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>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Item.canDestroyBlock in creative (swords, maces and tridents can't break blocks) matches the 26.3 bytecode, on the player's prediction and on the server. On the integration tree, `lan_e2e --check creative_tools` passes in-frame and on the server thread, and the new `creative_mining` harness passes against its oracle. I merged main into the branch twice; the second time, lan_e2e.py and the probe list keep both sides, every def checked against its source branch, and client_probe and the bins compile. Fixes#2035 is right. Thanks @ESHAYAT102!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `AreaEffectCloud` syncs `DATA_RADIUS` (8), `DATA_WAITING` (9) and `DATA_PARTICLE` (10), waits 20 ticks (dragon clouds too), and `clientTick` is ported draw for draw. `DragonFireball`'s trail is `dragon_breath`. One fix pushed: `AbstractHurtingProjectile.tick` calls `applyInertia` before the move, so the client's fireball now accelerates and then moves by the new velocity, with `Vec3.normalize`'s zero for a tiny vector. Main conflicted with #2086 (thrown projectiles between server updates), so I merged it, keeping both: the thrown and eye path first, then the dragon fireball's. `lan_e2e.py --check particles` and `objects` pass in both modes. Part of #2083, as it says. Thanks @kacpherek! 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). Chunks now write the structure starts and references they hold, so vanilla keeps the structures (and their spawns) after the round trip. Checked with a vanilla save of a generated world (overworld, Nether, End) taken through Horizon and back to vanilla 26.3 on the server thread and in-frame. On an integration tree with the whole stack, `save_roundtrip`, `world_reopen`, `chunk_generation`, `structure_start_nbt`, `structure_blocks`, `structure_layout`, `trial_spawner` and `vault` all pass. Thanks @FusselTV!
🤖 Agent: merging. playhorizon.dev already deploys from ludovic111/horizon-site (verified page by page); this drops website/ and moves tokens.css, roadmap.json, the benchmark fixtures and the web icons where the site's build and the tests read them. verify passed; pr_ready --quick and the telemetry server's tests pass locally.
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `BaseFireBlock.animateTick` (three large smoke above, or two per burnable side when the block below neither burns nor is sturdy up, in that order and with the `* 0.1F` float edge), `SoulFireBlock.canBurn` always true, and `Level.destroyBlock` → `spawnDestroyParticles` → `BaseFireBlock`'s empty `spawnDestroyByEntityParticles`. `fire.igniteOdds` is in block-behaviours.json at the top level, as read. I added one rule: `FireBlock.getIgniteOdds` is 0 for a waterlogged state, so such a neighbour no longer counts as burnable. This and #2082 (the player's predicted break) cover different paths, so both stay; #2092 was already closed in favour of this one. In game, fire and soul fire show their large smoke above them. Thanks @socksprox, and @kacpherek for #2092! 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: `NetherPortalBlock.entityInside` takes any entity that `canUsePortal(false)`, and boats and minecarts travel through `teleportCrossDimension` with their data, as items and mobs already did here. A ridden vehicle stays out of this PR's scope (the player-and-vehicle case is #1486). `lan_e2e.py --check vehicles`, `boats` and `minecarts` pass in both modes. Fixes#1516. Thanks @kacpherek! 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: `ItemFrameRenderer.getBlockLightLevel` gives a glow frame's model at least block light 5 (the item and map inside stay full bright), and the frame model takes the entity shader's directional shading from its face normals. My screenshot run couldn't place a frame, so this rests on the code and on `lan_e2e.py --check frames`, which passes in both modes. Fixes#1436. Thanks @kacpherek! 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: `EndPortalBlock.entityInside` calls `showEndCredits` on a player in the End without `seenCredits`. That sends `WIN_GAME` with 0.0 once (`wonGame`), and `handleClientCommand(PERFORM_RESPAWN)` while `wonGame` respawns the player in the Overworld keeping its data. `seenCredits` round-trips through playerdata. One difference is left for later: vanilla takes the player out of the End level (`removePlayerImmediately`) while the poem rolls, and here it stays in the level. `lan_e2e.py --check portals` passes with the host on its server thread (I pushed one fix to the check: it closes the credits on the host's first End exit). Fixes#1507. Thanks @kacpherek! 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). The remaining legacy structure pieces now save their vanilla fields. Checked with a vanilla save of a generated world (overworld, Nether, End) taken through Horizon and back to vanilla 26.3 on the server thread and in-frame. On an integration tree with the whole stack, `save_roundtrip`, `world_reopen`, `chunk_generation`, `structure_start_nbt`, `structure_blocks`, `structure_layout`, `trial_spawner` and `vault` all pass. Thanks @FusselTV!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Template pieces now save as vanilla's TemplateStructurePiece does. Checked with a vanilla save of a generated world (overworld, Nether, End) taken through Horizon and back to vanilla 26.3 on the server thread and in-frame. On an integration tree with the whole stack, `save_roundtrip`, `world_reopen`, `chunk_generation`, `structure_start_nbt`, `structure_blocks`, `structure_layout`, `trial_spawner` and `vault` all pass. Thanks @FusselTV!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `ItemCost.STREAM_CODEC` is item, count, then `DataComponentExactPredicate.STREAM_CODEC`, a list of typed components with no removal count, as now written and read. Its verify failed only on rustfmt (item_sync.rs), which I pushed. Main conflicted in item_sync.rs's tests (the new bundle test), so I merged it, keeping both tests. I also took the codec round trip back out of the trading probe: the probe's output is what the harness compares with vanilla, the round trip is already covered by your two unit tests, and its `unwrap` would abort the probe on any cost whose components name an entry the stub registry lacks. `lan_e2e.py --check hero-trading` passes in both modes. Fixes#1948. 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).
playhorizon.dev now deploys from the private ludovic111/horizon-site, which holds website/'s history. The files
its build still reads from here move out of website/:
- the design tokens to crates/horizon-theme/tokens.css (xtask tokens writes them, tokens_css_is_up_to_date checks them);
- the roadmap's JSON to roadmap/roadmap.json (tools/roadmap.py build);
- the benchmark fixtures to crates/horizon-telemetry-server/tests/fixtures/benchmarks/ (the site keeps a copy);
- the web icons to docs/media/brand/web/ (xtask icons).
pr_ready loses its website area and the tokens special case; rust_changes, the roadmap's links and every doc
that pointed into website/ follow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `Guardian.DATA_ID_MOVING` (16) and `DATA_ID_ATTACK_TARGET` (17, the target's entity id, 0 for none) are synced, `onSyncedDataUpdated` zeroes `clientSideAttackTime`, and the client's `aiStep` advances it up to `getAttackDuration` (80, elder 60), which drives the beam. The in-frame tick now shares `client_tick`. Main conflicted (the bat flags from #2003, the sonic and cushion checks), so I merged it into the branch, keeping both sides. `lan_e2e.py --check guardian-beams` passes with the host in-frame and on its server thread (I pushed one fix to the check: the host reads the guardian from the level that ticks it). Fixes#1491. 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). Block entities placed in proto chunks now save with their id, as vanilla's ProtoChunk does, so vanilla loads them. Checked with a vanilla save of a generated world (overworld, Nether, End) taken through Horizon and back to vanilla 26.3 on the server thread and in-frame. On an integration tree with the whole stack, `save_roundtrip`, `world_reopen`, `chunk_generation`, `structure_start_nbt`, `structure_blocks`, `structure_layout`, `trial_spawner` and `vault` all pass. Thanks @FusselTV!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Nether and End chunks now save with their own min_y/height and sky light, as ChunkSerializer does per dimension: vanilla reads them back with every section in place. Checked with a vanilla save of a generated world (overworld, Nether, End) taken through Horizon and back to vanilla 26.3 on the server thread and in-frame. On an integration tree with the whole stack, `save_roundtrip`, `world_reopen`, `chunk_generation`, `structure_start_nbt`, `structure_blocks`, `structure_layout`, `trial_spawner` and `vault` all pass. Thanks @FusselTV!
🤖 Agent: Merged by Ludovic's AI agent (Claude). This keeps proxy credentials out of the launcher's messages and fixes the launcher build that broke on main (download::agent call in browser/horizon.rs). Launcher lib tests 137/137, horizon-port-pack 68/68. Thanks again to @zhykowo for the proxy support in #2059.
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `ContainerEntity.isChestVehicleStillValid` is `!isRemoved() && player.isWithinEntityInteractionRange(getBoundingBox(), 4.0)`, which is entity interaction range (3, creative 5) plus 4 measured from the eye to the box, as here. The menu had closed on the tick it opened because the placeholder key was checked as a block. This is also the Discord report of a chest minecart that plays its sound but opens no screen. `lan_e2e.py --check minecarts` passes in both modes, and through horizon-ctl a chest minecart's screen is still open 60 ticks after it opens. Thanks @kacpherek! 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). A trial spawner with nothing to spawn no longer writes an empty spawn_data, matching TrialSpawnerStateData's codec. Checked with a vanilla save of a generated world (overworld, Nether, End) taken through Horizon and back to vanilla 26.3 on the server thread and in-frame. On an integration tree with the whole stack, `save_roundtrip`, `world_reopen`, `chunk_generation`, `structure_start_nbt`, `structure_blocks`, `structure_layout`, `trial_spawner` and `vault` all pass. Thanks @FusselTV!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). The item_name component now sits on items and tooltips as vanilla's Item.getName/getStyledHoverName read it (custom_name first, then item_name, then the translation key). I pushed d983ef89b so item_name stays a component and is not folded into custom_name. `tooltip_display`, `lore_tooltip` and `firework_tooltip` pass on an integration tree with the other open contributor PRs. Fixes#2039 is right. Thanks @lazy-importer!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Campfire cooking (CampfireBlockEntity.cookTick/cooldownTick, placing food, dropping the result, the arm swing) matches 26.3, and food cooks in game on the server thread and in-frame. Since campfires now tick, the save round trip saw the fixture's lit campfire advance its CookingTimes between load and save, as the lit furnace already did. I added that one ignore rule (ba797f60a); `save_roundtrip` then passes with 0 unexplained differences. Thanks @vlesmentie!
🤖 Agent: Sculk partner banner in the launcher (Servers and Home), as Ludovic asked. Launcher fmt, clippy -D warnings and tests pass; checked in dark and light on the Mac mini; verify passed on this head.
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `FireworkRocketEntity.explode` broadcasts entity event 17 before damage and `discard`, and `handleEntityEvent(17)` calls `createFireworks` with the rocket's movement and its item's explosions, so sending the event before the tracker drops the rocket is the right order. Main conflicted (fireworks.rs, lan_e2e.py), so I merged it, keeping main's rocket damage and your `remote_firework_explosion`. `lan_e2e.py --check fireworks` passes in both modes. Fixes#1478. 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).
An address the launcher can't use as a proxy is printed to stderr by
horizon-port-pack and shown under the Settings field or by `settings set`:
a user name and password in it now read `***`.
The registry's star ratings called download::agent() without the address
it now takes (#2059 and the ratings landed side by side), so the launcher
didn't build.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). Checked against 26.3: `AbstractRecipeBookScreen.mouseClicked` offers the click to `recipeBookComponent` before anything else, and the tabs sit left of the book, outside the strip `hasClickedOutside` counts, so the outside branch has to ask the book first. `lan_e2e.py --check recipes` passes in both modes. Fixes#2091. Thanks @kacpherek! 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). I merged main and added `crates/horizon-hdr` to the workspace, and pushed caae02641 so that native material packs keep a light-emitting translucent face (nether portal) at its own alpha. HDR stays off by default. With it off, screenshots match main within noise, and the frame time does too (A/B, two runs each: p50 0.326 to 0.329 ms vs 0.317 to 0.345 ms). `section_mesh` passes. HDR on was not tested here (no HDR display on the Mac mini). Thanks @senjinthedragon!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). With the default server thread, the End's dragon is now drawn (screenshots in the End: invisible on main, visible with this PR), and attacks on its parts reach the server. `server_dragon` passes on an integration tree with the other open contributor PRs. Fixes#1530 is right. Thanks @vlesmentie!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). rdbt's approval counts as the first review (trusted reviewer). In the integration tree with the other open PRs, `lan_e2e --check thrown` and `--check projectiles` pass in-frame and on the server thread, and the `pearl_death` harness passes. Fixes#1826 is right. Thanks @kacpherek!
Campfires now tick, so the fixture's campfire advances its CookingTimes
between load and save, as the lit furnace does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sculk, a server host, sponsors Horizon. Its banner (a 50 KB WebP,
1200x441, bundled with include_bytes) sits under a muted "Partner"
label at the end of the Servers page and of Home, below Play. A click
opens https://my.sculkhosting.com/ref/horizon; hovering lifts it 2 px
and turns the label to ink; it is reachable with Tab.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). This needs no decision from Ludovic: every field now edits at the caret, as vanilla's EditBox does. rdbt's line-break review: the book page is a MultiLineEditBox in 26.3 (javap), so I pushed MultilineTextField's own word rule. getPreviousWord/getNextWord split words at any Character.isWhitespace and start from the cursor clamped to the last character. The single-line EditBox.getWordPosition keeps counting spaces only. I also merged main in twice (multiplayer.rs labels, stations.rs). Integration tree: `lan_e2e --check books`, `signs` and `anvil` pass in-frame and on the server thread, the `creative` and `container_screen` harnesses pass, and the ui/creative/title/clipboard unit tests pass. Thanks @Joansitoh!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). I checked CushionItem.useOn and Cushion.interact against the 26.3 bytecode with rea_vanilla and pushed three leftovers. (1) An occupied cushion now returns PASS, as Cushion.interact does. (2) A cushion placed in fire burns at once (destroyIfInFire). (3) Placing always posts the entity_place game event before consuming the item. I also merged main into the branch (lan_e2e.py). The suffocation check matches vanilla's block definitions except closed shulker boxes, which are approximated as always suffocating (noted in a comment). Integration tree: `lan_e2e --check cushions` passes in-frame and on the server thread, and the `cushion_placement` and suffocation harnesses pass. Thanks @ESHAYAT102!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). The launcher contacts no new host: the existing requests (releases, Mojang files, Modrinth, CurseForge, the mod registry, pictures, Microsoft sign-in, Tailscale's API, modpack jars) now go through the player's HTTP proxy, or HTTPS_PROXY/HTTP_PROXY/ALL_PROXY when unset. Loopback and NO_PROXY hosts stay direct, and the Servers page's local API never uses the proxy. Tokens (Microsoft, Tailscale) only cross the proxy inside TLS (CONNECT), so the proxy can't read them, and none are logged. rdbt's NO_PROXY review is addressed: `horizon-port-pack`'s agents apply the same rules (`no_proxy_lists_hosts_and_their_subdomains`). On Linux at da54f988d: launcher lib tests 132 passed, `cargo clippy --all-targets -D warnings` clean, `horizon-port-pack` 68 tests passed and clippy clean. Labelled feature. One small note for later: an unusable proxy address is printed to stderr as typed, so a `user:pass@` in it would show in the log. Thanks @zhykowo!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). `tools/bundle_tooltip_e2e.py` passes in-frame. On the server thread (what players run) it showed every bundle full, holding more bundles, so I pushed two commits: (1) BundleContents' stream codec (component 52, a list of ItemStackTemplate, checked with javap) plus sending it to clients in `network_stack`; (2) `Hub::stack` gave every stack inside an item the outer item's id, which also turned a crossbow's charged projectiles into crossbows on clients. It is fixed now: each inner stack looks up its own id. With both commits the e2e passes on the server thread and in-frame (306 px tooltip frame; main shows 117 px and fails). The screenshot shows 32 stone with a half-full bar, as in vanilla. Unit tests: `bundle_contents_are_item_stack_templates` and `a_bundle_keeps_its_contents_across_the_wire`. Thanks @levzzz5154!
🤖 Agent: Gate check by Ludovic's AI agent (Claude). I checked the loot functions against vanilla 26.3 directly. A vanilla server placed 48 chests with fixed LootTableSeed values: 16 tables (end city, stronghold library, ancient city, mansion, dungeon, trial chamber rewards and more) times 3 seeds, then rolled them. Horizon then opened the same world. Every chest got the same items, counts, slots and enchantments as vanilla, enchant_randomly and enchant_with_levels included: 48/48 on the server thread and 48/48 in-frame. The only difference is the order of keys inside the enchantments map, which vanilla doesn't read. EnchantRandomly and EnchantWithLevels match the bytecode, and rdbt's single-level review point is covered by `a_single_level_enchantment_draws_no_level`. It is also built and run in an integration tree with the other open PRs. Thanks @levzzz5154!
VanillaMobAiOracle ran out of its 3g heap in PoiManager.tick on a busy CI
runner. Each scenario forces 9x9 chunks at a new origin; the live heap after
GC swings between 0.3 and 1 GB over a run, with peaks near 2 GB before GC.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Hub::stack resolved every item id of a stack to the outer item's, so a
bundle's contents and a crossbow's charged projectiles reached a server's
clients as more bundles and crossbows. The tooltip then drew a bundle of
bundles, full. Each stack inside now looks its own id up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: merged main (aec52c85e). pr_ready --quick passes (1308 horizon-client tests, clippy, 507 LOC). The welcome server mod e2e passes 35/35 in every mode: in-frame, server thread, LAN host in-frame and on its thread with a guest, dedicated with a guest. Verify is green; parity now runs on main (parity-scope). Queued.
linux-runner.sh takes its packages from apt where there is no Nix, the
registration token from RUNNER_TOKEN (a machine where gh must not be logged
in), and RUNNER_CPUS / RUNNER_MEMORY for a machine that does nothing else.
playhorizon, a partner's 8-thread server, joins parity_hosts.json.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖 Agent: Gate check by Ludovic's AI agent (Claude). The server fix in this PR landed first in #1779 (levzzz's PR, same change to `hub.rs` and `agent.rs`, with the same `lan_bot` test). I merged main into the branch and kept main's side, so what this PR now adds is its two unit tests: `sprint::tests::sprint_survives_releasing_the_key_through_a_jump` (the client sprint state survives the key coming up on the jump tick) and `living_motion::tests::sprint_jump_adds_look_direction_impulse` (LivingEntity.jumpFromGround's 0.2 impulse along the look, and none for a walk). Both pass, and `lan_e2e --check sprint-fov` passes in-frame and on the server thread in the integration tree. Labelled skip-release, because players see no change. Thanks @mc070309 for tracking the cause down and for the tests.
🤖 Agent: Merging. verify and parity passed; the new unit test covers the merge of both key forms, and save_roundtrip no longer sees duplicate RecipesUsed keys.
Ludovic Marie