[1.21.1] [Crash] Conflict with Botania (and possibly other mods)

[ArtemShustov]

I wanted to play on the new version of Minecraft, but ran into a crash when using ProductiveBees and Botania. (Botania hasn't officially released for 1.21.x, so I built the mod from 1.21.1-porting).
When loading into a world, after ~5 seconds or even immediately I get a crash:

java.lang.ClassCastException: class net.minecraft.core.BlockPos$MutableBlockPos cannot be cast to class net.minecraft.world.level.block.state.BlockState (net.minecraft.core.BlockPos$MutableBlockPos and net.minecraft.world.level.block.state.BlockState are in module minecraft@1.21.1 of loader 'TRANSFORMER' @765ffb14)

crash-2026-03-12_17.39.32-server.txt

I'm not a Java developer and don't fully understand how mixins work, but as far as I can tell the issue is caused by BeePollinateGoalMixin. (added in the 1.21.x branch, no issues on 1.20.1)
Your mixin signature:

private Predicate<BlockPos> VALID_POLLINATION_BLOCKS;
private Optional<BlockPos> findNearestBlock(Predicate<BlockPos> predicate, double distance)

Vanilla code:

private final Predicate<BlockState> VALID_POLLINATION_BLOCKS
private Optional<BlockPos> findNearestBlock(Predicate<BlockState> predicate, double distance)

Botania also uses BlockState and its code hasn't changed since 1.20.x.
I think the code should be rewritten to avoid issues in the future.

Environment:
Minecraft: 1.21.1
NeoForge: 21.1.219
Mods:

1. botania-neoforge-1.21.1-452-SNAPSHOT
2. curios-neoforge-9.5.1+1.21.1
3. geckolib-neoforge-1.21.1-4.8.4
4. Patchouli-1.21.1-92-NEOFORGE
5. productivebees-1.21.1-13.13.1

[JaisDK]

I don't shadow VALID_POLLINATION_BLOCKS, I make a new predicate to use, but I wonder if Mixin gets confused because it has the same name and tries to extend both of them.

The functionality they're implementing is another issue in itself as it will probably overwrite my flowering block requirements.

[furtabs]

Can confirm this is an issue
https://mclo.gs/c2J6mlT

[JaisDK]

productivebees-1.21.1-13.13.3.zip

does this one fix it? need to unzip it into the mods folder

[xPotato4u]

I just updated the mod from curseforge (modrinth still only has 13.13.0 in prismlauncher) and the Issue persists
https://mclo.gs/FnFq11v

[JaisDK]

the file I linked here is not on cf

[xPotato4u]

Okay, I thought they were the same because both files had the same Name and filesize. I downloaded the jar you provided, but it still crashed immediately with the same error as before.
https://mclo.gs/64Hjjia

[TheRealWormbo]

I don't shadow VALID_POLLINATION_BLOCKS, I make a new predicate to use, but I wonder if Mixin gets confused because it has the same name and tries to extend both of them.

I just saw this issue linked in our channel, and that hypothesis seems sound. Botania's code attempts to replace the vanilla predicate (or any modified version of it) by taking the value and appending an alternative condition. That on its own should not interfere with your code trying to completely replace the logic. The issue is probably that the Botania mixin sees the injected private field of the same name and doesn't quite know what to do. Your field probably needs a different name and/or the @Unique annotation to make sure it doesn't replace the vanilla one, since that has a different compile-time type, but the same runtime/"erased" type.

The functionality they're implementing is another issue in itself as it will probably overwrite my flowering block requirements.

It shouldn't, as far as I can tell, since the Botania mixin and the Productive Bees mixin want to operate on completely different data. That said, I think reading this and the original class made me realize Botania probably doesn't append the proper check in the first place, as e.g. it ignores waterlogged floating special flowers. Getting that to work with this mod is a whole different story, though.

[JaisDK]

In the linked jar I did just that, renamed it and put a Unique annotation on it

https://gist.github.com/JaisDK/474f84981c18572d097c2dd60fcd2897

[TheRealWormbo]

Hmm, that's strange. I have no idea why Botania's mixin still sees your predicate instance, then.

[ChiefArug]

I obtained a class export (from -Dmixin.debug.export=true) from someone who was having this issue and the issue is due to mixin's Overwrite-by-default behaviour.
In your mixin you declare private Optional<BlockPos> findNearestBlock(Predicate<BlockPos> predicate, double distance) with no annotations. Because the signature matches the signature of an existing method in the class (remember that generics are not a part of runtime signatures!), it overwrites that method. This means that when vanilla calls the method in findNearbyFlower it passes the Predicate to your new findNearestBlock that treats it like a Predicate<BlockPos>!
This works without botania because you used to also overwrite the field. However in the jar above where you fixed the field being overwritten, it will also crash in vanilla due to the vanilla predicate being passed to your overwritten findNearestBlock.

The solution here (in addition to what you have already done by renaming the field) is to put @Unique on your findNearestBlock method, and also add this to your mixin config to cause a hard crash when you accidentally overwrite a method:

  "overwrites": {
    "requireAnnotations": true
  },

[JaisDK]

Thanks for the input, let's see if that took care of the issue.

productivebees-1.21.1-13.13.4-pre.zip

[xPotato4u]

I tested it for 10 Minutes in the world where before it crashed everytime after 2 seconds, so far no crash, so it seems to be working.
Another Question though: Don't you need a ci.setReturnValue at the End to return the result of findNearestBlock?

[JaisDK]

that would make sense...