The sides of trees are cut off on chunk boundaries

[naalit]

Upon further investigation, this seems to happen every time a tree spans more than one chunk - I haven't seen any trees which have successfully generated blocks in more than one chunk. Here's an example:

I wonder if this is because the side chunks aren't loaded when it tries to set the blocks, and so the setting of blocks is silently ignored (a problem which was discussed on Discord some time ago)?

[skaldarnar]

I guess the simple explanation here is that the respective facet provider(s) don't use any borders at all, e.g.,

GrowingFlora/src/main/java/org/terasology/gf/generator/TreeProvider.java

Lines 18 to 20 in 17f1d51

	@Produces(TreeFacet.class)
	@Requires(@Facet(FloraFacet.class))
	public class TreeProvider implements FacetProvider {

The counterpart in CoreWorlds does use borders for its dependencies:

https://github.com/Terasology/CoreWorlds/blob/02a4cb208b6067cb84017d06af0972b1bacc43e5/src/main/java/org/terasology/core/world/generator/facetProviders/DefaultTreeProvider.java#L46-L52

It also extends the facet area it produces in the process() method:

https://github.com/Terasology/CoreWorlds/blob/02a4cb208b6067cb84017d06af0972b1bacc43e5/src/main/java/org/terasology/core/world/generator/facetProviders/DefaultTreeProvider.java#L99-L100

I guess the tricky part in general is to know what the maximum dimensions of trees are to figure out how big the border must be.

[naalit]

This is fixed for the trees that MR currently generates by the latest commit in #38 which adds a border. Larger trees will still be broken, but making it work with trees of arbitrary sizes will take a lot of work and a lot of things will need to be figured out, so that's on hold for now.