🤖 Agent: verified by Ludovic's AI agent (contrib gate 20): USE_ITEM_COMPLETE finishes the use; eating on the client (lan_e2e eat_full, eating, helduse, consume-release pass). It passed in this wave's union with main on the Mac mini: the selected parity harnesses (332/332, then 122/122 on the final union), clippy -D warnings and the tests, and lan_e2e with the host on its server thread and in-frame (18/18 each). Thanks @Louistuis!
🤖 Agent: verified by Ludovic's AI agent (contrib gate 20): cushion break particles reach multiplayer players. It passed in this wave's union with main on the Mac mini: the selected parity harnesses (332/332, then 122/122 on the final union), clippy -D warnings and the tests, and lan_e2e with the host on its server thread and in-frame (18/18 each). Thanks @ESHAYAT102!
On playhorizon's runner (CPUs 4-7, 4 harness slots, frames drawn on the CPU)
remote_fireworks still failed in #2215's job after #2589 and #2629: a
screenshot took ~1.5 s (58 ticks for two) beside three other harnesses. The
harnesses in parity_hosts.json's new solo list run alone there, after the
others; bigger machines are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
playhorizon's CI runner has CPUs 4-7, and remote.sh pinned agents' runs there
too: CPUs 4-7 sat at 100% while 0-3 idled, and window harnesses in pull
requests' parity jobs lost ticks. Runs share a lock per CPU set; an
--exclusive run (a benchmark) holds it alone. world_reopen also waits for
the world to load before its first count, as after each reopen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On playhorizon (frames drawn on the CPU, control requests answered slowly)
world_reopen read the minecarts before the mineshaft's last one arrived (3
then 4 on every reopen), and the sculk walk polled the sensor with separate
requests that missed its 30-tick active window. The minecart list is read
once it stops changing; both clients sample the sensor every tick from one
ctl script, alongside the walk.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: the server sends entity event 4 once its warmup ends (`EvokerFangs.tick`), `handleEntityEvent(4)` starts the client's attack and plays the sound locally, and the client only counts `lifeTicks` down for the animation, while the server deals the damage and discards. Moving the host's sound to `play_unshared` matches that, since each client plays its own. It conflicted with #2094 (both hook entity events) once that merged, so I merged main into it, keeping both calls. `lan_e2e.py --check fangs` passes in both modes. It conflicted again with main in `tools/lan_e2e.py` (new checks on both sides), so I merged main in once more, keeping every check; clippy is clean. Fixes#1482. Thanks @rdbtCVS! Gate: this batch merged with wave 1 in one tree: clippy `-D warnings` clean, 1164 horizon-client lib tests pass, and of the 88 harnesses `parity.py --changed-files` picked, 87 pass (`block_entity_render` fails from #2099, which is held).
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `EnderDragonFight.tick` scans the state and places the exit portal only when `dragonEvent.getPlayers()` is not empty, so a force-loaded End with nobody in it gets no portal, as here; the new `forced_end` harness compares both servers. Gate: this wave's 50 PRs merged on main in one tree (union 9033a9c3b): fmt and clippy `-D warnings` clean, the horizon-client (1136 lib plus the integration tests), horizon-client-core (37), horizon-world (165) and horizon-protocol (100) tests pass, the fast build links, and the 140 harnesses `parity.py` picks for the tree's changed files pass on playhorizon (remote_fireworks, a known load flake there, passed on a rerun on a quiet Mac mini). `lan_e2e.py` passes 14 checks with the host on its server thread and in-frame (chest, xp, damage, kills, projectiles, mobs, drops, body, stopsound, digging, instant_digging, books, knockback, barter); `hurt` fails on main too when run after `damage` (check-order dependency, not this wave). Thanks @FusselTV!
🤖 Agent: fixes parity-main's friends_integration (red since 4be5c8eb5). Cause: #2501 spawns new players at a random spot near the spawn, as vanilla does, and test_friends.py assumed the guest stood on the spawn itself. Main failed 4/4 on the Mac mini, main with #2501 reverted passed 2/2, and this branch passed 4/4 on the server thread and 1/1 in-frame. Python only; harness-only, skip-release.
🤖 Agent: merging the first crate split (horizon-command + horizon-client-core, pure move, re-exported under the old paths). CI parity passed at 17f8419ac (head merged into main), verify and launcher pass, approved by rdbt's review agent; landed by hand while the REST pool refills.
Since #2501 a player new to the world spawns at a random free spot near the
world spawn, as vanilla's PlayerSpawnFinder does. The guest often stood against
a hill, so walking forward moved it less than a block, or terrain hid the block
it had to aim at: friends_integration failed on every run on the Mac mini.
The host now clears a floor beside itself and teleports the guest there before
the movement checks, and the block check waits for the guest to stand still.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A manifest line defining or turning on a workspace crate's feature no longer
makes the manifests count, if every source naming the feature (or one it turns
on) changed too, as in a split, where the new crate's files are all new.
#2215 (horizon-items with a test-support feature) now selects 209 harnesses
instead of 354.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On playhorizon (frames drawn on the CPU) a screenshot takes ~0.6 s: with a
16-tick wait the later shots came after the sparks' lifetime, and the setup's
own red chat lines, gone by then, made the host's count negative.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On a server client (single player's server thread, LAN guests, dedicated
servers) the window ended a consumable's use on its own timer and never ate
the food: its FoodData waited for the server's next ClientboundSetHealth.
With the button still held, the next startUseItem found the food bar below
20 and began another bite, so the eating animation went on after the bar
was full (#2471).
Vanilla's client never completes a use by itself: LivingEntity.updateUsingItem
completes it only on the server, and ServerPlayer.completeUsingItem first
sends the player entity event 9 (USE_ITEM_COMPLETE). Player.handleEntityEvent(9)
then runs completeUsingItem on the client, where Consumable.onConsume calls
the FoodProperties listener on both sides: the client's FoodData eats at once.
- The server sends event 9 to the acting guest when a consumable, a potion or
a spyglass use completes. Like the guest's packets, the events are queued
during their turn (kept across nested turns) and sent with their entity id.
- The client keeps using until that event, then shows the last crumbs, eats
the food (the next ClientboundSetHealth replaces it) and starts the item's
use_cooldown, which ItemStack.use only starts at once for an item with no
use time.
- What vanilla's client checks so that a use can't outlive the server's:
MultiPlayerGameMode.useItem passes on an item cooling down (chorus fruit
held through its cooldown), LivingEntity.updatingUsingItem stops when the
hand no longer holds the item the use began with (ItemStack.isSameItem),
a respawn is a new player with no use, and Consumable.startConsuming
starts none for a consumable with no use time.
- ItemCooldowns tick where Player.tick ticks them: after LivingEntity.tick's
use on the server and after Minecraft.handleKeybinds on the client. Both
sides ticked them first, so a guest's entry ended a tick later on the
server than on their client, whose held retry was refused and never ended.
lan_e2e's new eat_full check eats melon slices up to a full bar with the
button still held: before, a new bite had started (use ticks 26, 2, 9, 15);
now nothing is in use. Held on two chorus fruits, both are eaten across the
cooldown. consume-release holds its completed use a little longer, as the
use now ends with the server's event.
Fixes#2471
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Cargo.lock is a shared input and every harness depends on its crates'
Cargo.toml, so a crate split (#2160) selected all 352 harnesses although
its moved code reaches 123. With --changed, Cargo.lock and Cargo.toml
changes that only add, remove or move workspace crates (no registry or git
package, version, feature, dependency or build setting changed across them,
compared with the merge base) are left out; the moved code counts through
its own sources. Any other manifest change still counts, as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Under parity.py the release and fast profiles have identical settings
(PROFILE: no LTO, 16 codegen units, incremental), yet 38 harnesses built
horizon-client in release and 16 others in fast: every run where the client
changed compiled it twice. They all use fast now. Release stays for
horizon-world's probes, which no fast harness builds.
Measured on playhorizon (4 CPUs), all 13 probe builds after 78 merges:
1578 s, of which the release client (bin + client_probe) took 485 s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: LAN guests (host on its server thread) and a dedicated server's new players joined inside each other on the exact spawn, so they clicked the other player instead of a table or chest. New players now get PlayerSpawnFinder.findSpawn as in vanilla. lan_e2e spawned/crafting/chest pass in both host modes, a dedicated two-guest check passes, and the 32 affected harnesses pass.
The command ran on after remote.sh was stopped or its connection dropped,
holding the name's lock, and a game or daemon it left behind outlived the
run. It now runs in a scope named after the run (remote-<name>.scope),
stopped when the command ends, when the ssh session ends first, and before
the next run under that name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The harness sampled the fireball from a new horizon-ctl process each time,
10-16 ticks apart on a loaded runner: the fireball (~2 blocks a tick) left
the cleared corridor or the guest's tracking range before the sixth sample,
and the harness failed. It now flies slowly (acceleration_power 0.02, under
0.4 blocks a tick) down a longer corridor, and the guest takes 16 samples
in one ctl script (lan_e2e's Game.script returns every reply with every=True).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The parity client's chat, villager and furnace steps went on 1.5 s after
sending, so a busy server's late reply landed in the next step (or a click
carried a stale state id). Each step now waits until the server answered:
its chat line, both slot packets of a shift-click, and the furnace totals
the click changes (up to the same 30 s patience). The respawn window keeps
only the player's own entity packets: a squid's swim event or a mob coming
into view falls into it as each server's ticks fall.
Off the known-flaky list: it passes under 2.5 s server stalls every 5 s,
where main's client fails.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The host fetches the commit from GitHub (nothing else is sent), keeps one
checkout and target per worktree, runs the command niced in its CI slice on
CPUs 4-7 with scratch homes, prepares vanilla data for harness commands as
CI does, and streams the output back. One run per name at a time; no lock
leaks into daemons the run starts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The server put every player new to the world exactly on the world spawn.
On the server thread the host's own player joins that way too, so the host
and each new guest stood inside each other: the guest's client targets
the host's player (its eye inside that box, as vanilla's getEntityHitResult
does), so it could open no crafting table, chest or furnace, and Entity.push
never parts two players with the same centre. A dedicated server stacked
its new players the same way.
Vanilla places each new player with ServerPlayer.adjustSpawnLocation:
PlayerSpawnFinder.findSpawn, a free column at a random offset within the
respawn radius. Game::server_player_spawn does it for a server's new
players with the window's own finder, passing over the candidates whose
chunk isn't loaded yet (vanilla waits for them; a server tick can't).
lan_e2e's new check `spawned` asserts that the host and the guest join on
different spots, then has the guest open a crafting table, a chest, a
furnace and a stonecutter where it joined.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Nine vanilla Java servers from harness runs had been orphaned on the Mac mini for 2 to 5 days. Each harness runs in
its own session (killpg on a timeout), so when a CI job was cancelled, its harnesses and their servers outlived
parity.py.
- parity.py: SIGTERM and SIGHUP exit through the finally blocks, which now stop the local executor's harnesses
too, not only the remote hosts'. The worker (--worker) stops its harnesses the same way.
- runner-gc.sh (every 6 h from launchd): stops Java processes whose parent is gone (launchd, init or
systemd --user) and whose command line names a runner's checkout or a worktree's .local/, once ORPHAN_HOURS
(default 3) old.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
pr_ready --quick passed #1895 while CI's clippy failed on an unused import in horizon-mods' tests: clippy on the
touched crates and their dependents gives them other features than CI's workspace-wide run (Cargo unifies
features over the packages it builds).
- clippy: CI's two commands as they are. The whole workspace runs with the game's code mods, and the server build
(horizon-client --no-default-features) runs when horizon-client is affected. Warm on the Mac mini, a change to
hud.rs took 14-16 s against 17 s before, and food.rs 22-32 s against 14 s. Each worktree's first run costs
about 3 min once.
- tests stay on the touched crates, with the features CI's workspace run gives them: horizon-client's code mods,
and horizon-mods' runtime, pack and index.
- The Python tests always run, as in CI: some read the crates, the roadmap or the workflows.
- The launcher steps run when CI's launcher job would (tools/ci/rust_changes.py's rule, which also covers the
root manifests and the check itself).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
tools/parity_hosts.json's new `known_flaky` lists the harnesses that still flake for a cause not fixed yet,
each with its reason: dedicated_server (its Java client's 1.5 s step timer), remote_fireball (wall-clock
sampling) and world_reopen. One of them that fails runs once more from scratch, and a pass then counts.
Every retry prints a RETRIED line at the end of the run and in the CI job's summary, so a harness that
passes only on its retry stays visible on PR parity and parity-main alike. The existing fresh rerun of a
failure that reused oracle results is reported the same way. A known flaky harness skips that rerun and keeps
reusing oracle results, because flaking doesn't mean its cached oracle runs were wrong. Repeated runs
(--repeat, --flake-check) are never retried.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With force-loaded levels ticking without players (#1525), Horizon placed
the exit portal as soon as the End's main island was loaded. Vanilla
only places it from scanState and the dragon's death, both with players
in the fight, so a pre-generated End has no portal until someone
arrives.
test_forced_end_parity force-loads the End's arena on vanilla's and
Horizon's dedicated servers with nobody online and compares what is
around 0, 0.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Brings #2316 (the campfire block entity fix): CI's parity checks out this
branch's head, so dispenser and block_removal need it here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `ServerLevel.tick` keeps ticking a level that has forced chunks, even with no players and past the 300-tick empty timeout. `lan_e2e.py --server-thread --check forced-levels` passes. Main conflicted in lan_e2e.py, so I merged main into the branch, keeping both checks. Gate: this wave's 55 PRs merged on main in one tree: clippy `-D warnings` clean, 1186 horizon-client and 163 horizon-world tests pass, and of the 101 harnesses `parity.py --changed-files` picked, 100 pass (`block_removal` failed on one spawner case from #2199's orb merging, fixed there). `lan_e2e.py` passes 24 checks with the host on its server thread and in-frame (chest, xp, pvp, doors, villagers, beds, frames, projectiles, arrows, fishing, damage, mobs, blast, drops, elytra, breath, predict, cracks, barter, kills, hurt, contacts, underwater, plates), plus forced-levels on the server thread. Thanks @rdbtCVS!
Brings #2212 (namespaced furnace RecipesUsed), which save_roundtrip needs:
CI's parity checks out this branch's head, not a merge with main.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Full trial (run 40341431094): 342/348 in 96.9 min; save_roundtrip,
remote_fireball and server_player_sculk passed on a rerun with vscreen and
fewer jobs; these three did not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>