fcdf14c0dc8cab514b953046cf8a47b994b34d53
20
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
  gromandClaude Opus 5
|
948373f7a4 |
probes: a prop shares a slot with one that agrees, not only one identical
The interior prop-probe table holds 8192 and cannot be widened — 13 mesh-tag bits, with every bit of that u32 already spoken for. It keyed on exact coefficient bits, so two props shared a slot only when no MOLR light reached either *and* their baked MODD colours matched to the last bit. Measured in Ironforge at grom's login spot: 3,883 distinct probes standing still, 5,543 at a far clip of 777, and his session saturated during the world load — after which every further interior prop falls back to the room's flat day/night law. The distinctness is the per-prop baked colour, not the prop's position: folding the probe on a 16 yd grid instead of each prop's own centre removed only 4.5% of the slots. Rounding the *key* onto a 1/32 step removes 35%. A shared slot keeps the first claimer's exact rows, so the quantization decides who shares and never reaches the light the shader reads. Verified by render, not by argument: that Ironforge storeroom at the new default is byte-identical to exact keying across 1.44M pixels, while starving the table outright (the new `WOW_PROBE_CAP=0`) moves 109,830 of them by up to 14/255 — so the lane does drive this view, and the sharing costs nothing measurable in it. The table now also counts the distinct probes it refused, and all three overflow warnings report it. That number is the whole question when the table saturates: a small overflow is answered by sharing harder, a large one only by more slots, and until now the log could not tell them apart. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SgstFWtn1RxEJ5kVrpgYwn |
||
|
pkuzicandClaude Opus 5.5
|
853c09f44d |
Merge upstream/main (samwhosung/benilla b17be271) into the graphics programme
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Sam SungandClaude Opus 5.5
|
fe16ed14b3 |
String literals in benilla-ui, -world and -formats carry no record numbers (#299)
Part of #92, for `benilla-ui`, `benilla-world` and `benilla-formats`. - 88 string literals drop private record and bug numbers, the dates of private incidents, and the workspace's name for the maintainer. They are assert and panic texts, log and printout lines, the `--` comments of the Lua prelude and of test fixtures, and one Lua assert. Each keeps its fact in its own words. - Two test names lose their numbers: `burning_steppes_lava_builds_a_magma_surface_at_the_reported_pin` and `join_channel_by_name_answers_the_row_custom_and_nil_legs`. - A test-only event, `B271_PROBE`, is now `ERROR_PROBE` in the two crates that register and fire it. A fixture's error text in `benilla-app`'s `ui_script` tests drops its bug number, together with the assertion that reads it. - Doc comments that named a client function by a disassembler's label (`FUN_0068a830`) now give its address, in `benilla-formats`, `benilla-assets` and one `benilla-app` file. - Every literal change keeps each literal's line count, proven with comments stripped and every literal masked. The two renamed tests differ only in their names. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> |
||
|
pkuzicandClaude Opus 5.5
|
da0f1a0d80 |
Merge upstream main (samwhosung/benilla fd386e75) into lighting
71 upstream commits: the repo-wide comment rewrite, the hover highlight as the scene ambient, M2 Mod/Mod2x batches drawing the bare texel, test and contribution changes. Conflicts were comment-vs-code in 36 files; where upstream changed only comments our code was kept, and wow_model.wgsl carries both our lighting lanes and upstream's new unlit/Mod/Mod2x paths. Follow-ups folded in: - our nine Advanced Graphics / water / fog options tests take upstream's wow_data_or_skip! guard - PostUpdate ambiguity ceiling re-measured under upstream's new census: 466, with the per-family breakdown in the comment - LIGHTING.md: new upstream base Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Sam SungandClaude Opus 5.5
|
87a92eb295 |
Comments in benilla-world: one line of fact each, no history, no record numbers (#121)
Part of #41: the comments of `benilla-world`, Rust and WGSL, are rewritten to the rule in `docs/METHOD.md`, which is what the code does and the reference fact behind it, in a line. 24,226 comment lines become 8,863. - **The mechanical strip.** The citations that were only a private record number go first, and each edit is shown to be a pure removal. - **Wrong facts corrected.** The rewrite corrects the comments that stated wrong facts. Among them: - `0x6b62e0` is the WMO group's liquid renderer, not a doodad drawer. - The reference creates a WMO's props once, when a group enters its streaming window (`0x698720` → `0x695aa0`). It then admits them each frame the portal walk visits the group (`0x685d70` → `0x6838f0`). The old text had them created per visible group. - The liquid query's comments now state the reference's order: the map-object leg first, then the ADT (`0x69b6d0`). They no longer call the Outdoors/Inside split a match. - The interior probe runs from the character, not the camera. - The reference lights ground clutter. The old comment said it draws unlit. - The model-particle "pool order" deviation was wrong. Each instance is its own mesh, depth-sorted every frame. - The reference's vertical field of view follows the aspect ratio (44.1° at 16:9). The comment now gives it beside our fixed 45°. - The effect shader's far wall is the live `farclip`, 177 to 777 yd, not a fixed 777 yd. - Dozens more turned up on the way: renamed or deleted items, docs sitting on the wrong item, stale counts and links. - **Deviation labels.** "Deviation:" now marks only a choice with a reason. An unbuilt behaviour says so plainly. - **Verification.** Every file strips to identical code, and `cargo fmt --check` is clean. Five second readers compared every file old against new, restored the facts the rewrite had dropped, and checked each new citation against the reference. Defects the rewrite surfaced outside the comments are filed as their own issues (#106, #114, #115, #116). The comments here state what the code does. String literals are untouched; they are #92. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> |
||
|
Sam SungandClaude Opus 5.5
|
7b66dac9db |
comments cite the reference by address, not by private note (#30)
Comments that pointed into the maintainer's private reverse-engineering notes (a note by name, a path, a section or finding id, the analysis's own name for a client function, a process word) now carry the fact itself, cited by the reference client's address, by vmangos file and line, or as "the reference". Nobody outside could follow those pointers. Comment text only, plus the test-assert and log strings that carried the same pointers. No claim changes: a comment the notes have since corrected keeps its wording here, and correcting it is separate work. The two reference generators, `scripts/gen-reference-globals.py` and `scripts/gen-reference-verb-events.py`, named the private notes' layout to find their inputs. They now take the input files as arguments (`--capture`, `--framexml`, `--names`, `--extents`, `--disasm`). Run on the same inputs, both write the committed tables row for row; only the two header lines that named the old environment variable change. Checked per file: - the code is identical: each file with its comments stripped compares equal before and after; a changed string literal is compared under a mask that proves only literal text moved and every literal kept its line count; - no client address the old text cited was dropped; - a second reviewer checked every added address against its derivation; - a detector for the notes' names, paths, ids, function names and process words finds none left; - `scripts/gates.sh` is green. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> |
||
|
pkuzicandClaude Fable 5.1
|
839bc58d1b |
Merge upstream main (samwhosung/benilla edec5b22) into lighting
Ports the lighting settings into upstream's single settings registry and the lighting plugins into game_plugins.rs. 1108 world+formats tests green; world scenes (inn, Northshire dusk, Stormwind) render clean. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
pkuzicandClaude Fable 5.1
|
7048e5b34f |
Merge upstream/main (samwhosung/benilla 66388278, 78 commits) into everwood-cs
Keeps both sides: upstream's settings rework, model_render rework, mesh-tag change-gating, static_gx node move + diagnostics span, WorldPointLight, brightness/weather options, weapon trails; and the MONKEY lighting stack (sun shadow lanes, dynamic interiors, cube-map torch shadows, daylight fixtures, spell lights, bake floor, K=12 exterior selection, debug panel). All MONKEY light lanes migrated from Bevy PointLight to upstream's WorldPointLight. 1091 world+formats tests green; all 8 WGSL probes OK; server-less smoke (inn, Northshire) clean. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
SamandClaude Fable 5.1
|
00150c55b5 |
a crash leaves a report, and the wire's counts are capped
- a panic writes benilla-config/Diagnostics/crash-<time>.txt with the build id, the location, a backtrace and the last 200 log lines, and names the file as its last line, so a bug report can attach it (2266) - every wire-count allocation hint is capped and the compressed update object reads through the cursor, so a malformed packet can no longer ask for gigabytes (2266) - the 23 in-world run conditions are one set, the 45 probe variables one registry, the install-gated tests say when they skip, and every gate is timed (2266) - one workspace-level lint allowance replaces 413 per-site attributes, and the decision corpus gets a lint in the gate (2266) Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
SamandClaude Fable 5.1
|
cde247fe39 |
a chest fades out, a held key survives an alt-tab, the blits go
- a looted chest fades out over two seconds on the reference's own smoothstep, from the alpha it is showing, instead of blinking away; every destroy and out-of-range exit takes the fade (2198, 2203) - opening a chest goes through the cast ladder, so mashing right-click no longer cancels the running cast bar with "Another action is in progress" (2199) - a held key resumes after an alt-tab: a repeat is a key we already have down, not the OS repeat bit, as the reference classifies it (2204) - both upscaling blits are skipped: the world's combine and the UI's gamma decode write their camera's target themselves, pixel-identical, and the costliest pass on the lab's RTX is gone (2206) - WOW_GPU_PASSES prints the journal's per-pass GPU split beside each row (2205) Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
pkuzicandClaude Fable 5.1
|
082f3ebe50 |
MONKEY: lighting round — fire/lamp lights, light lanes, room gating, static torch cache
Point lights - Synthesised lights for fire and lamp models that author none (benilla-formats fire_light.rs: flame route from the particle emitter's over-life colour, name and emissive-glass routes for lampposts/lanterns/candelabra; NoOmni braziers included — they are Orgrimmar's street lighting). Spawned on every lane: ADT doodads, WMO props in interior and exterior groups (2.5 yd dedupe against authored MOLT fixtures), GameObjects and creatures (carried). Cvar fireLightGain (synthetic only). - Light LANE decided by a down-ray at the light's position: exterior consumers (terrain, models, WMO exterior batches — new per-vertex point term) read exterior lights; the interior room lane reads interior fixtures. Fixes interior fixtures leaking onto lawns and street torches lighting nothing. - Interior attenuation: the authored MOLT reach replaces the flat 48 yd, with a soft long-tailed profile (constant 1.75 yd core so energy no longer scales with reach), cvar interiorAttenScale; a 1.6x default. - ROOM GATE: a WMO surface takes an interior fixture only if the fixture claims its group — claims = containment ∪ MOLR ∪ portal-adjacent rooms (2 hops), shared rule in benilla-formats room_claim.rs, packed per light into a 32 KB storage buffer; portal claims are weighted (1 across the door frame, fading into the next room) so doorways carry no seam; in-building exterior-class groups take claimed fixtures. Carried lights (GameObjects, pets) claim rooms from their world position once settled. Cvar interiorRoomGate. Torch shadows - Casters ranked by contribution at the player with hysteresis and fade weights (no pop-in); eligibility 4R+3 and the shadowDistance setting; caster geometry gathered around the fixtures, not the camera. - Static cube maps cached per fixture (rendered once), entity casters rendered per frame only for the nearest interiorShadowDynamic fixtures via a depth copy; up to 16 resident maps (interiorShadowCasters), interiorShadowSoft PCF radius. - Carried lights: rest-pose flame position (no hand-bone jitter), lane from the bearer's cross-fade weight, caster only while settled — the imp no longer strobes. Entities - Portal lane cross-fade (tag bits 15..18) and MATTE_INDOOR bit so a unit keeps its room lighting across a Bake/Matte flicker; hover highlight folded into the room lane; the blob shadow returns whenever the realtime sun lane is inactive (night, beyond shadow distance) with a crossfade at dusk/dawn. Robustness - A placement or WMO prop whose model fails to load is retired instead of pinning the loading screen forever (Stormwind's alliancebrasscannon_flat.m2). - benilla-extract: wmolights (MOLT + claims + group classes), wmolamps (corpus audit), m2firescan (synthesis routes). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
SamandClaude Fable 5.1
|
6e3483c1e7 |
the loading tip draws, its backdrop holds, and indoor trails go warm
- the loading screen's tip of the day is on screen: it was picked and saved but never painted, and once painted it was wiped by the login round trip (2083, 2081) - the backdrop no longer switches mid-load — logging back in to an expired dungeon held its art for a second and then flipped to the continent's; the art latches at the raise, keyed on the texture, not the map id (2081, 2087) - a weapon trail indoors takes the room's warm light instead of the sky's ambient, the wearer's own light being the reference's term (2086) - the glue dialog moves out of the login screen into the shared glue layer, so the select screen's refusals and the login's errors are one widget (2084) Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
pkuzicandClaude Fable 5.1
|
7309297f1c |
MONKEY: portal lane cross-fade for entities
An entity crossing a WMO portal no longer pops between the exterior day/night result and the dynamic room lane in one frame. The classifier ramps a per-anchor lane weight at 2/s (first resolve snaps, so a unit streaming in indoors never fades up from dark) and, while ramping, writes its parts in exterior material mode with the weight in tag bits 15..=18 beside the MATTE_INDOOR bit; the Bake material commits only at full weight. wow_model.wgsl mixes exterior-lit -> room-lit by that weight (forced to 1 in interior material mode, where those bits are the probe slot). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
pkuzicandClaude Fable 5.1
|
e254e6dce8 |
MONKEY: dynamic WMO interiors + cube-map torch shadows (surfaces + entities)
Interior rooms light per fragment from their authored fixtures instead of the MOCV bake (direct + fill + ambient, soft rolloff; live cvars interiorLight/interiorAmbient/interiorFill/interiorExposure/interiorShadows/ interiorDebug), TRANS/EXT batches blend to the day/night reference at the portal seam, and fixtures admit camera-independently (90 yd). Torch shadows are a from-scratch depth-map renderer (Bevy point shadows are dead under ClusterConfig::None): a shared Depth32Float 512^2 x 24-layer Image + depth-only pipeline + ViewNode rendering the static caster and a per-frame entity caster into 6 cube faces for the nearest <=4 fixtures (reverse-Z, GreaterEqual compare, 4-tap PCF, far fade). WMO surfaces receive through static_gx group 3; entities receive on the wow_model path through WowModelExt bindings 91/92/93 (depth array, comparison sampler, 1616-byte torch table) with a 0.15 yd normal offset against CPU-vs-GPU skinning self-shadow. The entity lane is gated on is_interior OR the new MATTE_INDOOR tag bit (exterior payload bit 14, written by the classifier's Matte law), so a unit whose verdict flickers Bake<->Matte at one spot keeps its room lighting. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
SamandClaude Fable 5.1
|
4040166138 |
the model pane takes the file's camera, and the who list sorts
- the `<Model>` widget's perspective leg: a pane whose file supplies the camera it selected renders through a camera of its own whose viewport is its own cell of the same tile atlas — the reference's structure, one target, a viewport per widget, depth cleared per widget — with the composite, packer and reaper untouched; the selection is RAW (`0x76cec0` reads the count off `MD20+0x124` and the record at `[model+0x3c4] + idx·0x84 + 0x80`, `cameraLookup` never consulted), so `M2Model::cameras` is the whole table in file order and `benilla-extract m2cam` prints it, reproducing wow-re's numbers to the digit; the camera is two fields because the reference has two — `+0x320` a PENDING index and `+0x31c` the installed one — and while a request is pending the pane paints nothing (`76d5f0`'s draw gate, now the paint list's own filter); an index past the count installs the NULL camera, which IS the orthographic leg every shipped in-game pane takes; the projection is the booths' byte-derived diagonal-FOV matrix at the pane's aspect, and the eye and target ride the model's root transform as `0x718960`'s publish does — so `SetFacing` cancels along with `SetPosition` and `SetModelScale`, a contradiction with a wow-re note that went out as a §5 round and came back agreeing: the up vector `0x7ac640` builds is model-space +Z at roll 0, the very axis the facing turns about, and the note's sentence is scoped to `<PlayerModel>`'s frozen camera; each perspective camera takes a render layer of its own from a pool of eight, the orthographic camera clears the atlas and every perspective one clears only depth; `SetLight` is a typed `CGLight` that folds each intensity into its colour, normalises a direction on write, quantizes to 8 bits, returns before the copy on `SetLight(0, …)` and reads the second intensity at index 8 on a short ambient block; the five fog verbs are built (`ClearFog` is `76f5c5 and [edi+0x3a4],-2`, bit 0 alone), so the `Model` UNBUILT list is empty for the first time since 1134; a pane that arms a light or fog takes a rig from a lazily-grown pool, and the tiles' material twins — never reaped since the cache was written — are; `WOW_PROBE_MODEL_CAMERA=1` reads the renderer's own camera back for the three cancellations, `pass=5 fail=0` (2027)
- the layout cache writes in the shutdown tail: an unlocked chat window's size and position did not survive `/logout` or `/reload` and `benilla-config/layout/` was never created (B353) — the lane was right (1.12.1 persists a chat window through `StartMoving`/`StartSizing` and the engine's userPlaced bit; no `SetChatWindowSaved*` verb exists in the 5875 FrameXML), but `UiLayoutPlugin` hung its flush on `OnExit(InWorld)`, which a `/reload` never leaves, its debounced writer was never registered in `Update` at all, and on the roots where the edge did fire the VM-reading saver raced `end_ui_session`'s VM replacement on the same unordered edge (measured: the placement moves with nothing but registration positions); the reference's UI shutdown `0x490bd0` has an ordered tail — `PLAYER_LOGOUT` (`0x490c2a`) → `layout-cache.txt` (`0x490c79`) → the flat saved file (`0x490c7e`) → the per-addon files → `AddOns.txt` — on the same five roots as SavedVariables, `/reload` included, so `save_now` joins `shutdown_ui_state` in that slot, one call site reached from every root, and its own edges are deleted; the debounce stays as the crash writer and only that, a deliberate divergence, and the chat-look flush next door is pinned `.before(end_ui_session)` (2029); the VM gets its font engine at the LOAD edge: the measurer was seated only from an `Update` system on the first frame the glyph atlas exists, but `ReloadUI()` mints a fresh VM and runs the in-game load in the same exclusive slot, so every `<OnLoad>` and every addon's file scope measured 0 through the whole reload (`LOAD MEASURE ZERO` live, 33.185 after) — `seat_text_measurer` is one construction site called from both edges; the options tabs fit once from `<OnShow>`, the `OptionsTab_FitWidth` OnUpdate poll and its `fitted` latch gone, because `GetStringWidth` answers in the region's own units and this window's 0.78 scale is applied in the window's own `<OnShow>` after every `<OnLoad>` — an OnLoad fit read 73.19 where the settled value is 76.47; the settled widths are unchanged to the last digit (76.467232 / 91.661896) and the one-frame 72/87 flash at every open goes; four comments that still called the text measure a frame late, and seven gossip-suite comments that put `UIPanelScrollFrameTemplate` in our `ScrollTemplates.xml` when it lives in the reference's `UIPanelTemplates.xml`, are corrected (2028)
- the `/who` sort is a seven-key chain, not "order by the clicked column" (B365): the reference keeps seven `{key, dir}` slots at `0xc2817c`, seeded zone → level → class → group → name → race → guild; `SortWho` (`0x5ad890`) promotes the clicked key to the front and flips its direction ONLY when it was already there — Name, Name reverses, Name, Level, Name restores what the first click left — and the comparator `0x5ada00` walks the chain as tie-breakers, most-recently-used first, through an ASCII fold (`A`–`Z` by `+0x20`, not a Unicode one), the three DBC arms comparing the localized name out of `ChrClasses`/`ChrRaces`/`AreaTable` where ours had sorted by raw id, and an unresolved row TIES (its miss jumps to `0x5adbb2`, the chain loop's own `inc esi`); the click fires `WHO_LIST_UPDATE` synchronously through `SignalEvent` at `0x5ad9ed`, so the binding promotes and sorts the snapshot and the handler repaints before the OnClick returns, where ours redrew a tick later — a redraw one click stale reads exactly like a header that does nothing; the chain is process state (initialised once from `0x401666`), so it survives a logout; the rider from the same carve: `SMSG_WHO` has two exits, and with `SetWhoToUI` clear more than three rows go to the frame with nothing printed and three or fewer to chat (`0x5adfcc jl`, `eax = 3`) where we printed every unclaimed answer; and wow-re's §11 round corrected the first build before it landed — the chat lines come out in WIRE order with the total line LAST, because `0x5ae0a1` composes each line inside the parse loop and the `qsort` at `0x5ae0e2` sits past its back-edge (2030)
- two lighting reports, one frame and one bit each: the surfacing hitch (B354) is a one-frame stale atmosphere — `update_time_lighting` is late by construction, waiting on the wire drain, the weather tick and the submersion verdict, and five `Update` systems that read the `WowLighting` it resolves declared no ordering, so the executor handed each of them last frame's value, deterministically and invisibly while the atmosphere drifts a minute at a time; at a submersion crossing it jumps, on the very frame the sky and cloud domes un-hide, so the first dry frame painted `LightParams` 27's underwater sky over a dry world (the clip's blue frame, (58, 137, 173) against the settled (53, 58, 34)) and `tick_clouds`' surfacing full-rebuild ran at the underwater cloud density, 0.0, rebuilding an empty field that crept back over 0.4 s — the fix for the pop-in handing the symptom straight back; `LightingConsumeSet` names the read side of `LightingResolveSet`, tied in one place, and four consumers join it, the FFX haze and warp are ordered after the verdict they had the same hole against, and two tests enforce it — zero ambiguities between the two sets, and a source scan over `benilla-world` that fails any `WowLighting` reader not in the set; `emit_minimap` is left stale knowingly because joining cost a 178th item through the world API wall; `WOW_FORCE_SUB=<frames>` holds the eye verdict submerged and releases it, so the crossing reproduces server-less (2032); the black furniture below a boat's deck (B373): `MeshTag` bits 6..=13 are the exterior payload's ground-shade byte and bits 6..=18 the interior payload's SH-probe slot, and `entity_shade` guarded the overlap by asking whether a part was on the interior classifier's Bake law — one of the TWO populations holding a probe slot; the other is a lit interior MODD prop, which a WMO-display GameObject parents under the net entity so it sails with the deck, putting it inside the light node's descendant walk — so every frame the boat's shade byte was written over each prop's slot, `(slot & 0x1f00) | byte`: a neighbouring prop's light where that row was live, a zeroed row — solid black — where it was not; the guard now asks the payload question (`probe_payload`) at both write sites, the tag's bits and the marker derived from one binding; measured live at Ratchet, 749 of 749 resident interior prop parts renamed before, 0 of 640 after, and both ledger leads refuted at the bytes — the MODD colours are healthy (`floor112` raises even `#010101` to diffuse 112) and the fold is rigid-motion invariant since every 1.12 transport authors zero MOLR lights; a prop's mouseover identity now names its lane, words and probe slot, and `WOW_PICK` prints `model_flags.z` (2031); the sea-crossing probe can now measure the seam a report names — an explicit `WOW_PROBE_DOCK` goes first, boards only once the body stands at the dock, carries a map id, takes only a candidate within boarding distance, and drops onto the deck rather than into the hull, because a taxi node's `z` is the WATERLINE for the sea ferries while their decks stand ~16 yd above it — and the seam re-anchor returns the cycle it moved to or `None`, so a healthy crossing no longer logs the server's 9.7 h uptime as a skew inside a 350 s cycle
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
||
|
SamandClaude Fable 5.1
|
f9293ffbe8 |
the lamppost halo stops stacking, and the stance bar seats on login
- the lamppost halo stops stacking after a short-range teleport: bevy registers `add_visibility_class::<Mesh3d>` as an add hook that pushes the mesh's class id onto the entity's `VisibilityClass` list on every (re)add, nothing dedups the list, the required component survives a `Mesh3d` removal, and `check_visibility` queues an entity once per entry — so 1949's parking, which removes `Mesh3d` after thirty hidden frames and re-adds it the frame the part is shown, drew a part one more time per hide-then-show cycle; opaque parts hide it at the same depth and colour, an additive glow card adds itself to itself (×2 after the first teleport that hid the lamp for half a second, ×3 after the next), and a loading screen despawns the tiles and respawns fresh entities, which is why a far teleport and a relog never showed it; the unpark queues one dedup after the hook, the test registers bevy's own hook and requirement in a bare world and reads 2 entries without it and 1 with, and on the rig, with the away camera facing the lamp past a 177 yd far clip, a build without the dedup read `stacked_parts=278` after one hop back where the landed build reads 0 (2003, 2005); the instruments that excluded the other mechanisms stay — the inspector's glow-card line names the material's additive bit, fade/unlit flags, depth write, texture residency, tag alpha and live scale, `STACKED n` on the card and `stacked_parts=` on the probe line count the duplicated class lists, and screenshot and pick bursts now keep the rig window awake, because an occluded window recaptures one stale drawable and a day of A/Bs had read identical to the decimal
- `perf::blend_check` reads the half of a glow card's look the material cannot: per `Transparent3d` item, the pipeline descriptor's colour blend against the material's additive marker — the marker folds the radial alpha into the colour in the shader and gives the pipeline its pure (ONE, ONE) add, and 0071's wrong-pipeline class was immunised only for opaque batches; mismatches count into the perf pill as a red `blend xN` the frame they happen, ride the probe line as `blend_mismatch=`, and are tracked across frames and named — object, texture, which pipeline, PERSISTING; its own first harvest was the tracker: an additive material realized this frame is extracted, prepared and drawn this frame, and a tracker reading the `Added` event a frame late called every first draw a mismatch, so it runs after `AssetEventSystems` like bevy's own specialization check; and the hook-accumulation sweep states the class so it can be swept: bevy 0.18.1's insert path fires `on_add` only when the component was absent, `on_replace` on every outgoing value, `on_remove` on removal only — a push on `on_add` accumulates across remove→re-add, a free on `on_remove` leaks across an insert-overwrite; every hook in the bevy crates we link (the visibility class on `Mesh3d`/`Mesh2d`/`Sprite`/`Text2d`/`MeshletMesh3d`, the cluster class on every bevy light) and every removal of a hooked component in our tree was read, parking's `Mesh3d` is the only accumulating pair, the UI quad pools reuse entities by insert-over and fire no add hook, and the one asymmetric freer, `PropProbeSlot`, freed its probe slot on `on_remove` — an insert over a live slot leaked it for the session, held off by a remove-then-insert discipline at the one re-seat site; it frees on `on_replace` now (`RigSkin`'s shape), the re-seat is a plain insert, and a test overwrites a live slot and reads it freed (2005)
- the stance bar loads after everything the managed-position pass reads: the stock `ShapeshiftBar_OnLoad` shows the bar, its OnShow runs `UIParent_ManageFramePositions`, and the stock pass indexes `ReputationWatchBar` unguarded — our manifest seated the bar beside the other bar files where the reference's toc has it at l.74, after the reputation frame at l.57, so the pass raised at `UIParent.lua:1618` on every login and never reached the arm that seats the shelf: the red login error and the sliver above Defensive Stance (the shelf's plate at its XML anchors, a few pixels off its seat, healed by any bottom-bar toggle re-running the pass) were one defect from 1988, whose retired copy of the pass had guarded every read — and the three-form case 2000 could not reproduce; the bar sits at the manifest's tail right before `WorldStateFrame.xml`, whose own OnLoad runs the pass and indexes the shelf textures the stance bar declares — the reference's own reason for l.74 before l.89; a manifest test states the law with the pairs read off the pass's body, and `load_default_ui` now folds the VM's handler errors into its failures, because a raise one call below an OnLoad landed only in the VM's error list and shipped through the manifest test, the full-manifest harnesses and the smoke's WARN alike (2001)
- `UnitName` answers `UNKNOWNOBJECT`, never nil, for a resolved unit whose name is still in flight: a hunter casting Call Pet with the stable window closed got `PetStable.lua:129: attempt to concatenate a nil value`, because stock `PetStable_OnEvent` repaints on `UNIT_PET` without asking whether the frame is shown and the paint concatenates `UnitName("pet")` the instant `UnitExists("pet")` is true, while the pet's name is a `CMSG_PET_NAME_QUERY` round trip away; the reference's `UnitName 0x517020` pushes nil in exactly two cases — the `"player"` fast path (`0x517083`) over an empty name buffer, and a token that resolves to GUID 0 (`0x5170c0`) — and every other path ends in a string: `CGUnit_C::GetUnitName 0x609210` looks the pet up in `petnamecache.wdb` by `UNIT_FIELD_PETNUMBER`, re-sends the query on a miss or a stale stamp, and falls to `0x609324`, the GlobalString `UNKNOWNOBJECT` ("Unknown") read through `0x703bf0`, or the literal `"Unknown Being"` at `0x860fa4` when the global is missing or empty; the binding now reads the global out of the VM's `_G` at call time, so a translated `GlobalStrings.lua` translates this too, the stock file is untouched, and 1688's "nil for an unresolved name is faithful" is corrected — the reference paints `Unknown Level 58 Boar` into the hidden frame and the next repaint reads the name; `GameTooltip:SetUnit` still titles a pending-name unit with an empty line, a next move (2002)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
||
|
SamandClaude Fable 5.1
|
1f3ec2d538 |
the waterline stops stepping, and mobs in far rooms lose the keep's fog
- the step in the water at the shore, which travelled with you as you walked: `WATER_BIAS` (−2e4) exists to give the water pass its sort slot, and bevy's `StandardMaterial` also packs that f32 into the rasterizer's depth bias, so every water surface in the game shipped with a −20000 depth pull — on the one world surface that meets the ground at a grazing angle over tens of yards that does not settle a tie, it moves the waterline, and a float depth buffer's bias unit doubles at every exponent boundary, so two water triangles either side of one stopped agreeing about where the shore is, at a fixed distance from the eye; the bias is out of the rasterizer (a ~17 px jog at the reported Stranglethorn beach, present before and absent after) - the foam wake went dull under the next water chunk and painted onto dry sand: it sorted against ONE water surface with a coplanar epsilon (the chunk 33 yd nearer the eye outranked it thirty times over, while the reference drains every liquid queue before it draws foam, `0x6816d0`), and its coplanar settle was a 0.03 yd geometric LIFT, which on a 29% beach is 0.10 yd of dry sand painted and proportionally more on every flatter shore; it is a rung above the whole water band now and sits exactly on the plane, and the RE on its depth state came back: a depth-buffer bias (`footstepBias × [0x810390]`), depth-tested LEQUAL with write off — the "−2048 GL units polygon offset" our comment justified the lift with existed nowhere in wow-re - a unit standing in a room the camera's chain does not reach wore the building's fog while the walls around it wore the scene's: the reference's unit test has a second conjunct (`[P+0x98]`, the same `[0xca7f00]` chain gate the walls use), so the attach ray now returns the room it claimed and the gate is re-read every frame because it moves with the CAMERA — at Shadowfang, from inside the pinned room, three true interiors are in frame across the courtyards and a mob in g13, 91 yd out, wore the keep's teal at ~84% against a scene fog of ~3% - two instruments for the shore class of report: `water_here <map> <x> <y>` reads a shoreline offline (the wet-cell lattice, the plane on it, the terrain under it, and the four scans that tell an authored defect from a rendered one), and `WOW_CAPTURE=waterfx WOW_WFX_AT=x,y,z` wades the foam rig in the REAL streamed water at a pin, where the synthetic square had no bank to clip against Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
SamandClaude Fable 5
|
e9e38bf6ec |
your corpse is your own dead body, and particles take the room's light
- corpses render: a corpse create spawned a real entity, latched its guid for the reclaim send, and drew nothing — one missing EntityKind variant folded it into the catch-all and four downstream lanes then failed by default, not even the debug cube fired; the reference forks once, on the bones flag: a fresh body builds a character composite from the corpse's OWN appearance bytes and 19 item slots through the same compositor a living player is dressed by (its own hide-helm/hide-cloak bits honoured, the ranged slot never dresses, and weapons never dress because the client looks a packed display word up as a guid — we reproduce the outcome, not the dead lookup), while a bone pile takes its race and sex's DeathSkeleton, sixteen of which ship; the pose is one settled arm — Dead, or Drowned when the body lies more than 0.66666669 yd under a liquid surface — and a corpse is not a unit, so clicking one no longer sends a selection for a guid no unit owns
- a lit particle takes its MODEL's committed light node, not the sky's: indoors the reference lights an object's particles from the object's own node — the footprint's baked floor colours on the fixed interior axis plus the room's placed lights — never the day/night sun; a particle quad carries exactly one normal, world up, one constant per draw, so the whole node collapses to three floats, published off the same fold the model's mesh renders from so a mesh and its own particles can never disagree about the room they stand in; in Onyxia's Lair the 78 lit falling-rock emitters on the 26 anchors that classify interior now take the room's light
- and that lit lane is FIXED FUNCTION, not the mesh's soft SH curve: the shipped client null-binds both shader slots for a particle batch and re-zeroes the shader mask at the head of every batch, so a model and its own particles share the same committed light words and legitimately differ in the curve applied to them — the mesh's wrapping lobe against the hardware's max(N·L, 0) is 0.900 vs 0.869 on the key axis, up to the three nearest room lights arrive diffuse-only under GL's own 1/(0.7d + 0.03d²) falloff (a light at or below the emitter cannot reach it at all), and the clamp sits on the product, never the multiplier, because it is GL that clamps the lit vertex colour
- the cinematic letterbox has no aspect test in it: CinematicFrame.lua's if guards a recompute, not whether there are bars — below 4:3 the reference's bar textures simply stand at their declared 1024x128 on the client's native 1024x768 sheet, which is height/6, the same formula evaluated at exactly 4:3 — so one expression covers every window shape, min((h − min(w/2, h))/2, h/6): a 2:1 picture when wide, bars at 4:3 too; and a shown keyboard frame swallows EVERY key, not the ten names the host used to feed it — you could walk around underneath your own intro; the proof is the reference's own handler, which has to RunBinding("SCREENSHOT") by hand to get one key back; ESC still skips, and a W already held when the shot starts keeps running, which is faithful too
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
||
|
SamandClaude Fable 5
|
d99435beb8 |
the performance hunt, day one: the frame sheds 1.85 ms
- UI layout solves what moved, not what exists — hovering a full bag no longer re-solves the whole interface - world albedo textures stop keeping a second copy of themselves in RAM - collision gets a coarse grid for oversized slabs (32x fewer triangles tested), idle units stop re-asking the ground every frame, and anchors spawn on first consumer instead of by the thousand - aura and loot watchers scan when the store changes rather than on the clock, and static-transform tracking and plate reaping join the sweep Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> |
||
|
SamandClaude Fable 5
|
bdef9cd790 |
benilla-world: the engine becomes its own crate
- the world engine splits from the game into a benilla-world crate — 112 files, and the wall is ratcheted to zero names of the game inside the engine - the engine's public doorway is counted and gated at 32 items, and the render materials, shaders and asset cache move down to benilla-assets - the game's files leave the engine's asset root — game:// is its own space — and a spike binary boots the engine with no game attached - weather rides along: snowfall emission is denominated in the reference's frame cap, and a long teleport retires the stranded precipitation field instead of leaving it behind Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> |