Registerable WorldFormat api by NotStirred · Pull Request #1795 · chunky-dev/chunky

NotStirred · 2024-12-07T06:25:03Z

I mainly want comments on general structure, everything else is bikesheddable. I'll go over everything again and polish it up before review.

Lots of changes here, so far I've only looked at loading worlds from the gui, haven't looked at reloading from the previous
session

Chunk, World, Dimension are now abstract classes, the previous implementation is now JavaChunk etc.
there is also a skeleton of registerable WorldFormats. Each is asked if a directory is valid for it, if the user selects it, the format is asked to load the world.

The ultimate goal would probably be to have Chunk be an implementation detail of java world, but that's a much larger task as it would require a rewrite of parts of the map view and scene chunk loading

leMaik

I like the overall direction and the abstractions. 👍
Just some small suggestions, maybe we can simplify some code patterns while we're at it.

NotStirred · 2025-08-23T17:37:43Z

Still need to figure out the region situation, as currently a dimension needs regions for it to be viewable on the map

Additionally we could provide a simplified api (for things like schematics etc.) which implements regions, regionchangewatcher, chunkdata stuff. But I think this PR is big enough already.

all are now abstract

NotStirred · 2026-06-27T20:13:25Z

Not sure we need to go with the virtual regions approach as discussed before #1795 (comment)

The changes required to support either virtual regions, or a per-world-format impl are both extremely small

Regions only escape dimensions in map view in one place, and scene in one place.

leMaik · 2026-06-27T21:47:25Z

As long as we can somehow select areas in worlds, that's fine. Regions might be a thing that is special to java worlds anways. Even being limited to chunks is a very technical thing. Why would a user want to be limited to loading a 16x16 blocks grid.

We can figure that out later, but the whole chunk/region selection could be a UI-only thing.

leMaik

Mostly minor comments on the API. Overall this refactoring is still great and it's nice to see how CubicChunks, Java edition and, in the future, bedrock support fit together.

It seems like the whole regions and chunks stuff is mostly needed for the UI (outside of java world loading, of course)?

leMaik · 2026-06-27T21:50:08Z

+/** For worlds that have multiple dimensions, and fully support the map view */
+public interface WorldFormat extends Registerable {
+  /**
+   * This method will be called on every possible world directory (typically this is every directory in `.minecraft/saves`).
+   *


The javadoc here doesn't explain what this class does or what the isValid method does.

leMaik · 2026-06-27T21:55:08Z

    world.currentDimension().removeChunkTopographyListener(this);
-    World newWorld = World.loadWorld(world.getWorldDirectory(), currentDimensionId,
-        World.LoggedWarnings.NORMAL);
-    newWorld.currentDimension().addChunkTopographyListener(this);
+    world.loadDimension(currentDimensionId);
+    world.currentDimension().addChunkTopographyListener(this);


Removing the listeners, loading the same dimension again, and adding the listeners could be handled internally in e.g. world.currentDimension().reload() or World.reloadCurrentDimension()

WorldMapLoader and World were tightly coupled before, so this might be out-of-scope for this PR.

leMaik · 2026-06-27T21:58:14Z

-      mapView.setYMax(256);
-    }
+    mapLoader.loadWorldFromDirectory(PersistentSettings.getLastWorld());
+    IntIntPair heightRange = mapLoader.getWorld().currentDimension().heightRange();


From an API standpoint, maybe this should be HeightRange with getYMin() and getYMax(), that way we could have javadoc and document whether the values are inclusive or exclusive (and maybe have some util methods).

leMaik · 2026-06-27T21:59:21Z

-    } else {
-      yMin.setRange(0, 256);
-      yMax.setRange(0, 256);
+    if (world != null) {


Should we disable the sliders if no world is loaded and/or set them to some default ranges and values?

leMaik · 2026-06-27T22:23:18Z

+  /**
+   * Region X chunk width
+   */
+  int CHUNKS_X = 32;


Is this independent of the world format? If this is just a UI thing for selection, we should move it into the UI classes.

Yes, it's used in scene's chunk loading to batch by region, and in the map view for per-region selecting

As well as being used in JavaRegion for its internal state

leMaik · 2026-06-27T22:24:39Z

+import java.io.IOException;
+import java.nio.file.Path;
+
+/** For worlds that have multiple dimensions, and fully support the map view */


Since "has no dimensions" is the same as "has one dimension" and that would be supported, I guess we could lower the restrictions. The map view could draw schematics too, so that's nothing the world format should be bothered about.

Maybe change it to `A world format that Chunky can load."

leMaik · 2026-06-27T22:26:06Z

+    if (providedWorlds.size() > 1) {
+      // Maybe allow the user to select which?
+      // This method is called from a variety of different popup/menu situations, is this ^ possible?
+      Log.warn(String.format("The directory %s has multiple valid world formats: %s", dir.getAbsolutePath(), String.join(", ", providedWorlds.keySet())));


Load a random one, that'll be fun in #help 🤣

Currently the order is based on arbitrary hashmap ordering.

Though two world formats saying the same world is compatible in both would be... interesting.

leMaik · 2026-06-27T22:29:34Z

  }

+  @Override
+  public Set<Dimension.Identifier> availableDimensions() {


getAvailableDimensions and getDefaultDimension

leMaik · 2026-06-27T22:31:24Z

+  protected int gameMode = 0;
+  protected final long seed;

  /** Timestamp for level.dat when player data was last loaded. */


*Timestamp when player data was last... I don't know. Loaded? Saved?

👍 Also this should be in JavaWorld

leMaik reviewed Jul 28, 2025

View reviewed changes

NotStirred force-pushed the feature/world_format branch 2 times, most recently from 200298a to bcf8d12 Compare August 23, 2025 17:28

NotStirred marked this pull request as ready for review August 23, 2025 17:31

NotStirred changed the title ~~WIP: Registerable WorldFormat api~~ Registerable WorldFormat api Aug 23, 2025

NotStirred added 5 commits June 27, 2026 20:59

Major refactor of world, dimension, and chunk

c0651ed

all are now abstract

Basics of WorldFormat interface

2f74b6a

Replace all world loading with worldformat

c409c38

Make WorldFormats registerable

86a8740

Rename MC_ classes to Java_

56e05d9

NotStirred force-pushed the feature/world_format branch from bcf8d12 to 56e05d9 Compare June 27, 2026 20:06

NotStirred requested a review from leMaik June 27, 2026 20:08

leMaik requested changes Jun 27, 2026

View reviewed changes

NotStirred added 5 commits June 28, 2026 11:19

Create HeightRange and use it in place of the awkward IntIntPair

733ddb1

Various method renames

af78724

Move player data timestamp into JavaWorld

f17f3fd

MC->Java rename misc. cleanup

195a5e3

Add more detailed javadoc to WorldFormat apis

700fad2

NotStirred requested a review from leMaik June 28, 2026 10:25

Uh oh!

Conversation

NotStirred commented Dec 7, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

leMaik left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

NotStirred commented Aug 23, 2025

Uh oh!

NotStirred commented Jun 27, 2026

Uh oh!

leMaik commented Jun 27, 2026

Uh oh!

leMaik left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

NotStirred commented Dec 7, 2024 •

edited

Loading