Hello,

 

I have a strange problem with a tickable TileEntity.

My tile is checking a blockstate property of it's block each update tick, which normally works fine.

But when the chunk is unloaded on client side, there is one tick where the tileentity is still updated, but the block at it's position is (already) an air block.

 

Here is my tile:

https://github.com/TeamLapen/Vampirism/blob/e0a40c903fd8481a5d5c93cb327b78e497462cff/src/main/java/de/teamlapen/vampirism/tileentity/TileCoffin.java#L75

and the block:

https://github.com/TeamLapen/Vampirism/blob/e0a40c903fd8481a5d5c93cb327b78e497462cff/src/main/java/de/teamlapen/vampirism/blocks/BlockCoffin.java#L56

 

When moving away from a placed coffin and it is unloaded, there is one tick with an air block at the position and my code logs the stacktrace:

at de.teamlapen.vampirism.blocks.BlockCoffin.isHead(BlockCoffin.java:57) [blockCoffin.class:?]

at de.teamlapen.vampirism.tileentity.TileCoffin.update(TileCoffin.java:75) [TileCoffin.class:?]

at net.minecraft.world.World.updateEntities(World.java:1936) [World.class:?]

at net.minecraft.client.Minecraft.runTick(Minecraft.java:1880) [Minecraft.class:?]

at net.minecraft.client.Minecraft.runGameLoop(Minecraft.java:1114) [Minecraft.class:?]

at net.minecraft.client.Minecraft.run(Minecraft.java:401) [Minecraft.class:?]

at net.minecraft.client.main.Main.main(Main.java:118) [Main.class:?]

at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) ~[?:1.8.0_51]

at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) ~[?:1.8.0_51]

at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) ~[?:1.8.0_51]

at java.lang.reflect.Method.invoke(Method.java:497) ~[?:1.8.0_51]

 

I have verifiered the problem with another TileEntity as well. (At serverside there seems to be no problem (with remote check removed of cause).

 

Is there a bug in my code or is this intended by Forge or is it a Forge bug?

I guess it is a forge bug, see post below.

 

Minecraft Forge: v12.16.0.1865

 

Thanks

Alright I took a second look at the Forge/Minecraft code and might have an idea where this problem is coming from:

 

The world#updateEntities method checks if the block is loaded before ticking it, but it uses World#isBlockLoaded(BlockPos) for that, which calls #isBlockLoaded(BlockPos,true), which also considers empty chunks as loaded. Apparently unloaded chunks on client side are (sometimes) converted to empty chunks. So the update loop thinks the block is loaded, and updates the tile entity even though the tile's block is already unloaded. Unfortunatly the tile entity (which is marked to be removed by Chunk#onChunkUnload) is removed AFTER the update method was called.

 

 

Can somebody confirm this?

I want to create a Forge issue, but I would prefer if someone could confirm this issue beforehand. Maybe I'm missing something