Move more data into sections, world height implementation

[paulevsGitch]

This small PR will make possible to change level height. It includes:

• Storing metadata and both light arrays inside chunk sections
• Storing heightmaps in short format instead of byte format
• Loading data from old worlds into new worlds
• Storing world height in level.dat and providing it in the interface

Bonus: short world (height set to 80 blocks with NBT Explorer):

[GameHerobrine]

c-custom world height =o?

[paulevsGitch]

c-custom world height =o?

Some sort of. It will require some more work - splitting chunk section rendering and some changes in light code

[paulevsGitch]

I added test terrain generator. It will be used in dev environment and when world height is larger than 128. It should be removed before next release (I marked is for removal, not whole mixin, only some functions)

[calmilamsy]

You should probably move your test stuff to the testmod workspace instead.

[paulevsGitch]

True, I will fix this in next commit

[GameHerobrine]

hm does the last commit means i can try it =o?

[paulevsGitch]

Yes, you can try it if you want

[paulevsGitch]

Should world height be per-dimensional or per-world? Current implementation is per-world

[theishiopian]

Per dimension would be cool

[calmilamsy]

Dimensional would be preferred, or even better, both, though I'm not sure of a use-case for the latter.

[theishiopian]

Maybe the world one is the default, which dimensions can override?

[calmilamsy]

Yea, likely something like that.

[GameHerobrine]

hm custom dimension height

[paulevsGitch]

Maybe the world one is the default, which dimensions can override?

This actually makes it per-dimensional. There is no reason to store both values for world and dimension, so there can be only dimensional height with default value of 128. If this will be done as interface - any mod can implement it and change height into desirable one. It will be also possible for mods to change vanilla height with mixins as mixins can override interface functions of another mixins.

It will be not to hard to implement as each chunk already ask a world for section count, so now it will ask dimension about that

[paulevsGitch]

Mobs are now spawning on any world height, items works correctly, the last visible issue that is not fixed yet is light. I'm not sure why it is not completely correct (looks like there are more hardcoded constants somewhere in the code), so any help is welcome

P.S.: I optimised some parts of light calculations. They are not source if the issue, but light should work faster than in vanilla for any sections count

Edit: I forgot about move height from level to dimension, but this is not harderst part

[paulevsGitch]

Looks like I fixed light. Next step will be to move world height into dimension settings

[paulevsGitch]

StAPI now can support per-dimensional height, and any dimension can define custom height limit.

Example for custom dimension:

public class CoolDimension extends Dimension implements StationDimension {
    @Override
    short getDefaultLevelHeight() {
        return 256;
    }
    // Other code
}

Example for vanilla dimension (with Mixin):

@Mixin(Overworld.class)
public class OverworldMixin implements StationDimension {
    @Override
    short getDefaultLevelHeight() {
        return 256;
    }
}

I found a strange bug when Nether with vanilla generator will cause infinity world loading, when generator is custom this will not happen. I'm not sure why this happens, looks like it is not dependent from world height as it will stuck on vanilla height range too

[GameHerobrine]

wonder how will it work with aether

[paulevsGitch]

I found a strange bug when Nether with vanilla generator will cause infinity world loading, when generator is custom this will not happen. I'm not sure why this happens, looks like it is not dependent from world height as it will stuck on vanilla height range too

I fixed that. The reason was that Nether generator is different from Overworld and tried to directly put blocks byte data into chunks. As chunks have only states that failed and resulted as empty chunks. Now this is fixed

[paulevsGitch]

I think that this PR is ready for review

[mineLdiver]

Wrong import rules?

[mineLdiver]

I think I've submitted this review wrong.

Let's just use vanilla import rules then

[mineLdiver]

Pointed out some minor issues, but otherwise the PR is ready to be merged.

[mineLdiver]

Seems good, merging.