Vanilla teleports a respawning player once; the second, identical move
showed up in the dedicated_server, pearl_death and cursor_death harnesses.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The host runs a guest's trident release on its copy of the guest, which
sets the spin for 20 ticks. The copy never ran the client's movement
tick, so the count never went down: the copy stayed in the spin pose
(eye 0.4), and every arrow or projectile the guest shot was spawned from
the leg until something else reset it.
LivingEntity.aiStep counts autoSpinAttackTicks down on the server too.
Count it down in the guest's server tick, before updatePlayerPose.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Orchestrator's review of #2859: a nightly can hold pull requests nobody
reviewed yet, and their build scripts run in `build`. game-build.yml gets a
`trusted` input, true only from release.yml:
- the Rust cache (GitHub-hosted Linux and Windows): its own key for an
untrusted build (`nightly-<name>`), never saved; a release restores only
what a release saved;
- macOS: an untrusted build's exporters and build jobs run on the runner
that already runs pull requests' parity jobs (macOS + horizon-game), never
the release runner (MAC_RUNNER) nor parity-main's;
- a trusted build doesn't read the machine's shared sccache, which pull
requests' jobs write (runner-env.sh: HORIZON_NO_SCCACHE).
tests/test_nightly.py pins all three.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
nightly-2: a Nightly can include pull requests not reviewed yet, picked by
trusted reviewers (docs/release-flow.md "Nightly channel"). The Settings row
says so, in English, French and Russian.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Single player joins its integrated server as a client, so `Minecraft.tick`
takes the `tick_remote` path and never reaches `tick_mobs`, where the
subtitle overlay's tick had its only call. The overlay therefore never
started listening to the sound engine: no sound notified it, and the
closed captions stayed empty with "Show Subtitles" on or off.
Call `tick_subtitles` from `tick_remote_level` beside `tick_hud`, as the
local path does.
Co-Authored-By: ZCode (Z.ai) <noreply@z.ai>
Step A of the horizon-client crate split. `Chunk.block_entities` held
`containers::BlockEntity`, whose payloads were defined across containers,
brewing, signs, sculk, mobs::spawner, mobs::trial_spawner, vault, beehive
and command_block, so `chunk` reached most of the gameplay code.
The enum and every payload type now live in `chunk/block_entity.rs` as
plain data (definitions and derives). Every impl block stays in its
feature module, and each module re-exports its types under the old path,
so callers don't change. Also:
- `mobs::Who` moves to `horizon_client_core::who` (the vibration `Info`
names it); mobs re-exports it.
- `skins::NameOrId` moves beside `SkullSkin`, which holds it; skins
re-exports it.
- `Spawner.next`, the `Next` enum and `Selector.current` become
`pub(crate)`, since their struct now lives in another module.
- The vault, spawner and trial_spawner harnesses record
`chunk/block_entity.rs`.
No behaviour change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Ludovic's decision (2026-10-27, from kacpherek's proposal under #2779): the
Nightly is where contributors test pull requests before the gate.
kacpherek, rdbtCVS and levzzz5154 (.github/nightly-reviewers.txt) and
Ludovic pick them with `/nightly 2640 2642`, `add`, `remove`, `clear`,
`status` or `dry-run` in any issue or pull request.
- nightly-command.yml (issue_comment, main's copy): checks who asked, saves
the set on nightly-state, starts `nightly build`, answers "🤖 Nightly: …".
It never checks out or merges a pull request.
- nightly-build.yml: `pick` builds main + the set (newest heads and merged or
closed ones dropped at the scheduled run), pushes it as
nightly/build-<date>-<n>, records it in builds.jsonl; build, sign and
publish are unchanged; `answer` tells the thread how it ended.
- Notes say a Nightly can include pull requests not reviewed yet, picked by
trusted reviewers, and list each with its author and who picked it.
- docs/release-flow.md: who can pick, how, how Ludovic removes or pauses them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The pieces of nightly-2 (`/nightly 2640 2642`): parsing a comment, the
reviewers list (.github/nightly-reviewers.txt, with a `paused` line), the
refusals (paths CODEOWNERS gives to Ludovic, closed or merged PRs, more than
30, more than 6 builds a day), the set and its commands (Ludovic's remove
keeps reviewers from adding a PR back), the nightly-state branch (pushed with
retries when another run saved first), and the assembly: each PR head merged
onto main in order with `git merge-tree --write-tree`, objects only, so no
file, hook or filter of a pull request reaches the runner. Equal inputs give
the same commit. A conflict names the two pull requests (or main) and the
paths.
Nothing calls it yet: the next PR wires it into nightly-build.yml and a
`nightly command` workflow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Read against the vanilla 26.3 sources (Fire the enchanted_item advancement trigger at the enchanting table) and proven in contrib-gate-21's union of 72 kacpherek PRs: 95 parity harnesses on playhorizon and the Mac mini, clippy, 1534 tests, lan_e2e 23 checks in-frame and with --server-thread; main merged in since, clippy clean. Thanks @kacpherek!
🤖 Agent: Read against the vanilla 26.3 sources (An axe turns an unoxidized copper golem statue back into a golem) and proven in contrib-gate-21's union of 72 kacpherek PRs: 95 parity harnesses on playhorizon and the Mac mini, clippy, 1534 tests, lan_e2e 23 checks in-frame and with --server-thread; main merged in since, clippy clean. Thanks @kacpherek!
playhorizon's runner fails remote_fireball and remote_fireworks even alone and
at bare main (3b19c386c: 2 of 2 runs and the retry), while the Mac mini passes
them; every fork PR whose parity lands there stopped on them. The host's
exclude list only kept harnesses off workers sent there, not off the runner's
own jobs: platform_skips now adds the local host's exclude, skipped with its
reason unless named. The two harnesses join playhorizon's list; the Mac
mini's runner and the nightly audit (macOS) keep running them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Read against the vanilla 26.3 sources (Block ambient effects: levers, repeaters, dripstone, wither roses, End gateways, lightning rods) and proven in contrib-gate-21's union of 72 kacpherek PRs: 95 parity harnesses on playhorizon and the Mac mini, clippy, 1534 tests, lan_e2e 23 checks in-frame and with --server-thread. Thanks @kacpherek!
🤖 Agent: Merged main in (conflicts resolved, see the merge commit); proven in union R1b (branch union/cg19-r1b): clippy -D warnings, tests, the harnesses it can affect, lan_e2e in-frame and on the server thread.
🤖 Agent: Merged main in (conflicts resolved, see the merge commit); proven in union R1b (branch union/cg19-r1b): clippy -D warnings, tests, the harnesses it can affect, lan_e2e in-frame and on the server thread.
Upstream maps U+043E (о) and U+043F (п) and not their capitals, so a Russian
heading drew О and П from the fallback font: they came out smooth and taller
than the pixel letters around them (screenshots in the PR).
Both come from the font's own drawing. О is the Latin O's glyph under the
Cyrillic codepoint, which is the same round shape this font gives о: its
lowercase is the O with the top shifted down. П is that weight's lowercase п
with everything above the baseline lifted to the cap height the font uses there
(Н's top), 175 units in SemiBold, 177 in Medium and 171 in Bold, since each
weight draws its own metrics. Both now sit on the bbox of the O and Н of their
own weight.
О keeps the O's hinting instructions; П is a new outline with none, which a
pixel design does not need at the sizes the launcher draws it. GDEF, GPOS and
GSUB are byte-identical to the files they replace.
Co-Authored-By: Hermes Agent (Nous Research) <noreply@nousresearch.com>
🤖 Agent: Merged main in (conflicts resolved, see the merge commit); proven in union R1b (branch union/cg19-r1b): clippy -D warnings, tests, the harnesses it can affect, lan_e2e in-frame and on the server thread.
🤖 Agent: Proven in union R1 (branch union/cg19-r1, main + 83 of kacpherek's PRs): the 98 harnesses it can affect, clippy -D warnings, tests, lan_e2e in-frame and on the server thread.
🤖 Agent: Proven in union R1 (branch union/cg19-r1, main + 83 of kacpherek's PRs): the 98 harnesses it can affect, clippy -D warnings, tests, lan_e2e in-frame and on the server thread.
On a remote (non-Horizon) server the use-item tick spawned the crumb
particles but never played the consumable's eat sound, so only the burp
at the end was heard. Call emit_consume for the periodic emit, as
tick_used_item does on the player's own server, so the bites sound too.
Fixes#2747.
Co-Authored-By: Devin (SWE-2 High) <158243242+devin-ai-integration[bot]@users.noreply.github.com>