- RGBA16 and R11F_G11F_B10F buffers get their own formats where the device has them
(16-bit normalized render targets need the adapter's format features): the float stand-in
kept NaN and values past 1, and BSL's TAA history went black for good.
- Textures share one sampler variable per kind (linear, nearest, repeat, nearest-repeat,
comparison): devices have as few as 16 sampler slots, Complementary's terrain program
declares 23 textures.
- A gbuffers pass reading the depth buffer or a buffer it draws into reads a blank texture
instead (a feedback loop: undefined in GL, refused by wgpu, and a crash here).
- A bind group layout that fails validation skips the program with the error's details
instead of ending the game.
- shadowModelView / shadowProjection (and inverses) for programs that read them before the
shadow pass exists: an orthographic camera 160 blocks each side, toward the shadow light.
- HORIZON_SHADER_PACK_SKIP=a,b leaves programs out (a developers' aid to bisect a pack).
Clean-room: implemented from cleanroom/shader-packs-spec.md revision 1, sections 6.1, 6.7,
5.2; no other loader's code read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`uniform.<type>.<name>` and `variable.<type>.<name>` of shaders.properties (3.4): an
expression parser with C precedence, the function table (math, vectors, `if`, `between`,
`equals`, `smooth`, `random`...), evaluated on the CPU once per frame in file order after the
frame's own uniforms. `smooth` keeps its state per call site and halves the distance to its
target every fade time (OQ-26 working answer). A name neither the frame nor the pack defines
reads 0 and is listed once in the log.
Clean-room: implemented from cleanroom/shader-packs-spec.md revision 1, section 3.4; no other
loader's code read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
naga keeps only constant initializers of globals, so every initialized global that is not
`const` (and every conditional one) gets its value from a function of ours that main calls
first, one per declarator, converting as GLSL's initialization does (`vec2 v = atlasSize`).
`vec3 a = x, b = y;` no longer loses `b`.
Program-enable expressions (3.3) now see the pack's switch options that are on by default
(8.1): `#define NAME` with nothing after it, tested with `#ifdef`/`#ifndef` anywhere in the
pack (OQ-13 working answer). BSL's `deferred` (ambient occlusion) and its TAA ran with options
off before.
Clean-room: implemented from cleanroom/shader-packs-spec.md revision 1, sections 3.3, 8.1;
no other loader's code read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
predicate_property was the same fifteen lines in block_feature and
state_provider; it is worldgen::bound_property now. The mojang link helpers are
unix-only in use but worth testing everywhere, so they are allowed to look dead
elsewhere, which clears the launcher's -D warnings red on Windows. The bots
harness recovers from a poisoned lock instead of panicking.
The moderation lists on a dedicated server, command storage, boss bars, the
world clocks, the difficulty and the emergency save before a crash all used
let _ = and lost the error. write_dat also swallowed the rename that moves the
current file aside, which made the following rename fail on the existing path
with nothing reporting it.
PrimedTnt keeps Entity's NO_OP InterpolationHandler and ticks on a client
too: a position packet puts it there at once, its own tick moves it on
(gravity, collisions, bounces, drag) and burns the fuse down, and a client
discards it when the fuse runs out. Horizon interpolated it towards each
update instead, so it trailed about two ticks behind the server.
lan_e2e's falling check now drops primed TNT too and checks that the
guest has it exactly where its fall from rest puts it for the fuse the
server sent.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
link cannot symlink a directory on Windows, so mirror has to recurse into one
itself, but it only did so when a pack had filters. An untouched directory fell
through to the link branch and vanished from the overlay, which took
textures/font and the panorama with it and made a pack that overrides a font
crash the client before its window.
A resource pack written while it streams (Nexo's server behind
resourcepack.gommehd.io, gommehd.net's required pack) leaves every local
header a stub, every uncompressed size at 0xFFFFFFFF and the END record's
entry counts and disk number at 0 and 0xFFFF. Java tolerates all three;
the zip crate's ZipArchive refuses the END record as multi-disk and empty,
so a required server pack failed its reload and the join never got through
(each retry also rebuilt the renderer, once a second).
Count the real central-directory entries and normalize the END record's
disk and count fields when they disagree with it, and bound the inflate
output when an entry's size is the 0xFFFFFFFF sentinel.
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
scoreboard::canonical_component already re-encodes a component as vanilla
does, list form included, so decorate_name uses it instead of its own copy.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A component list is its first element with the rest as siblings, so
decorateName merges the frame colour under that element's style. The
parsed list moved a styled first element down a level, which gave the
description the frame colour instead of the title's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Advancement.decorateName merges the frame colour under the title's own
style and appends the description after its siblings. Built from the
parsed Text, a styled root with siblings had its style moved down a level,
so the description showed in the frame colour instead of the title's
(a gold, bold title gave a green description). The hover is now built on
the title's JSON, and the local spans are decoded from that JSON, so an
in-frame player sees the description coloured as a client does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A server sends an item's and a falling block's position only every 20 ticks.
As a client (multiplayer, a LAN guest, single player on its server thread)
Horizon moved them only when a position arrived, three steps towards it, so a
dropped item or falling sand hung in the air and then jumped down once a
second (#1568).
Vanilla runs ItemEntity.tick and FallingBlockEntity.tick on the client too
(gravity, collisions, drag), takes their motion from AddEntity
(recreateFromPacket) and SetEntityMotion (Entity.lerpMotion), and snaps them to
each position packet: neither overrides Entity's NO_OP InterpolationHandler.
The client now does the same, with the same ticks single player in-frame uses;
only the server removes them or places the block.
lan_e2e --check falling: the guest samples both every 2 ticks over 16 ticks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A loopback test listens on both [::1] and 127.0.0.1 and checks the
connection lands on IPv4 although IPv6 is listed first; it fails without
the sort. Another checks the fallback to IPv6 when IPv4 refuses. When every
address fails, the error is now the first address's (the one vanilla tries)
rather than an IPv6 fallback's, which is often only "no route to host".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The effect capped downward gravity but left the fall distance to accumulate, so the landing tick still dealt fall damage. Clear it at the start of check_fall_damage, where both local travel and the server thread's replay of a guest's move pass.
Co-Authored-By: opencode (deepseek-v4.1-flash-free) <noreply@opencode.ai>
CustomBossEvents.onPlayerConnect adds a joining player to its bars before any
entity, dragon fight or raid tells it about theirs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The item-tab hotbar loop only set the tooltip, so the nine slots never
showed the white hover fill the 5x9 grid uses. Apply the same fill,
including over empty slots, before picking the tooltip item.
Fixes#1341
Co-Authored-By: opencode (deepseek-v4.1-flash-free) <noreply@opencode.ai>
The host wrote no trim component, so single player on its server thread,
LAN guests and dedicated-server players saw armour without its trim.
network_stack now takes the connection's synced-registry lookup for any
registry and writes Component::Trim when both names are in it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A server sends its packs in the configuration phase, while the joining
screen covers the frame; vanilla's ToastManager draws over every screen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The host diffed each viewer's boss events into a Change enum, then turned the
changes into packets for guests and, through a second match, into setter calls
on the window's own overlay; the client turned packets into the same setters a
third way. The diff now yields ClientboundBossEventPacket values, guests get
them as packets and the local overlay applies them with BossOverlay::apply,
the same function a server's packets reach. Gone: Change, Snap, Sent,
apply_local, boss_packet and the overlay's nine-argument add and setters.
UUIDUtil.uuidFromIntArray joins uuid_ints in lan::playerdata.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The settings screen's own options (controller, vibration, dead zones and the
rest) are not vanilla options and have no vanilla keys, so their captions
were English literals. They now read horizon.options.* from assets/horizon/lang,
which the client bundles as one more language source placed first, so a pack
or the prepared root can still override it. English and Russian ship here.
The pre-world overlay said "Joined as {name}; loading terrain..." after
Event::Joined. Vanilla's LevelLoadingScreen only ever draws the
multiplayer.downloadingTerrain text there — use that translation, no
name.
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
The trim decode sat inside the read-past fallback and matched the kind
there; it now has a TRIM arm, and the fallback only reads past. Long
comment lines wrapped.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
What the toast showed differed from vanilla 26.3 in several ways:
- the title and the failure line had no arguments: download.pack.title takes
the request number and the batch size, download.pack.failed the failures
and the batch size;
- sizes read "1.3 KiB" style; com.mojang.realmsclient.Unit.humanReadable gives
"1.3 KB" ("512 B" below 1 KiB, half-up rounding), checked against the jar;
- each update didn't restart SystemToast's 5 s display time, so a download
longer than 5 s lost its toast, and the Notification Time option was ignored;
- a finished batch removed the toast at once instead of forceHide's slide out;
- the failure title sat at y 7 in white; SystemToast draws titles yellow, and a
title without a message at y 12;
- a second batch replaced the first one's channel and lost its events.
The download thread now reports HttpUtil's listener calls (requestStart,
downloadStart with the Content-Length, downloadedBytes, requestFinished), and
each batch gets a PackNotifier (createDownloadNotifier) with its own toast. The
toast is a shared toasts::SystemToast; the system background moved to
toasts.rs with a y, so the pack toasts stack under an advancement toast.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With a pack whose gbuffers_terrain (or a fallback) loads, solid and
cutout terrain is drawn by it into the pack's buffers, in place, between
prepare and deferred, and Horizon's own solid terrain draw is skipped.
- depthtex0 is now a Depth32Float attachment: the blit writes the
world's depth there (GL window depth, GL rows), terrain tests and
writes against it, so entities and terrain occlude each other; later
programs sample it as before.
- The vertex glue reads Horizon's TerrainVertex on both draw paths
(instanced, and flat with the page tables in bind group 1) and gives
atlas coordinates, lightmap coordinates in 0-240, the vertex colour,
mc_midTexCoord at the sprite's centre and mc_Entity = -1. Normals,
tangents and block ids come later.
- texture/gtexture/tex sample the block atlas, lightmap the lightmap;
atlasSize is set.
- Occlusion culling is off while a pack draws terrain (it reads the
depth of Horizon's terrain draw), and flat draws stay on.
- Uniforms a program reads that are not provided yet are reported.
- Program building is shared by quads and terrain (build_pass, bind).
- terrain-test: Horizon's own pack for this slice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Advancement.decorateName appends a new line and the description to the title's copy;
adv_name replaced the title's extra list, so a title built from siblings (a
plugin's {"text":"","extra":[...]}) reached clients as just the description.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The screen built them from literals, so the buttons stayed English in every
language while the world select screen next to them was translated. Each has
a vanilla key the prepared resources already carry; Friends is Horizon's own
and reads horizon.menu.friends with the English text as fallback.
Main's SRV resolver (#1578) added Connection::connect_server; both connect paths now
go through connect_addresses, which puts IPv4 first, as ServerAddressResolver.SYSTEM's
InetAddress.getByName does for the original host and the SRV target alike.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A proxy switching servers (Velocity, BungeeCord) sends the client back to
configuration. The network thread acknowledged it, but the game thread kept
ticking the old level and its client_tick_end went out right after the
acknowledgement, where the server reads it as configuration packet 13.
Velocity drops the client: "A packet did not decode successfully", shown as
"An internal error occurred in your connection".
Vanilla's handleConfigurationStart clears the level before it acknowledges,
so nothing of play follows. Session::send now drops play packets while the
connection is in configuration, checked under the writer lock that the
acknowledgement is sent with.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Mesher::mesh keeps model order, which the leaf-culling test pairs with
the baked faces; the section and tile workers sort what they upload.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
load_post takes the game's terrain vertex inputs and, given them, loads
the first program of gbuffers_terrain's fallback chain that translates
(gbuffers_terrain, textured_lit, textured, basic). Program loading is
shared by the post-processing and gbuffers programs.
The translator no longer rewrites a variable, parameter or member that
has the name of a sampler (`float specular` beside `uniform sampler2D
specular`), which kept Complementary's gbuffers programs from compiling.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Two halves of FireworkRocketEntity on a server were missing:
- SetCreativeModeSlot encoded only damage, so a creative-picked rocket
lost its minecraft:fireworks component (flight_duration and explosions
gone); the wire codec also had no decoder for it or firework_explosion,
so a stack carrying them failed to decode at all. The components now
follow Fireworks.STREAM_CODEC/FireworkExplosion.STREAM_CODEC both ways
(component ids 71/70).
- A remote rocket never read DATA_ATTACHED_TO_TARGET,
DATA_ID_FIREWORKS_ITEM or DATA_SHOT_AT_ANGLE, so the server-attached
boost rocket was an inert marker: attached_to now comes from
EntityDataSerializers.OPTIONAL_UNSIGNED_INT (the id plus one), the item
feeds the explosion list, and FireworkRocketEntity.tick's
isFallFlying boost applies to the local player.
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Packs sample blocks from one 2D atlas; Horizon keeps them in sprite
arrays. While a pack is loaded, every layer and its mip levels are
copied into a grid, and the animator reports the layers it rewrote so
only those are copied again. The sprite arrays gain COPY_SRC.
The gbuffers pass binds it in the next slice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
translate_gbuffers places a vertex-input glue the game writes (it knows
its own vertex formats) in front of a pack's gbuffers program, sets the
compatibility matrices from gbufferModelView/gbufferProjection and the
lightmap texture matrix, and gives declared attributes the game's values
(zero when it has none).
Nothing calls it yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Give integrated and dedicated startup a synchronous loading entry point without client deferral or integrated-server selection. Keep client orchestration and background loading unchanged.
Co-Authored-By: OpenAI Codex GPT-6.1 <noreply@openai.com>
Separate client, integrated-server and dedicated-server roles from connection state. Use roles for startup, presentation work and server dimension loading while preserving in-frame client simulation.
Co-Authored-By: OpenAI Codex (GPT-6.1) <noreply@openai.com>
A mounted mob answered "Found no elements matching Passengers": the command
built the entity with the flat serializer, so `Passengers` was never in the
tag, even though the game had the relationship and saving a world kept it.
Persistence already writes the tree (`Game::mob_tree_nbt`, which also carries
the leash), and vanilla's command path uses the same one, so the command reads
it now.
Fixes#1519.
Co-Authored-By: Hermes Agent (Nous Research) <noreply@nousresearch.com>
- A buffer is cleared only when a program reads it before one writes all
of it; the clears left ride in the blit pass as attachments the blit
shader leaves alone, instead of passes of their own.
- The blit writes depthtex0 only when some program reads it.
- Attachments a pass overwrites whole (the blit's colortex0 and depth,
every composite output, final's window) start cleared rather than
loaded, which spares a tiled GPU reading them back first.
- The blit's bindings are kept until the world target changes size.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
boss_event_round_trips missed its #[test] attribute, failing clippy's
dead-code lint; unknown_packets_keep_their_payload decoded boss_event,
which this PR gives a real decoder, so the garbage payload no longer
comes back Unknown — it uses player_look_at, which still has none.
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>