavily being worked up on

kindred fractal Sep 10, 2023, 9:45 PM

#

I know but I dread a temp fix becoming permanent 😄

mental carbon Sep 10, 2023, 9:45 PM

#

And it would surprise me to not see this area changed in the next pre1

kindred fractal Sep 10, 2023, 9:45 PM

#

Hopefully

mental carbon Sep 10, 2023, 9:45 PM

#

Again

#

IMHO the change is fine as is

#

It sets a good example for modders

kindred fractal Sep 10, 2023, 9:45 PM

#

Yeah that's fine

mental carbon Sep 10, 2023, 9:45 PM

#

And realistically I want to pipe the ops through to the recipes as well

#

They are already there

#

Just not into the ingredients

#

I think that would be great to have the full registry ops there

#

To be able to do recipe ingredients based on for example custom data driven registry entries

kindred fractal Sep 10, 2023, 9:46 PM

#

Yeah all the stuff is there...

mental carbon Sep 10, 2023, 9:47 PM

#

So yes

#

Some commits are still on the horizon here

#

Okey so heads up on the condition stuff

#

I am still designing the decoder to deal with the conditions in the data based registry

#

I need to recode and retool some stuff to create a pure decoder for it

#

Mojang stupidly made the method that does the parsing take a decoder

#

But then when it comes to calling the method

#

They only pass codecs into the method as decoders

lean warren Sep 10, 2023, 10:34 PM

#

Orion, ya done goofed: https://github.com/neoforged/Kits/commit/9c5b8b1bfa9677d059a210255b02e77e94961e73#diff-7baf4a63471af873e33f0b8df9721b61994a816bfa9a4cbbb72706322196da46R186
(In case you're wondering, I have access because I helped with the 1.20 port)

rich void Sep 10, 2023, 10:37 PM

#

lean warren Orion, ya done goofed: <https://github.com/neoforged/Kits/commit/9c5b8b1bfa9677d...

what happened?

lean warren Sep 10, 2023, 10:38 PM

#

A typo, basically

mental carbon Sep 10, 2023, 10:39 PM

#

lean warren Orion, ya done goofed: <https://github.com/neoforged/Kits/commit/9c5b8b1bfa9677d...

Where?

lean warren Sep 10, 2023, 10:40 PM

#

You wrote manager.(ConnectionProtocol.LOGIN), it's missing the method name

mental carbon Sep 10, 2023, 10:40 PM

#

Yes that is on purpose

#

I don't know what they method name is

#

And I got kicked for not pushing yesterday evening

#

So this is my current state

lean warren Sep 10, 2023, 10:40 PM

#

Ah, ok, don't mind me then 😄

mental carbon Sep 10, 2023, 10:40 PM

#

Wir the name missing

dull bison Sep 11, 2023, 5:04 AM

#

As of now I can see 8 files with errors left

tranquil salmon Sep 11, 2023, 6:04 AM

#

The food bar has 6 different textures for empty, half, and full and a version for the hunger effect for each

mental carbon Sep 11, 2023, 6:16 AM

#

Yep

#

They split the widgets texture

#

Which is the biggest source of errors still

dull bison Sep 11, 2023, 6:19 AM

#

there are only 2 classes left that are not rendering stuff

IForgeAdvancementBuilder
SimpleChannel

tranquil salmon Sep 11, 2023, 6:19 AM

#

I am working on finishing the ForgeGui class

dull bison Sep 11, 2023, 6:20 AM

#

we definitly need to have a discussion about what to do with IForgeGuiGraphics

#

because a lot of methods it uses are now internal and work differently

tranquil salmon Sep 11, 2023, 6:23 AM

#

also for some reason the condition codecs are registered before the registry

mental carbon Sep 11, 2023, 6:31 AM

#

Who is working on what?

dull bison Sep 11, 2023, 6:32 AM

#

I just commented out the custom blitNineSlicedSized methods in IForgeGuiGraphics

mental carbon Sep 11, 2023, 6:35 AM

#

I am trying to figure out the advancement builder

#

But its documentation is severely lacking.....

#

I think the concept "canBuild" is gone

dull bison Sep 11, 2023, 6:36 AM

#

it is checking if the parent exists before creating

mental carbon Sep 11, 2023, 6:36 AM

#

Yeah i realise that

tranquil salmon Sep 11, 2023, 6:37 AM

#

While I was working on that I decided to remove the FPS part of FPS_GRAPH since it now shows either the network or fps graph

#

And I didn't have to make more fields protected that time

mental carbon Sep 11, 2023, 6:38 AM

#

Ah I think I figured out a patch for this

tranquil salmon Sep 11, 2023, 6:40 AM

#

12 compile errors left

dull bison Sep 11, 2023, 6:41 AM

#

of ffs AbstractSelectionList is not a Widget

mental carbon Sep 11, 2023, 6:47 AM

#

Fixed the advancement builder

#

Okey

#

I am going to tkae a look at simplechannel

#

Simple channel fixed

#

UnicodeGlyph button fixed

#

@tranquil salmon or @dull bison what are you working on ?

#

I have 2 files remaining on my system

#

I have 1 file remaining

#

Who is doing ForgeSlider?

dull bison Sep 11, 2023, 6:51 AM

#

not me

tranquil salmon Sep 11, 2023, 6:51 AM

#

not me either

mental carbon Sep 11, 2023, 6:51 AM

#

I will take care of it then

#

It is a simple render call 😄

#

Okey all errors fixed

#

Compiling

dull bison Sep 11, 2023, 6:53 AM

#

push stabolb

tranquil salmon Sep 11, 2023, 6:53 AM

#

push

mental carbon Sep 11, 2023, 6:53 AM

#

Give me a second double checking my work :XD

#

It is thinking about it......

tranquil salmon Sep 11, 2023, 6:57 AM

#

I edited the build.gradle in a commit a couple days ago that removes the uuid argument

tranquil salmon Sep 11, 2023, 6:58 AM

#

tranquil salmon I edited the build.gradle in a commit a couple days ago that removes the uuid ar...

@mental carbon

mental carbon Sep 11, 2023, 7:08 AM

#

#

<@&1067092163520909374> The obligatory porting screenshot 😄

dull bison Sep 11, 2023, 7:49 AM

#

@mental carbon I did the dispatchUnsafe stuff

tranquil salmon Sep 11, 2023, 8:31 AM

#

@mental carbon Fixed scrolling downwards

mental carbon Sep 11, 2023, 8:34 AM

#

Ah

#

Did I use the wrong value there?

dull bison Sep 11, 2023, 8:55 AM

#

we need to fix the patch for conditions in datapack registries

mental carbon Sep 11, 2023, 8:58 AM

#

Why?

#

I thought I fixed that

dull bison Sep 11, 2023, 8:59 AM

#

in datapack reg stuff the conditions key is "forge:conditions" not "conditions"

mental carbon Sep 11, 2023, 8:59 AM

#

Ahhh

#

Okey

#

Why did they do that?

warm root Sep 11, 2023, 9:23 AM

#

Mods for 1.20.1 will be working?

mental carbon Sep 11, 2023, 9:25 AM

#

Unsure

#

Some yes

#

Some no

dull bison Sep 11, 2023, 9:26 AM

#

hmmm Unknown registry key in ResourceKey[minecraft:root / minecraft:condition_codecs]: forge:mod_loaded

fallow sundial Sep 11, 2023, 9:26 AM

#

why is that registry under mc

mental carbon Sep 11, 2023, 9:27 AM

#

dull bison hmmm `Unknown registry key in ResourceKey[minecraft:root / minecraft:condition_c...

Missing registration?

dull bison Sep 11, 2023, 9:27 AM

#

I'm checking

warm root Sep 11, 2023, 9:29 AM

#

mental carbon Unsure

Will it be possible to test this version?

mental carbon Sep 11, 2023, 9:29 AM

#

Not untill it is released

warm root Sep 11, 2023, 9:31 AM

#

It's a pity, but I hope I don't have to rewrite all my mods

mental carbon Sep 11, 2023, 9:55 AM

#

Did anyone of you have luck with the escape button?

dull bison Sep 11, 2023, 9:56 AM

#

net/minecraftforge/common/crafting/conditions/ConditionalOps.java:187 returns an errored dataresult

mental carbon Sep 11, 2023, 9:56 AM

#

Maybe the wrong codec?

#

Or something with the wrong ops?

dull bison Sep 11, 2023, 9:57 AM

#

the ops is the ConditionalOps and the codec is the ContextRetrievalCodec

mental carbon Sep 11, 2023, 9:58 AM

#

How is that able to return an errored state?

#

?!?

dull bison Sep 11, 2023, 9:58 AM

#

Not a JSON object: [{"type":"forge:mod_loaded","modid":"not_forge"}] is the error in the dataresult

mental carbon Sep 11, 2023, 9:59 AM

#

Ah that is an array

#

Question is where it is throwing that

dull bison Sep 11, 2023, 10:01 AM

#

question is why is it even trying to parse that object since we already have the parsed conditions at that point

mental carbon Sep 11, 2023, 10:03 AM

#

No clue

tranquil salmon Sep 11, 2023, 10:06 AM

#

Were you getting the BootstrapMethodError?

#

@mental carbon

#

So it's exclusive to eclipse then

#

It doesn't like the lambda here: https://github.com/neoforged/Kits/blob/1.20.2/src/main/java/net/minecraftforge/network/NetworkEvent.java#L195
Invalid receiver type class java.lang.Object; not a subtype of implementation type interface net.minecraftforge.network.INetworkDirection

dull bison Sep 11, 2023, 10:16 AM

#

the experiments screen is broken

tranquil salmon Sep 11, 2023, 10:19 AM

#

Seems like there was a patch to ecj for it in june here https://github.com/eclipse-jdt/eclipse.jdt.core/commit/0b6a65a66ed4ae0b3771e50662bdaa5f76ce7036 but hasn't been released yet

#

also worth noting the next eclipse release is in a couple days

dull bison Sep 11, 2023, 10:23 AM

#

the escape issue also affects e with the inventory but only inside the game

#

escape works in the world creation screen

mental carbon Sep 11, 2023, 10:25 AM

#

Yeah

#

Something with the input management

dull bison Sep 11, 2023, 10:25 AM

#

creative inv background is broken

#

creative tab sorting should be validated (I had issues since it became a registry)

rich void Sep 11, 2023, 10:28 AM

#

did it become a registry this update?

dull bison Sep 11, 2023, 10:28 AM

#

no I think that was 1.20

fallow sundial Sep 11, 2023, 10:29 AM

#

yes that was 1.20

#

aslo wdym by "validated"?

dull bison Sep 11, 2023, 10:29 AM

#

spawn eggs being on page 2 with test mods enabled

#

is this correct?

normal wigeon Sep 11, 2023, 10:31 AM

#

WTF

fallow sundial Sep 11, 2023, 10:31 AM

#

dull bison spawn eggs being on page 2 with test mods enabled

well the test mod pushes tabs between vanilla https://github.com/neoforged/NeoForge/blob/1.20.x/src/test/java/net/minecraftforge/debug/CreativeModeTabTest.java

dull bison Sep 11, 2023, 10:35 AM

#

then this is the only issue in the creative inv

mental carbon Sep 11, 2023, 10:35 AM

#

Likely a misalligned patch

#

Or I called the wrong method when I tried to fix it

fallow sundial Sep 11, 2023, 10:47 AM

#

tfw useless supplier

tranquil salmon Sep 11, 2023, 10:48 AM

#

So the escape key seems to be being processed both when you press it and release it

mental carbon Sep 11, 2023, 10:49 AM

#

Aha

dull bison Sep 11, 2023, 10:49 AM

#

fallow sundial tfw useless supplier

if it is useless nuke it

tranquil salmon Sep 11, 2023, 10:54 AM

#

Fixed the issue

dull bison Sep 11, 2023, 10:54 AM

#

the escape one?

tranquil salmon Sep 11, 2023, 10:54 AM

#

yes

#

https://github.com/neoforged/Kits/commit/e0d7bad82e6a6dfaa5a4c95f4fd4e551a6510165

mental carbon Sep 11, 2023, 10:55 AM

#

nICE!

tranquil salmon Sep 11, 2023, 10:58 AM

#

They created a commit to fix it in june here: https://github.com/eclipse-jdt/eclipse.jdt.core/commit/0b6a65a66ed4ae0b3771e50662bdaa5f76ce7036
that commit was created about a week after a release and the next release is in a couple days

#

Also there are no advancements

fallow sundial Sep 11, 2023, 11:13 AM

#

I want to get rid of the generic mess in channels (bifunction and tobooleanbifunction and all that crap) and make them their own FIs. but i'm thonk ing on whether FIs should still have the I prefix

dull bison Sep 11, 2023, 11:15 AM

#

my take on that is no I for FIs

tranquil salmon Sep 11, 2023, 11:17 AM

#

Isn't https://github.com/neoforged/Kits/blob/1.20.2/projects/forge/src/main/java/net/minecraft/advancements/Advancement.java#L93 meant to return null when conditions aren't matched instead of when they are?

arctic sphinx Sep 11, 2023, 11:18 AM

#

Yes, unless it got redesigned

mental carbon Sep 11, 2023, 11:19 AM

#

oeps XD

#

Yopu are right coehlrich

#

My mistake

rich void Sep 11, 2023, 11:19 AM

#

question: is the Forge API gonna be seperate now?

mental carbon Sep 11, 2023, 11:20 AM

#

mental carbon Sep 11, 2023, 11:20 AM

#

rich void question: is the Forge API gonna be seperate now?

No

#

#1136320550168436798 message <@&1067092163520909374> and we are in world. most stuff is working

kindred hatchBOT Sep 11, 2023, 11:20 AM

#

oriononline

[Jump to referenced message.](#1136320550168436798 message)

tranquil salmon Sep 11, 2023, 11:22 AM

#

This doesn't look like an iron sword

#

For each item added to a crafting table you get a random item in the output slot

arctic sphinx Sep 11, 2023, 11:24 AM

#

something something the pen is mighter than the sword

fallow sundial Sep 11, 2023, 11:26 AM

#

ezpz

mental carbon Sep 11, 2023, 11:26 AM

#

tranquil salmon This doesn't look like an iron sword

Ingredient match

fallow sundial Sep 11, 2023, 11:26 AM

#

only 14 files changed to get rid of the context suppliers thinkies

tranquil salmon Sep 11, 2023, 11:27 AM

#

Also it seems like animals follow you even if you aren't holding anything

mental carbon Sep 11, 2023, 11:27 AM

#

Yeah that is all tied to ingredients

#

So we need to look into what gets send around

dull bison Sep 11, 2023, 11:28 AM

#

tranquil salmon This doesn't look like an iron sword

yup

mental carbon Sep 11, 2023, 11:28 AM

#

Allthough not the anymimals

fallow sundial Sep 11, 2023, 11:29 AM

#

dull bison yup

nice recipe

tranquil salmon Sep 11, 2023, 11:44 AM

#

@mental carbon Is the 2nd part of this or meant to be true if one or more item stacks are empty or all are empty
https://github.com/neoforged/Kits/blob/1.20.2/projects/forge/src/main/java/net/minecraft/world/item/crafting/Ingredient.java#L146

#

Currently it's true if none of the item stacks are empty

mental carbon Sep 11, 2023, 11:44 AM

#

That should indeed be allMatch

#

Actually

#

I don't know

#

The way getItems() works is that it returns a list of itemstacks

#

It might need to be allMatch

#

But i am not 100% on that

drowsy condor Sep 11, 2023, 11:53 AM

#

fallow sundial tfw useless supplier

Finally 😩

rich void Sep 11, 2023, 11:54 AM

#

?

fallow sundial Sep 11, 2023, 11:57 AM

#

the loot tables becoming codecs made adding conditions to loot tables much easier harold

#

datagenning loot tables with conditions on the other hand

#

that's going to be fun

#

fun for another time thinkies

kindred fractal Sep 11, 2023, 12:00 PM

#

You should also make the pools conditional for extra fun

fallow sundial Sep 11, 2023, 12:00 PM

#

ehh actually

#

fair point actually

rich void Sep 11, 2023, 12:00 PM

#

so how are loot tables done now? still with a JSON?

mental carbon Sep 11, 2023, 12:01 PM

#

fallow sundial the loot tables becoming codecs made adding conditions to loot tables much easie...

So can I now get a thank you....... Cause yesterday evening you sounded completly oposed to this path 😄

fallow sundial Sep 11, 2023, 12:02 PM

#

I still oppose to deserializing conditions with registry access stabolb

mental carbon Sep 11, 2023, 12:02 PM

#

I know

#

But you have to give it to the solution with the codec

#

It is very nice to just tag the codec on to an existing codec, and bamn now you have a conditional codec

fallow sundial Sep 11, 2023, 12:02 PM

#

¯_(ツ)_/¯

#

where would I put a codec util method

mental carbon Sep 11, 2023, 12:03 PM

#

ForgeExtraCodecs

#

Has most of them

#

Unless it is related to conditional ops

#

Then they are in the ConditionalOps class or in the ICondition interface

true pivot Sep 11, 2023, 12:04 PM

#

That was a fast port 👀

mental carbon Sep 11, 2023, 12:04 PM

#

Well

#

It was however a rather complicated one

#

And we are not done yet

#

We have issues with some networking ,. especially on dedicated servers, and with recipes still

dull bison Sep 11, 2023, 12:09 PM

#

wth Map entry '#' : No key tag in MapLike[{"type":"forge:intersection","children":[{"tag":"minecraft:planks"},{"tag":"minecraft:non_flammable_wood"}]}]; No key item in MapLike[{"type":"forge:intersection","children":[{"tag":"minecraft:planks"},{"tag":"minecraft:non_flammable_wood"}]}]; Not a json array: {"type":"forge:intersection","children":[{"tag":"minecraft:planks"},{"tag":"minecraft:non_flammable_wood"}]}

fallow sundial Sep 11, 2023, 12:18 PM

#

much codec

tranquil salmon Sep 11, 2023, 12:32 PM

#

lethal bane Sep 11, 2023, 12:32 PM

#

shipit

daring ginkgo Sep 11, 2023, 12:46 PM

#

when you put your portal in the hydrolic press

fallow sundial Sep 11, 2023, 12:47 PM

#

time to dive into networking

lethal bane Sep 11, 2023, 12:48 PM

#

maty is gonna implement the config phase 👀

fallow sundial Sep 11, 2023, 12:48 PM

#

wish me luck, I'll hate this

lethal bane Sep 11, 2023, 12:48 PM

#

good luck, let me send people back into config phase

dull bison Sep 11, 2023, 12:51 PM

#

fallow sundial time to dive into networking

what is broken with networking?

#

or are you speaking of nuking the supplier

fallow sundial Sep 11, 2023, 12:52 PM

#

I've already done that

#

I'm talking about the common configuration phase

mental carbon Sep 11, 2023, 1:02 PM

#

@fallow sundial Please donÄt

#

I am currently trying to fix an issue with with packets received by the server having the wrong channel id

fallow sundial Sep 11, 2023, 1:18 PM

#

don't what thinkies

#

also requiresFullSynchronization being inside the ingredient type is limiting. stuff like compound ingredients would need that to be dynamic based on the subingreidnets

mental carbon Sep 11, 2023, 1:24 PM

#

fallow sundial also `requiresFullSynchronization` being inside the ingredient type is limiting....

Yeah, you are right

#

I need to refactor that again later today

#

Will clean that up

fallow sundial Sep 11, 2023, 1:25 PM

#

well I was in the process of doing it

mental carbon Sep 11, 2023, 1:25 PM

#

Ah then please go ahead

#

Don't let me stop you 😄

#

If you can fix the bug with the network on a dedicated server

#

Then please fix that as well

#

Somehow the resource name in the payload is weird

#

And is empty

fallow sundial Sep 11, 2023, 1:26 PM

#

hmmm

mental carbon Sep 11, 2023, 1:27 PM

#

But it works on an integrated server

#

So currently I am pretty stumped

fallow sundial Sep 11, 2023, 1:27 PM

#

well did you check the buffer

#

integrated server packets are copied

mental carbon Sep 11, 2023, 1:27 PM

#

As far as I can tell the buffer is fine

#

But I am 100% on that

fallow sundial Sep 11, 2023, 1:28 PM

#

also may I ask why did you choose to have the ingredient know its type instead of having it override codec() (as is tradition with all mojang registry codecs)

#

it means you can't inherit from custom ingredients unless that ingredient also exposes a constructor with a type

#

which is.. funky

mental carbon Sep 11, 2023, 1:29 PM

#

fallow sundial also may I ask why did you choose to have the ingredient know its type instead o...

Because ingredients have 2 codecs

#

And anything with more then just a simple codec, has historically had a type

dull bison Sep 11, 2023, 1:30 PM

#

door rendering is borked

mental carbon Sep 11, 2023, 1:30 PM

#

dull bison door rendering is borked

Is that the same issue as with the end-portal, maybe?

dull bison Sep 11, 2023, 1:31 PM

#

could be looks like broken face culling

mental carbon Sep 11, 2023, 1:31 PM

#

Weirdly, it looks like that did not update with respects to the changed blockstate.....

fallow sundial Sep 11, 2023, 1:32 PM

#

mental carbon And anything with more then just a simple codec, has historically had a type

I mean yeah, but it having a reference to its type is unneeded, when you can just do

dull bison Sep 11, 2023, 1:32 PM

#

it culls any side that is at the edge of the block

mental carbon Sep 11, 2023, 1:33 PM

#

fallow sundial I mean yeah, but it having a reference to its type is unneeded, when you can jus...

Yet, we don't really do that anywhere....... So........

#

I don't see a reason to start here.....

#

The codec for a recipe is not on the recipe

#

it is on the serializer

#

The exceptions to that mechanic, is actually stuff like our conditions

#

Which have the codec() directly on them

#

And maybe datadriven registries

fallow sundial Sep 11, 2023, 1:34 PM

#

mental carbon Yet, we don't really do that anywhere....... So........

yes? the entirety of worldgen

#

basically all codec registries

mental carbon Sep 11, 2023, 1:34 PM

#

Yeah but these are not codec registries......

fallow sundial Sep 11, 2023, 1:35 PM

#

it's a registry of a record of two codecs, the amount of codecs doesn't change the fact that it's a codec registry

mental carbon Sep 11, 2023, 1:35 PM

#

Does it matter?

#

Like you have been nitpicking about this kind of stuff over the last few days

fallow sundial Sep 11, 2023, 1:35 PM

#

fallow sundial it means you can't inherit from custom ingredients unless that ingredient also e...

^

dull bison Sep 11, 2023, 1:35 PM

#

thinkies

mental carbon Sep 11, 2023, 1:36 PM

#

Might I ask you, to for now, focus your energy on something like actually fixing some bugs that we still have

#

We can discuss the changes then later in the actuall review

fallow sundial Sep 11, 2023, 1:45 PM

#

¯_(ツ)_/¯

dull bison Sep 11, 2023, 1:55 PM

#

Issues:

Rendering

Blocks with broken face culling:
- Cauldron
- Doors
- Fence Gates
- Flower Pot
- Repeater
- Comparator
- Enchanting Table
- End Portal Frame
- Campfire & Soul Campfire
- End Rod
- Jack o'Lantern
- Sculk Sensor & Calibrated Sculk Sensor
- Sculk Catalyst
Stained Glass Panes are opaque

Chests are not placeable

fallow sundial Sep 11, 2023, 1:56 PM

#

Chests are not placeable

#

ah yes

#

make it h1

mental carbon Sep 11, 2023, 1:57 PM

#

dull bison # Issues: ## Rendering - Blocks with broken face culling: - Cauldron - Doors...

Can we put this somewhere were we can edit this?

#

Like a HackMD

lean warren Sep 11, 2023, 1:57 PM

#

dull bison # Issues: ## Rendering - Blocks with broken face culling: - Cauldron - Doors...

This sounds like some shape comparison issue in Block.shouldRenderFace() to me since it seems to specifically affect non-full faces. At a glance I can't see any issue in that method itself though

mental carbon Sep 11, 2023, 1:57 PM

#

Cause we still have some network issues to figure out

fallow sundial Sep 11, 2023, 1:57 PM

#

is it chests specifically or is it any block item

lean warren Sep 11, 2023, 1:58 PM

#

Are you running normal forgedev or with test mods?

dull bison Sep 11, 2023, 1:58 PM

#

lean warren This sounds like some shape comparison issue in `Block.shouldRenderFace()` to me...

it culls all faces that touch the edge of the block

fallow sundial Sep 11, 2023, 1:58 PM

#

also can someone get me a 1.20.1 world zip

lean warren Sep 11, 2023, 1:59 PM

#

dull bison it culls all faces that touch the edge of the block

Yes, but judging from the block list only those ones that don't occupy the whole area of the face, otherwise I would expect the whole world to be "empty"

dull bison Sep 11, 2023, 2:00 PM

#

lean warren Yes, but judging from the block list only those ones that don't occupy the whole...

nope for the door you only see the face on the inside of the block all others are culled (even the full ones)

lean warren Sep 11, 2023, 2:02 PM

#

Hmm, ok, that could be caused a similar way. I'll have to dig a bit when I'm home, I have a few ideas

lean warren Sep 11, 2023, 2:02 PM

#

lean warren Are you running normal forgedev or with test mods?

Btw, Schurli ^

dull bison Sep 11, 2023, 2:02 PM

#

lean warren Btw, Schurli ^

with test mods

lean warren Sep 11, 2023, 2:03 PM

#

Try running without them, there's a dumbass test mod that conditionally cancels chest placement screm

dull bison Sep 11, 2023, 2:03 PM

#

https://hackmd.io/@lE5fCiE4QeW5unaq_y0MQA/SknN6qn03/edit

fallow sundial Sep 11, 2023, 2:03 PM

#

dull bison with test mods

bad idea

#

can confirm, works without test mods

lean warren Sep 11, 2023, 2:04 PM

#

👌

dull bison Sep 11, 2023, 2:06 PM

#

fallow sundial can confirm, works without test mods

the chest or the rendering?

fallow sundial Sep 11, 2023, 2:06 PM

#

btw schurli we already have a neoforged team hackmd 😛

#

chest placing

dull bison Sep 11, 2023, 2:06 PM

#

fallow sundial btw schurli we already have a neoforged team hackmd 😛

why did orion ask me to create one then?

fallow sundial Sep 11, 2023, 2:06 PM

#

you're in the org, you should see it

#

https://hackmd.io/@neoforged/SkXxkj30h

HackMD

1.20.2 porting todo - HackMD

hhhh

#

if not i can get you an invite link

dull bison Sep 11, 2023, 2:10 PM

#

I can edit but I am not in the team

fallow sundial Sep 11, 2023, 2:11 PM

#

hmm

dull bison Sep 11, 2023, 2:11 PM

#

now I am

#

everyone that finds an issue pls add it to the document
if you work on something mark it in the document
when you fix an issue mark it as such in the document

fallow sundial Sep 11, 2023, 2:15 PM

#

powered lamps also don't render

#

dull bison Sep 11, 2023, 2:16 PM

#

unpowered do?

fallow sundial Sep 11, 2023, 2:16 PM

#

yes

mental carbon Sep 11, 2023, 2:16 PM

#

fallow sundial powered lamps also don't render

With or without test mods?

fallow sundial Sep 11, 2023, 2:17 PM

#

without

mental carbon Sep 11, 2023, 2:17 PM

#

Interesting

#

Could be many things.....

#

Sadly

rich void Sep 11, 2023, 2:17 PM

#

u can't binary search this, right?

dull bison Sep 11, 2023, 2:18 PM

#

same thing with redstone ore

mental carbon Sep 11, 2023, 2:18 PM

#

Problems with light emitting full blocks?

#

Check glowstone

dull bison Sep 11, 2023, 2:19 PM

#

yup sea lantern and glowstone

lethal bane Sep 11, 2023, 2:19 PM

#

lean warren Try running without them, there's a dumbass test mod that conditionally cancels ...

someone pr to invert that condition

dull bison Sep 11, 2023, 2:20 PM

#

so we have 2 different problems here the culling and the not rendering of light emitting full blocks

lethal bane Sep 11, 2023, 2:20 PM

#

so that placing chest doesn't work against diamond blocks thinkies

fallow sundial Sep 11, 2023, 2:22 PM

#

I wonder if tesselating without AO is broken in general

dull bison Sep 11, 2023, 2:23 PM

#

dull bison yup sea lantern and glowstone

same for froglights

fallow sundial Sep 11, 2023, 2:24 PM

#

can you even disable ao in the settings menu

dim sonnet Sep 11, 2023, 2:24 PM

#

No, only smooth lighting

fallow sundial Sep 11, 2023, 2:24 PM

#

nice

rich void Sep 11, 2023, 2:25 PM

#

isn't ambient occlusion needed for blocks like cactus?

fallow sundial Sep 11, 2023, 2:27 PM

#

heh I was right

#

tesselatew/oAO is indeeded broken

#

and I also know why

#

let's see if this fixes it

rich void Sep 11, 2023, 2:28 PM

#

let's see

fallow sundial Sep 11, 2023, 2:31 PM

#

Catjam

mental carbon Sep 11, 2023, 2:31 PM

#

?

#

What was it?

#

That whole area was a cluster fuck

#

The patcher layered 3 patches there over each other

#

I likely fucked something up trying to untangle it

#

Nice catch!

fallow sundial Sep 11, 2023, 2:34 PM

#

my fix probably fixed face culling

dull bison Sep 11, 2023, 2:34 PM

#

now mark them as fixed in the HackMD

#

after testing

fallow sundial Sep 11, 2023, 2:37 PM

#

yep

#

stained glass panes are fine on my end

#

berries are broken too

#

I only get damage once when entering them

#

and every time I move

#

but not if I stand still

dull bison Sep 11, 2023, 2:42 PM

#

that is normal vanilla behaviour

fallow sundial Sep 11, 2023, 2:42 PM

#

is it? thinkies

lethal bane Sep 11, 2023, 2:44 PM

#

iirc yes

#

you don't get pricked if you don't move into thorns

fallow sundial Sep 11, 2023, 2:48 PM

#

well then

#

also can anyone get me a 1.20.1 world

jagged agate Sep 11, 2023, 2:49 PM

#

lethal bane _you don't get pricked if you don't move into thorns <:squirrelhat:5352587627229...

Yep. It checks for if entity has velocity.

fallow sundial Sep 11, 2023, 2:51 PM

#

fallow sundial also can anyone get me a 1.20.1 world

preferably one with an entity that has an effect

dull bison Sep 11, 2023, 2:51 PM

#

time for the debug world

#

there is a hole between the black wool and the dandelion

fallow sundial Sep 11, 2023, 3:09 PM

#

okay so I can make the datafixer take the forge WSD into account

#

but it's going to be ugly harold

dull bison Sep 11, 2023, 3:10 PM

#

datafixer are always ugly

fallow sundial Sep 11, 2023, 3:11 PM

#

fair point

mental carbon Sep 11, 2023, 3:13 PM

#

fallow sundial okay so I can make the datafixer take the forge WSD into account

WSD?

fallow sundial Sep 11, 2023, 3:14 PM

#

saveddata

#

same thing

#

WSD is worldsaveddata (the mcp name)

mental carbon Sep 11, 2023, 3:14 PM

#

Ah okes

#

Why is it needed?

#

Did DFU remove it?

fallow sundial Sep 11, 2023, 3:15 PM

#

mental carbon Why is it needed?

because otherwise you can't update a forge world with custom mob effects

#

or well you can but all effects disappear

mental carbon Sep 11, 2023, 3:16 PM

#

Why are custom effects stored in SD?, when they are already hardcoded on the entities/itemstacks additionally?

fallow sundial Sep 11, 2023, 3:16 PM

#

this isn't that bad actually thinkies

fallow sundial Sep 11, 2023, 3:17 PM

#

mental carbon Why are custom effects stored in SD?, when they are already hardcoded on the ent...

name to id mappings are saved in the forge WSD because of the saving feature of forge registries that exists because of mob effects?

mental carbon Sep 11, 2023, 3:17 PM

#

Aaah so the forge reg data is stored in there?

fallow sundial Sep 11, 2023, 3:17 PM

#

yes

mental carbon Sep 11, 2023, 3:17 PM

#

There is one fly in the ointment though: DFU can not access a registry

#

So you would need to figure out a way to get that information in there....

#

IMHO It might be worth it to figure out if the existing DFU Fixer for the mob effects can be redirected to read the Forge field

#

Instead of trying to shoe horn this system into it

#

Cause I am pretty sure we write the full name already

dull bison Sep 11, 2023, 3:25 PM

#

vanilla blocks all work some of the testmod blocks need fixing

fallow sundial Sep 11, 2023, 3:25 PM

#

mental carbon Cause I am pretty sure we write the full name already

we do but it's not enough

#

we don't patch beacons to use the name

mental carbon Sep 11, 2023, 3:26 PM

#

But then you are out of luck anyway are you not?

#

Given that you have no idea what the id is supposed to be?

#

There might be an argument to be had here

#

To say: well we can introduce a hardcut here, or at least a softcut on custom beacon effects

#

Cause I do not think it is a good idea to read in the entire registry in DFU

#

The point of DFU is that it is completely independent of the registry

fallow sundial Sep 11, 2023, 3:40 PM

#

actually, serverlifecyclehooks#getcurrentserver works fine during block/entity fixing. it's player and item fixing that doesn't have the server available

#

and both of those are patched in forge to have the name written

mental carbon Sep 11, 2023, 3:41 PM

#

fallow sundial and both of those are patched in forge to have the name written

Not really

#

The problem is that they all use codecs now to deal with effect list

#

#

We could patch in here

fallow sundial Sep 11, 2023, 3:42 PM

#

no no we don't need to patch that

mental carbon Sep 11, 2023, 3:42 PM

#

But that is just creating TechDebt

fallow sundial Sep 11, 2023, 3:42 PM

#

for entities and stews we can redirect to the forge field

mental carbon Sep 11, 2023, 3:42 PM

#

In DFU?

#

Or what?

fallow sundial Sep 11, 2023, 3:42 PM

#

yes

mental carbon Sep 11, 2023, 3:42 PM

#

Yeah that is what I was coing to point out

#

Good 😄

fallow sundial Sep 11, 2023, 3:43 PM

#

and for beacons we can attempt to get a mapping from the current registry because those have been filled with the old ID by the time the fix runs

#

it's just players specifically (and their inventory) being annoying to deal with as those are stored in the level.dat so there's no server when they're fixed

#

but chunks (blocks and entities) are on-demand after a server was created

#

https://tenor.com/view/this-is-fine-gif-24177057

Tenor

#

@lean warren smol request if you have time. write me a mod registering like 10 effects in a random order, load it up in 1.20.1, give the effect to a cow, a player, and a beacon (I'll run this as a class in forgedev so you're fine to patch VALID_EFFECTS in the beacon BE), and make a suspicious stew out of it and put it in a chest

mental carbon Sep 11, 2023, 3:47 PM

#

I really don't like these kinds of fixes

fallow sundial Sep 11, 2023, 3:48 PM

#

who does harold

#

okay question

muted hemlock Sep 11, 2023, 3:52 PM

#

do we even need to fix this

fallow sundial Sep 11, 2023, 3:52 PM

#

aaaagh

#

fuck

muted hemlock Sep 11, 2023, 3:52 PM

#

it's not the first time that modded saves have lost data due to DFU

mental carbon Sep 11, 2023, 3:52 PM

#

^

#

And mojang fixed it for the future

#

So I am not sure we should even fix this

fallow sundial Sep 11, 2023, 3:53 PM

#

muted hemlock it's not the first time that modded saves have lost data due to DFU

yes but that's because it's not feasible to have everyone provide a datafixer for their BE

#

in this case however it's up to neo, and we can fix it

mental carbon Sep 11, 2023, 3:54 PM

#

fallow sundial yes but that's because it's not feasible to have everyone provide a datafixer fo...

The amount of data loss is negligible

lean warren Sep 11, 2023, 3:54 PM

#

muted hemlock it's not the first time that modded saves have lost data due to DFU

Or even been broken by it entirely (looks at 1.18.2)

mental carbon Sep 11, 2023, 3:54 PM

#

There might be an argument to say:

#

Well we do not fix this

#

We did not do for 1.18.2

#

Or any of the other 100 things that mojang broke

fallow sundial Sep 11, 2023, 3:55 PM

#

that's a stupid argument, if we let worlds be broken before it means we can't do better this time?

pine ember Sep 11, 2023, 3:56 PM

#

We decided not to handle this, doing so allowed us to remove all the code related to persisting registry ids to disk. A 3d party mod could be made to help players who do have a world they wish to upgrade, upgrade but I doubt anyone will actually care tbh.

#

Im happy to be proven wrong tho 😄

mental carbon Sep 11, 2023, 3:58 PM

#

@fallow sundial How many other places still use the int ids for on disk storage?

#

Is it only the chunk palette

#

Once it reaches a certain size?

tacit vale Sep 11, 2023, 3:59 PM

#

mental carbon <@561254664750891009> How many other places still use the int ids for on disk st...

wdym? In 1.20.2 nowhere

mental carbon Sep 11, 2023, 3:59 PM

#

Okey

#

Then I would argue against fixing this

#

And removing all the code that provides that API functionality

tacit vale Sep 11, 2023, 3:59 PM

#

see my registry rework

mental carbon Sep 11, 2023, 3:59 PM

#

Yeah I thought yo

tacit vale Sep 11, 2023, 3:59 PM

#

it still needs some work but it's mostly there

mental carbon Sep 11, 2023, 3:59 PM

#

But did you check the chunk palette

tacit vale Sep 11, 2023, 3:59 PM

#

I checked everything yeah

#

any int ids are only networking

mental carbon Sep 11, 2023, 4:00 PM

#

That had previously still a mode were if it exceeded a specific size it fell back to the global registry map

tacit vale Sep 11, 2023, 4:00 PM

#

I posted like 2 TODOs it still needs earlier in this thread

#

The branch is on my GitHub fork of neoforge if anyone wants to use it

mental carbon Sep 11, 2023, 4:01 PM

#

Sadly

#

The global palette are still in use

dull bison Sep 11, 2023, 4:01 PM

#

tacit vale it still needs some work but it's mostly there

I'd argue you could do a branch on the Kits repo that is rebased on 1.20.2 already

mental carbon Sep 11, 2023, 4:01 PM

#

See PalettedContainer#Strategy

#

Which if the palette exceeds 8 bits

#

it will use the global registry and its int ids

#

WHich means they still are needed and need to be written to disk

#

Cause in a modded world it is actually possible to get more then 256 blockstates in a chunk

#

However I would argue that the HashMap strategy should be extended to cover more bit sizes

#

Maybe even the linear one as well

#

fallow sundial Sep 11, 2023, 4:04 PM

#

we can still do this and forget about the beacon

#

(for those without access)

mental carbon Sep 11, 2023, 4:05 PM

#

#

For DFU yeah that is fine maty

kindred fractal Sep 11, 2023, 4:06 PM

#

mental carbon WHich means they still are needed and need to be written to disk

That sounds wrong

mental carbon Sep 11, 2023, 4:06 PM

#

But my point with respect to the palette and its size still stands

mental carbon Sep 11, 2023, 4:06 PM

#

kindred fractal That sounds wrong

But that is not wrong

#

I researched this area of the code extensively for C&B

#

And ran into this snag

kindred fractal Sep 11, 2023, 4:06 PM

#

@pine ember that's concerning if it's true

mental carbon Sep 11, 2023, 4:07 PM

#

If you have more 2^8 different states in your chunk, the palette on disk will be empty

#

See the GlobalPalette

#

And then it will pull from the IdMapper for the blockstate map

fallow sundial Sep 11, 2023, 4:08 PM

#

fallow sundial (for those without access)

(can someone look over that code again btw, there's a chance I might have inversed parameters or something, a second pair of eyes doesn't hurt)

mental carbon Sep 11, 2023, 4:08 PM

#

This is normally not a problem in Vanilla

#

Given that the amount of blockstates is sufficiently low

#

And especially important: always in the same order

#

New blocks are always at the end

#

So this rarely happens

#

But I think if you cram a debug world in a single chunk you can still trigger it

#

Regardless

pine ember Sep 11, 2023, 4:09 PM

#

kindred fractal <@98473211301212160> that's concerning if it's true

We havent saved block ids to disk for a long time now.

lean warren Sep 11, 2023, 4:10 PM

#

This is about blockstate IDs though and not block IDs

mental carbon Sep 11, 2023, 4:10 PM

#

lean warren This is about block**state** IDs though and not block IDs

Which are tied to each other!

#

Especially on forge

lean warren Sep 11, 2023, 4:11 PM

#

Yes, my point is that block IDs have indeed not been an issue for a while (at least from what I know) while the state IDs have due to the palette thing you just mentioned

mental carbon Sep 11, 2023, 4:11 PM

#

Yep :d

#

Block ids are as such only implicitly a problem

#

Since they determine the order in the block state map

winged hull Sep 11, 2023, 4:12 PM

#

dull bison # Issues: ## Rendering - Blocks with broken face culling: - Cauldron - Doors...

so is it just transparent blocks whose culling is broken?

mental carbon Sep 11, 2023, 4:12 PM

#

dull bison Sep 11, 2023, 4:12 PM

#

winged hull so is it just transparent blocks whose culling is broken?

that's already fixed

mental carbon Sep 11, 2023, 4:13 PM

#

And that means that the id map that is used is and will be different between two launches with different mods

pine ember Sep 11, 2023, 4:14 PM

#

We sort the blocks here

#

It has its own issues, but mostly works

mental carbon Sep 11, 2023, 4:14 PM

#

pine ember It has its own issues, but mostly works

Yeah but it instantly breaks in the same situation

#

Remove a mod

#

And all blockstates are broken

#

If you have more then the 256 different blockstates in that chunk

#

Granted that is a big if

#

But in a modded world with 200 mods that is not unthinkable

pine ember Sep 11, 2023, 4:15 PM

#

yeah, I wasn't aware of that 🤔

#

Is it 256 in the section (thats 161616 if im correct?) or 256 in the whole chunk?

mental carbon Sep 11, 2023, 4:16 PM

#

Whole chunk

pine ember Sep 11, 2023, 4:16 PM

#

Ill need to take a look into this later.

mental carbon Sep 11, 2023, 4:16 PM

#

Palettes are defined on a chunk level

#

not on the section if I remember properly

#

No it seems to be on the section now

#

That would make it significantly harder

#

But not really impossible

#

Sections are 4096 blockstates in size

lean warren Sep 11, 2023, 4:18 PM

#

mental carbon No it seems to be on the section now

1.20.1 has it per section as well

mental carbon Sep 11, 2023, 4:18 PM

#

That is still much larger then the 256 but still

#

I wonder why they did this this way

pine ember Sep 11, 2023, 4:18 PM

#

So yeah, I think this could be an issue for us. But its not a new issue.

mental carbon Sep 11, 2023, 4:18 PM

#

They must have realised that they could easily make an optimal palette using logarithmic growth

#

Which could cover any size they or us would need

lean warren Sep 11, 2023, 4:20 PM

#

mental carbon

Where is this snippet from? Considering it's a static initializer, it's most likely irrelevant to modded blocks (doesn't change anything about the issue though)

mental carbon Sep 11, 2023, 4:20 PM

#

Blocks.java

winged hull Sep 11, 2023, 4:20 PM

#

palletes suck. When generating dimensions in code. It breaks them on join and you have to mixin into the pallete getter to return zero when it wants to return -1.

mental carbon Sep 11, 2023, 4:20 PM

#

Forge replaces the IdMap with a registry link

#

But that registry link does exactly the same thing

dull bison Sep 11, 2023, 4:21 PM

#

FFS MOJANK they changed blockstate and model datagen again

mental carbon Sep 11, 2023, 4:21 PM

#

I realised what is the issue

#

They only use 4 bits for the size

lean warren Sep 11, 2023, 4:21 PM

#

mental carbon Blocks.java

Ah, does Forge replace that static block with the rebuildCache() method?

mental carbon Sep 11, 2023, 4:21 PM

#

Yeah

muted hemlock Sep 11, 2023, 4:21 PM

#

dull bison FFS MOJANK they changed blockstate and model datagen again

🤷‍♀️ it's not like they could have made it any less comprehensible

mental carbon Sep 11, 2023, 4:22 PM

#

I am going to take a look at this later

mossy bobcat Sep 11, 2023, 4:22 PM

#

There's a reason I made all my own stuff for blockstate and model generation for Excavated Variants instead of trying to bundle the vanilla stuff up nicely

muted hemlock Sep 11, 2023, 4:22 PM

#

exactly

mental carbon Sep 11, 2023, 4:22 PM

#

^

#

I have a custom variable growth palette

#

For C&B

quiet talon Sep 11, 2023, 4:40 PM

#

mental carbon Why did they do that?

The eventual plan was unification to forge:conditions instead of the raw conditions to avoid potential name collisions

fallow sundial Sep 11, 2023, 4:44 PM

#

yeah conditions is intrusive as people can't use it for their own codecs

kindred fractal Sep 11, 2023, 4:45 PM

#

quiet talon The eventual plan was unification to `forge:conditions` instead of the raw `cond...

Yeah, would be nice to unify everything

fallow sundial Sep 11, 2023, 4:47 PM

#

we can change it to forge: and use conditions for datapack backwards compat based on the pack version

mental carbon Sep 11, 2023, 4:52 PM

#

We could give the codec the option to check more then one key

dull bison Sep 11, 2023, 6:54 PM

#

writing ingredient with codec produces the wrong output ... it should be flat with the type but it is nested in a value

muted hemlock Sep 11, 2023, 7:04 PM

#

dull bison writing ingredient with codec produces the wrong output ... it should be flat wi...

like, it's doing this? ```json
{
"type": "thing",
"value": {
"field": "etc"
}
}

#

dispatch codecs have a trap you have to watch out for

#

trying to remember exactly what causes it...

#

there's a bit that checks if x instanceof MapCodecCodec

#

yeah, each subcodec must be a MapCodecCodec

#

err, it doesn't have to be, but if it's not, it serializes ala the above format

#

now, how do you make a MapCodecCodec? there's two main ways:

all codecs produced by RecordCodecBuilder are MCCs
fieldOf("name").codec() produces a MCC

#

important bit: xmap does not produce an mcc

#

if you're using fieldof to make a single-field codec, you have to xmap first, then codec()

#

    
    public static record IntHolder(int n)
    {
        public static final Codec<IntHolder> WRONG = Codec.INT.fieldOf("n").codec().xmap(IntHolder::new, IntHolder::n);
        public static final Codec<IntHolder> RIGHT = Codec.INT.fieldOf("n").xmap(IntHolder::new, IntHolder::n).codec();
    }

dull bison Sep 11, 2023, 7:23 PM

#

yea Codec#either fucks that up

mossy bobcat Sep 11, 2023, 7:24 PM

#

Basically, xmap your MapCodec not your Codec and it'll all be happy. That particular bit of weirdness has been a real pain for me

muted hemlock Sep 11, 2023, 7:25 PM

#

dull bison yea Codec#either fucks that up

it shouldn't thonk

dull bison Sep 11, 2023, 7:25 PM

#

I know that but I can't do either with a mcc and a normal codec

mossy bobcat Sep 11, 2023, 7:26 PM

#

muted hemlock it shouldn't <:thonk:932605815221788672>

It will if you do it after the map codec step

muted hemlock Sep 11, 2023, 7:26 PM

#

err where are you using the either

#

are you using the either for one of the dispatcher's subcodecs or is the dispatch codec one of the two eithered codecs

dull bison Sep 11, 2023, 7:27 PM

#

both actually

muted hemlock Sep 11, 2023, 7:27 PM

#

I guess only the former would actually cause this problem, so let's consider that one

#

what two formats are you trying to let this subcodec have

#

if you only have one field but you can interpret the value two different ways, then that's easily done:

eitherCodec.fieldOf("etc").xmap(to, from).codec();

dull bison Sep 11, 2023, 7:38 PM

#

problem is the either is outside of the .codec() call

muted hemlock Sep 11, 2023, 7:38 PM

#

can you just show what you have then

dull bison Sep 11, 2023, 8:00 PM

#

    public static final Codec<CompoundIngredient> CODEC = ExtraCodecs.withAlternative(
          Ingredient.CODEC_LIST.fieldOf("children").codec(),
          ExtraCodecs.withAlternative(
                Ingredient.CODEC_LIST.fieldOf("ingredients").codec(),
                Ingredient.CODEC_LIST
          )
    ).xmap(CompoundIngredient::new, CompoundIngredient::getChildren);

kindred hatchBOT Sep 11, 2023, 8:01 PM

#

dull bison ``` public static final Codec<CompoundIngredient> CODEC = ExtraCodecs.withAl...

Created Gist at the request of @kindred fractal: https://gist.github.com/NeoCamelot/04bdb8e9c5b223ba15671cb49424c346

kindred fractal Sep 11, 2023, 8:01 PM

#

Ok that's a cool feature ^^

muted hemlock Sep 11, 2023, 8:04 PM

#

okay, just to be clear, these are the three json formats it can have? ```json
"ingredient": {
"type": "forge:compound",
"children": []
}

"ingredient": {
"type": "forge:compound",
"ingredients": []
}

"ingredient": []

dull bison Sep 11, 2023, 8:04 PM

#

yes

muted hemlock Sep 11, 2023, 8:04 PM

#

okay, there's two things wrong here

#

firstly that third one cannot be part of the dispatch codec

dull bison Sep 11, 2023, 8:05 PM

#

orion I think we can not use the compound ingredient through the dispatch but need to use an outer either

muted hemlock Sep 11, 2023, 8:05 PM

#

secondly, for the first two, I would either drop one of them, or give them two different subcodecs (two different type ids)

muted hemlock Sep 11, 2023, 8:07 PM

#

dull bison orion I think we can not use the compound ingredient through the dispatch but ne...

the third one cannot be done via dispatch subcodec but it can be done via the outer either, yeah

#

no type field = can't be a dispatch subcodec

dull bison Sep 11, 2023, 8:08 PM

#

muted hemlock secondly, for the first two, I would either drop one of them, or give them two d...

I know how to do it without the third

muted hemlock Sep 11, 2023, 8:09 PM

#

is the third a vanilla format? or is that something forge adds

dull bison Sep 11, 2023, 8:11 PM

#

it was the previous forge format afaik

#

the 2 named ones are the new formats

muted hemlock Sep 11, 2023, 8:12 PM

#

if it's not a vanilla ingredient format then IMO drop it and make people use the explicit type

dull bison Sep 11, 2023, 9:39 PM

#

orion we need to fix ingredients with a list of values (reading works but writing errors)

mental carbon Sep 11, 2023, 9:41 PM

#

dull bison orion we need to fix ingredients with a list of values (reading works but writin...

Okey I will tackle it tomorrow morning asap

#

Somebody else then needs to deal with the networking issue on dedicated servers

dull bison Sep 11, 2023, 9:42 PM

#

I will check the testmod rendering things

tacit vale Sep 11, 2023, 10:06 PM

#

dull bison I'd argue you could do a branch on the Kits repo that is rebased on 1.20.2 alrea...

Me personally I don't think I have enough time to do that right now

quiet talon Sep 11, 2023, 10:10 PM

#

Plus the reg refactor has to be rebased on DH

tacit vale Sep 11, 2023, 10:12 PM

#

well that still needs to be merged which means reviewed if not already

mossy bobcat Sep 11, 2023, 10:21 PM

#

@muted hemlock @dull bison You should be able to use Codec.mapEither to make a MapCodec that works like either, right?

muted hemlock Sep 11, 2023, 10:21 PM

#

maybe

#

that wasn't the problem here though

mossy bobcat Sep 11, 2023, 10:21 PM

#

I think it'd solve the problem though?

#

Oh, except for that direct list one, which needs t ogo outside of the delegate codec anyways

#

But the other part can be Codec.mapEither(Ingredient.CODEC_LIST.fieldOf("children"), Ingredient.CODEC_LIST.fieldOf("ingredients")).xmap(...whatever...).codec()

#

But yeah that third one makes no sense because you kinda have to specify a type key in order to delegate

dull bison Sep 12, 2023, 4:32 AM

#

I am already using mapEither but as you said it doesn't work for tge list one which actually is a vanilla feature ... for reading it has a fallback but for writing it errors

dull bison Sep 12, 2023, 9:03 AM

#

I think I fixed ingredients (I hope so) and I am currently working on forges loot table provider

dull bison Sep 12, 2023, 11:22 AM

#

I think we are ready to cleanup and gen patches and then switch to pre2

mental carbon Sep 12, 2023, 11:30 AM

#

Test the dedicated server

dull bison Sep 12, 2023, 12:07 PM

#

I will need to start in offline mode

#

you started to debug this

mental carbon Sep 12, 2023, 12:18 PM

#

I got that fixed hold on

fallow sundial Sep 12, 2023, 12:19 PM

#

I expected
that "I" doesn't fit

mental carbon Sep 12, 2023, 12:20 PM

#

fallow sundial > I expected that "I" doesn't fit <:harold:932605815406342174>

?

#

Pushed btw

dull bison Sep 12, 2023, 12:28 PM

#

wtf a unicode char as packet name

mental carbon Sep 12, 2023, 12:29 PM

#

That is the empty string

#

Something is sending the server a weird packet

dull bison Sep 12, 2023, 12:50 PM

#

it's an answer packet

arctic sphinx Sep 12, 2023, 1:02 PM

#

Snapshot?

#

<@&1067092163520909374> https://www.minecraft.net/en-us/article/minecraft-1-20-2-pre-release-3

mental carbon Sep 12, 2023, 1:03 PM

#

dull bison it's an answer packet

Yeah but why is it decoded so weirdly

mossy bobcat Sep 12, 2023, 1:05 PM

#

arctic sphinx <@&1067092163520909374> <https://www.minecraft.net/en-us/article/minecraft-1-20-...

Ooh they fixed something with records?

arctic sphinx Sep 12, 2023, 1:05 PM

#

Greg (I think it was?) pr'd to Proguard a fix for empty records having metadata stripped

mossy bobcat Sep 12, 2023, 1:06 PM

#

Ah, nice

arctic sphinx Sep 12, 2023, 1:06 PM

#

It was based on an issue Jas (VF) raised to them

#

Ah Gegy, https://github.com/Guardsquare/proguard-core/pull/118, my appologies harold

tranquil salmon Sep 12, 2023, 1:08 PM

#

arctic sphinx <@&1067092163520909374> <https://www.minecraft.net/en-us/article/minecraft-1-20-...

It's not even 2am yet in nz

true pivot Sep 12, 2023, 1:08 PM

#

greg harold

dull bison Sep 12, 2023, 1:09 PM

#

mental carbon Yeah but why is it decoded so weirdly

we are reading the resource location too early there is another var int in there before it

arctic sphinx Sep 12, 2023, 1:09 PM

#

true pivot greg <:harold:932605815406342174>

Had the name in my mind, likely from the meme/spam that was posted in general earlier

lethal bane Sep 12, 2023, 1:09 PM

#

dull bison we are reading the resource location too early there is another var int in there...

kek

muted hemlock Sep 12, 2023, 1:10 PM

#

hoglins can now be bread

dull bison Sep 12, 2023, 1:11 PM

#

hoglins can now what?

true pivot Sep 12, 2023, 1:11 PM

#

muted hemlock hoglins can now be bread

it still concerns me how they broke that

kindred fractal Sep 12, 2023, 1:11 PM

#

🥖

lethal bane Sep 12, 2023, 1:15 PM

#

https://www.minecraft.net/en-us/article/minecraft-1-20-2-pre-release-3

#

ah wait

#

I'm slow

#

screm

#

I should just also apply for the role

arctic sphinx Sep 12, 2023, 1:16 PM

#

I would hand it out but the last time I did it gave access to some back channels harold

true pivot Sep 12, 2023, 1:16 PM

#

hehe

lethal bane Sep 12, 2023, 1:16 PM

#

arctic sphinx I would hand it out but the last time I did it gave access to some back channels...

yeah, I know harold

#

but why

tranquil salmon Sep 12, 2023, 1:19 PM

#

#builds I was wondering why it hadn't sent a message when I checked turns out I just had to wait a bit longer

daring ginkgo Sep 12, 2023, 1:23 PM

#

arctic sphinx I would hand it out but the last time I did it gave access to some back channels...

I apply for the role the moment it doesnt give back channels :p

lethal bane Sep 12, 2023, 1:23 PM

#

same

daring ginkgo Sep 12, 2023, 1:24 PM

#

especially if it means I will never have to navigate the minecraft site because I do not know how to use that

mental carbon Sep 12, 2023, 1:25 PM

#

dull bison we are reading the resource location too early there is another var int in there...

Can you hop into a call?

#

Cause I was sure I checked that

dull bison Sep 12, 2023, 1:26 PM

#

the erroring packet is:
id: 2
transactionId: 1
payloadPresent: true
packetIndex: 1
id: fml:loginwrapper

lethal bane Sep 12, 2023, 1:26 PM

#

#builds perfect shipit

dull bison Sep 12, 2023, 1:26 PM

#

I know what we are missing ... we are not using readNullable

tranquil salmon Sep 12, 2023, 1:26 PM

#

CompoundTag patch didn't apply

mental carbon Sep 12, 2023, 1:28 PM

#

dull bison the erroring packet is: id: 2 transactionId: 1 payloadPresent: true packetIndex:...

Yeah I know that

dull bison Sep 12, 2023, 1:28 PM

#

fixed

mental carbon Sep 12, 2023, 1:28 PM

#

Seriously?!?

#

That was all it needed the readNullable?

dull bison Sep 12, 2023, 1:29 PM

#

at least this issue is fixed now

mental carbon Sep 12, 2023, 1:29 PM

#

OIkey

#

Lets see if you can get on the server

dull bison Sep 12, 2023, 1:29 PM

#

mental carbon That was all it needed the readNullable?

yes because that writes a boolean

mental carbon Sep 12, 2023, 1:29 PM

#

Oef

dull bison Sep 12, 2023, 1:31 PM

#

I'm on the server

mental carbon Sep 12, 2023, 1:31 PM

#

Nice so

dull bison Sep 12, 2023, 1:35 PM

#

pushed

mental carbon Sep 12, 2023, 1:40 PM

#

Okey

#

So that was the last issue if I remember

#

So I will go through and create the patches

#

AKA do the cleanup

tranquil salmon Sep 12, 2023, 1:42 PM

#

Attempted to update the CompoundTag patch from mobile

tacit vale Sep 12, 2023, 1:44 PM

#

nice

tacit vale Sep 12, 2023, 1:45 PM

#

tranquil salmon Attempted to update the CompoundTag patch from mobile

Wait, how did you even know what to change?

tranquil salmon Sep 12, 2023, 1:46 PM

#

1.20.2-pre2 to 1.20.2-pre3 diff

tacit vale Sep 12, 2023, 1:46 PM

#

ah

mental carbon Sep 12, 2023, 1:47 PM

#

https://gist.github.com/marchermans/65300b65387addeb13eaa74cfc8a7093#file-gistfile1-txt-L199-L449 @dull bison

Gist

gist:65300b65387addeb13eaa74cfc8a7093

GitHub Gist: instantly share code, notes, and snippets.

tranquil salmon Sep 12, 2023, 2:03 PM

#

#builds

mental carbon Sep 12, 2023, 2:11 PM

#

Nice

mental carbon Sep 12, 2023, 2:51 PM

#

https://github.com/neoforged/Srg2Source.git

GitHub

neoforged/Srg2Source

Contribute to neoforged/Srg2Source development by creating an account on GitHub.

#

https://github.com/neoforged/Bus/commit/cb38ccd41626ea2bb5f8c350a3d57b20c59a6fe3#diff-660a0483d27a7df9ef700886a52d2714ac135f25af6889269798e06d37fff664

GitHub

7.0 and Neoify · neoforged/Bus@cb38ccd

dull bison Sep 12, 2023, 3:13 PM

#

can't publish to maven local proguard fails

#

(also need push access to the repo)

rich void Sep 12, 2023, 3:19 PM

#

mental carbon https://github.com/neoforged/Srg2Source.git

what's this gonna be? I was wondering, but there is nothing there rn

dull bison Sep 12, 2023, 3:20 PM

#

rich void what's this gonna be? I was wondering, but there is nothing there rn

the same as under MinecraftForge but without lex

arctic sphinx Sep 12, 2023, 3:20 PM

#

rich void what's this gonna be? I was wondering, but there is nothing there rn

Srg2Source is a source code remapping tool

rich void Sep 12, 2023, 3:21 PM

#

interesting

rich void Sep 12, 2023, 3:43 PM

#

mental carbon https://github.com/neoforged/Bus/commit/cb38ccd41626ea2bb5f8c350a3d57b20c59a6fe3...

so eventbus is bus now? that's it?

fallow sundial Sep 12, 2023, 3:44 PM

#

mental carbon https://github.com/neoforged/Srg2Source.git

we have an action for creating hard forks including tags

#

https://github.com/neoforged/actions/actions/workflows/fork_repository.yml

#

dull bison Sep 12, 2023, 3:46 PM

#

then delete it and re fork it

#

and give me push access so I can push the neoify commit

mental carbon Sep 12, 2023, 3:50 PM

#

Ah okey

#

Will do in a minute

rich void Sep 12, 2023, 3:56 PM

#

tranquil salmon 1.20.2-pre2 to 1.20.2-pre3 diff

what changed?

mental carbon Sep 12, 2023, 4:06 PM

#

Actually that is taking me a bit longer

#

Can somebody run the action for schurli

inland mesa Sep 12, 2023, 4:25 PM

#

so it looks like they are ramping down and doing polishing phase? which means maybe rc1 tomorrow/thursday and 1.120.2 next tuesday maybe?

rich void Sep 12, 2023, 4:27 PM

#

did they finish the stuff they were working on in rc1?

inland mesa Sep 12, 2023, 4:28 PM

#

rc1?

#

they released pre3 today

rich void Sep 12, 2023, 4:28 PM

#

I meant pre1 xD

inland mesa Sep 12, 2023, 4:29 PM

#

can't remember what they were doing in pre1 so I'll leave that to the porters

rich void Sep 12, 2023, 4:29 PM

#

sum codec stuff

inland mesa Sep 12, 2023, 4:29 PM

#

I don't think they have made any more code structure changes in pre3, only bugfixing

rich void Sep 12, 2023, 4:30 PM

#

ok

mental carbon Sep 12, 2023, 4:50 PM

#

@dull bison Did some body run the action for you?

dull bison Sep 12, 2023, 4:51 PM

#

nope

mental carbon Sep 12, 2023, 4:55 PM

#

Okey let me do it for you then

#

Running now

#

Done

dull bison Sep 12, 2023, 5:03 PM

#

I still need push access

mental carbon Sep 12, 2023, 5:04 PM

#

Yeah I am trying to figure that out

#

@fallow sundial Your action is broken

#1.20.2

Issues:

Rendering

Chests are not placeable