Blocks

Blocks are essential to the Minecraft world. They make up all the terrain, structures, and machines. Chances are if you are interested in making a mod, then you will want to add some blocks. This page will guide you through the creation of blocks, and some of the things you can do with them.

One Block to Rule Them All

Before we get started, it is important to understand that there is only ever one of each block in the game. A world consists of thousands of references to that one block in different locations. In other words, the same block is just displayed a lot of times.

Due to this, a block should only ever be instantiated once, and that is during registration. Once the block is registered, you can then use the registered reference as needed.

Unlike most other registries, blocks can use a specialized version of DeferredRegister, called DeferredRegister.Blocks. DeferredRegister.Blocks acts basically like a DeferredRegister<Block>, but with some minor differences:

• They are created via DeferredRegister.createBlocks("yourmodid") instead of the regular DeferredRegister.create(...) method.
• #register returns a DeferredBlock<T extends Block>, which extends DeferredHolder<Block, T>. T is the type of the class of the block we are registering.
• There are a few helper methods for registering block. See below for more details.

So now, let's register our blocks:

//BLOCKS is a DeferredRegister.Blocks
public static final DeferredBlock<Block> MY_BLOCK = BLOCKS.register("my_block", () -> new Block(...));

After registering the block, all references to the new my_block should use this constant. For example, if you want to check if the block at a given position is my_block, the code for that would look something like this:

level.getBlockState(position) // returns the blockstate placed in the given level (world) at the given position
        .is(MyBlockRegistrationClass.MY_BLOCK);

This approach also has the convenient effect that block1 == block2 works and can be used instead of Java's equals method (using equals still works, of course, but is pointless since it compares by reference anyway).

危险

Do not call new Block() outside registration! As soon as you do that, things can and will break:

• Blocks must be created while registries are unfrozen. NeoForge unfreezes registries for you and freezes them later, so registration is your time window to create blocks.
• If you try to create and/or register a block when registries are frozen again, the game will crash and report a null block, which can be very confusing.
• If you still manage to have a dangling block instance, the game will not recognize it while syncing and saving, and replace it with air.

Creating Blocks

As discussed before, we start by creating our DeferredRegister.Blocks:

public static final DeferredRegister.Blocks BLOCKS = DeferredRegister.createBlocks("yourmodid");

Basic Blocks

For simple blocks which need no special functionality (think cobblestone, wooden planks, etc.), the Block class can be used directly. To do so, during registration, instantiate Block with a BlockBehaviour.Properties parameter. This BlockBehaviour.Properties parameter can be created using BlockBehaviour.Properties#of, and it can be customized by calling its methods. The most important methods for this are:

• destroyTime - Determines the time the block needs to be destroyed.
  • Stone has a destroy time of 1.5, dirt has 0.5, obsidian has 50, and bedrock has -1 (unbreakable).
• explosionResistance - Determines the explosion resistance of the block.
  • Stone has an explosion resistance of 6.0, dirt has 0.5, obsidian has 1,200, and bedrock has 3,600,000.
• sound - Sets the sound the block makes when it is punched, broken, or placed.
  • The default value is SoundType.STONE. See the Sounds page for more details.
• lightLevel - Sets the light emission of the block. Accepts a function with a BlockState parameter that returns a value between 0 and 15.
  • For example, glowstone uses state -> 15, and torches use state -> 14.
• friction - Sets the friction (slipperiness) of the block.
  • Default value is 0.6. Ice uses 0.98.

So for example, a simple implementation would look something like this:

//BLOCKS is a DeferredRegister.Blocks
public static final DeferredBlock<Block> MY_BETTER_BLOCK = BLOCKS.register(
        "my_better_block", 
        () -> new Block(BlockBehaviour.Properties.of()
                .destroyTime(2.0f)
                .explosionResistance(10.0f)
                .sound(SoundType.GRAVEL)
                .lightLevel(state -> 7)
        ));

For further documentation, see the source code of BlockBehaviour.Properties. For more examples, or to look at the values used by Minecraft, have a look at the Blocks class.

备注

It is important to understand that a block in the world is not the same thing as in an inventory. What looks like a block in an inventory is actually a BlockItem, a special type of item that places a block when used. This also means that things like the creative tab or the max stack size are handled by the corresponding BlockItem.

A BlockItem must be registered separately from the block. This is because a block does not necessarily need an item, for example if it is not meant to be collected (as is the case with fire, for example).

More Functionality

Directly using Block only allows for very basic blocks. If you want to add functionality, like player interaction or a different hitbox, a custom class that extends Block is required. The Block class has many methods that can be overridden to do different things; see the classes Block, BlockBehaviour and IBlockExtension for more information. See also the Using blocks section below for some of the most common use cases for blocks.

If you want to make a block that has different variants (think a slab that has a bottom, top, and double variant), you should use blockstates. And finally, if you want a block that stores additional data (think a chest that stores its inventory), a block entity should be used. The rule of thumb here is that if you have a finite and reasonably small amount of states (= a few hundred states at most), use blockstates, and if you have an infinite or near-infinite amount of states, use a block entity.

Block Types

Block types are MapCodecs used to serialize and deserialize a block object. This MapCodec is set via BlockBehaviour#codec and registered to the block type registry. Currently, its only use is when the block list report is being generated. A block type should be created once for every subclass of Block. For example, FlowerBlock#CODEC represents the block type for most flowers while its subclass WitherRoseBlock has a separate block type.

If the block subclass only takes in the BlockBehaviour.Properties, then BlockBehaviour#simpleCodec can be used to create the MapCodec.

// For some block subclass
public class SimpleBlock extends Block {

    public SimpleBlock(BlockBehavior.Properties properties) {
        // ...
    }

    @Override
    public MapCodec<SimpleBlock> codec() {
        return SIMPLE_CODEC.value();
    }
}

// In some registration class
public static final DeferredRegister<MapCodec<? extends Block>> REGISTRAR = DeferredRegister.create(BuiltInRegistries.BLOCK_TYPE, "yourmodid");

public static final DeferredHolder<MapCodec<? extends Block>, MapCodec<SimpleBlock>> SIMPLE_CODEC = REGISTRAR.register(
    "simple",
    () -> simpleCodec(SimpleBlock::new)
);

If the block subclass contains more parameters, then RecordCodecBuilder#mapCodec should be used to create the MapCodec, passing in BlockBehaviour#propertiesCodec for the BlockBehaviour.Properties parameter.

// For some block subclass
public class ComplexBlock extends Block {

    public ComplexBlock(int value, BlockBehavior.Properties properties) {
        // ...
    }

    @Override
    public MapCodec<ComplexBlock> codec() {
        return COMPLEX_CODEC.value();
    }

    public int getValue() {
        return this.value;
    }
}

// In some registration class
public static final DeferredRegister<MapCodec<? extends Block>> REGISTRAR = DeferredRegister.create(BuiltInRegistries.BLOCK_TYPE, "yourmodid");

public static final DeferredHolder<MapCodec<? extends Block>, MapCodec<ComplexBlock>> COMPLEX_CODEC = REGISTRAR.register(
    "simple",
    () -> RecordCodecBuilder.mapCodec(instance ->
        instance.group(
            Codec.INT.fieldOf("value").forGetter(ComplexBlock::getValue),
            BlockBehaviour.propertiesCodec() // represents the BlockBehavior.Properties parameter
        ).apply(instance, ComplexBlock::new)
    );
);

备注

Although block types are basically unused at the moment, it is expected to become more important in the future as Mojang continues moving towards a codec-centered structure.

DeferredRegister.Blocks helpers

We already discussed how to create a DeferredRegister.Blocks above, as well as that it returns DeferredBlocks. Now, let's have a look at what other utilities the specialized DeferredRegister has to offer. Let's start with #registerBlock:

public static final DeferredRegister.Blocks BLOCKS = DeferredRegister.createBlocks("yourmodid");

public static final DeferredBlock<Block> EXAMPLE_BLOCK = BLOCKS.registerBlock(
        "example_block",
        Block::new, // The factory that the properties will be passed into.
        BlockBehaviour.Properties.of() // The properties to use.
);

Internally, this will simply call BLOCKS.register("example_block", () -> new Block(BlockBehaviour.Properties.of())) by applying the properties parameter to the provided block factory (which is commonly the constructor).

If you want to use Block::new, you can leave out the factory entirely:

public static final DeferredBlock<Block> EXAMPLE_BLOCK = BLOCKS.registerSimpleBlock(
        "example_block",
        BlockBehaviour.Properties.of() // The properties to use.
);

This does the exact same as the previous example, but is slightly shorter. Of course, if you want to use a subclass of Block and not Block itself, you will have to use the previous method instead.

Resources

If you register your block and place it in the world, you will find it to be missing things like a texture. This is because textures, among others, are handled by Minecraft's resource system. When adding a new block in Minecraft, you should either write or generate the following files:

• A blockstate file
• A block model
• A translation
• A loot table
• Some block tags, e.g. for mining

For all of the above, also reference the files and data generators of similar vanilla blocks.

Using Blocks

Blocks are very rarely directly used to do things. In fact, probably two of the most common operations in all of Minecraft - getting the block at a position, and setting a block at a position - use blockstates, not blocks. The general design approach is to have the block define behavior, but have the behavior actually run through blockstates. Due to this, BlockStates are often passed to methods of Block as a parameter. For more information on how blockstates are used, and on how to get one from a block, see Using Blockstates.

In several situations, multiple methods of Block are used at different times. The following subsections list the most common block-related pipelines. Unless specified otherwise, all methods are called on both logical sides and should return the same result on both sides.

Placing a Block

Block placement logic is called from BlockItem#useOn (or some subclass's implementation thereof, such as in PlaceOnWaterBlockItem, which is used for lily pads). For more information on how the game gets there, see the Interaction Pipeline. In practice, this means that as soon as a BlockItem is right-clicked (for example a cobblestone item), this behavior is called.

• Several prerequisites are checked, for example that you are not in spectator mode, that all required feature flags for the block are enabled or that the target position is not outside the world border. If at least one of these checks fails, the pipeline ends.
• BlockBehaviour#canBeReplaced is called for the block currently at the position where the block is attempted to be placed. If it returns false, the pipeline ends. Prominent cases that return true here are tall grass or snow layers.
• Block#getStateForPlacement is called. This is where, depending on the context (which includes information like the position, the rotation and the side the block is placed on), different block states can be returned. This is useful for example for blocks that can be placed in different directions.
• BlockBehaviour#canSurvive is called with the blockstate obtained in the previous step. If it returns false, the pipeline ends.
• The blockstate is set into the level via a Level#setBlock call.
  • In that Level#setBlock call, BlockBehaviour#onPlace is called.
• Block#setPlacedBy is called.

Breaking a Block

Breaking a block is a bit more complex, as it requires time. The process can be roughly divided into three stages: "initiating", "mining" and "actually breaking".

• When the left mouse button is clicked, the "initiating" stage is entered.
• Now, the left mouse button needs to be held down, entering the "mining" stage. This stage's methods are called every tick.
• If the "continuing" stage is not interrupted (by releasing the left mouse button) and the block is broken, the "actually breaking" stage is entered.

Or for those who prefer pseudocode:

leftClick();
initiatingStage();
while (leftClickIsBeingHeld()) {
    miningStage();
    if (blockIsBroken()) {
        actuallyBreakingStage();
        break;
    }
}

The following subsections further break down these stages into actual method calls.

The "Initiating" Stage

• Client-only: InputEvent.InteractionKeyMappingTriggered is fired with the left mouse button and the main hand. If the event is canceled, the pipeline ends.
• Several prerequisites are checked, for example that you are not in spectator mode, that all required feature flags for the ItemStack in your main hand are enabled or that the block in question is not outside the world border. If at least one of these checks fails, the pipeline ends.
• PlayerInteractEvent.LeftClickBlock is fired. If the event is canceled, the pipeline ends.
  • Note that when the event is canceled on the client, no packets are sent to the server and thus no logic runs on the server.
  • However, canceling this event on the server will still cause client code to run, which can lead to desyncs!
• Block#attack is called.

The "Mining" Stage

• PlayerInteractEvent.LeftClickBlock is fired. If the event is canceled, the pipeline moves to the "finishing" stage.
  • Note that when the event is canceled on the client, no packets are sent to the server and thus no logic runs on the server.
  • However, canceling this event on the server will still cause client code to run, which can lead to desyncs!
• Block#getDestroyProgress is called and added to the internal destroy progress counter.
  • Block#getDestroyProgress returns a float value between 0 and 1, representing how much the destroy progress counter should be increased every tick.
• The progress overlay (cracking texture) is updated accordingly.
• If the destroy progress is greater than 1.0 (i.e. completed, i.e. the block should be broken), the "mining" stage is exited and the "actually breaking" stage is entered.

The "Actually Breaking" Stage

• Item#canAttackBlock is called. If it returns false (determining that the block should not be broken), the pipeline moves to the "finishing" stage.
• Player#canUseGameMasterBlocks is called if the block is an instance of GameMasterBlock. This determines whether the player has the ability to destroy creative-only blocks. If false, the pipeline moves to the "finishing" stage.
• Server-only: Player#blockActionRestricted is called. This determines whether the current player cannot break the block. If true, the pipeline moves to the "finishing" stage.
• Server-only: BlockEvent.BreakEvent is fired. If canceled or getExpToDrop returns -1, the pipeline moves to the "finishing" stage. The initial canceled state is determined by the above three methods.
  • Server-only: PlayerEvent.HarvestCheck is fired. If HarvestCheck#canHarvest returns false or the BlockState passed into the break event is null, then the initial exp for the event will be 0.
  • Server-only: IBlockExtension#getExpDrop is called if PlayerEvent.HarvestCheck#canHarvest returns true. This value is passed to BlockEvent.BreakEvent#getExpToDrop to be used later in the pipeline.
• Server-only: IBlockExtension#canHarvestBlock is called. This determines whether the block can be harvested, i.e. broken with drops.
• IBlockExtension#onDestroyedByPlayer is called. If it returns false, the pipeline moves to the "finishing" stage. In that IBlockExtension#onDestroyedByPlayer call:
  • Block#playerWillDestroy is called.
  • The blockstate is removed from the level via a Level#setBlock call with Blocks.AIR.defaultBlockState() as the blockstate parameter.
    • In that Level#setBlock call, Block#onRemove is called.
• Block#destroy is called.
• Server-only: If the previous call to IBlockExtension#canHarvestBlock returned true, Block#playerDestroy is called.
  • Server-only: Block#dropResources is called. This determines what drops from the block when mined.
    • Server-only: BlockDropsEvent is fired. If the event is canceled, then nothing is dropped when the block breaks. Otherwise, every ItemEntity in BlockDropsEvent#getDrops is added to the current level.
• Server-only: Block#popExperience is called with the result of the previous IBlockExtension#getExpDrop call, if that call returned a value greater than 0.

Ticking

Ticking is a mechanism that updates (ticks) parts of the game every 1 / 20 seconds, or 50 milliseconds ("one tick"). Blocks provide different ticking methods that are called in different ways.

Server Ticking and Tick Scheduling

BlockBehaviour#tick is called in two occasions: either through default random ticking (see below), or through scheduled ticks. Scheduled ticks can be created through Level#scheduleTick(BlockPos, Block, int), where the int denotes a delay. This is used in various places by vanilla, for example, the tilting mechanism of big dripleaves heavily relies on this system. Other prominent users are various redstone components.

Client Ticking

Block#animateTick is called exclusively on the client, every frame. This is where client-only behavior, for example the torch particle spawning, happens.

Weather Ticking

Weather ticking is handled by Block#handlePrecipitation and runs independent of regular ticking. It is called only on the server, only when it is raining in some form, with a 1 in 16 chance. This is used for example by cauldrons that fill during rain or snowfall.

Random Ticking

The random tick system runs independent of regular ticking. Random ticks must be enabled through the BlockBehaviour.Properties of the block by calling the BlockBehaviour.Properties#randomTicks() method. This enables the block to be part of the random ticking mechanic.

Random ticks occur every tick for a set amount of blocks in a chunk. That set amount is defined through the randomTickSpeed gamerule. With its default value of 3, every tick, 3 random blocks from the chunk are chosen. If these blocks have random ticking enabled, then their respective BlockBehaviour#randomTick methods are called.

Random ticking is used by a wide range of mechanics in Minecraft, such as plant growth, ice and snow melting, or copper oxidizing.