On some Linux setups the game was silent. The device opened, but its data callback ran once and stopped, with no error. The music position never moved.
This is cpal 0.17.1's ALSA bug (RustAudio/cpal#1099, "Data callback called only once on ALSA"), fixed in 0.17.3 by setting playback's `avail_min` to one period. benilla's stream starts only at the start threshold, which is exactly that case. cpal goes to 0.17.3.
alsa 0.11 and bevy_audio's unbuilt cpal 0.15 cannot share a lock (`links = "alsa"`). A lock-only stand-in for alsa 0.9 settles that, documented in `third_party/alsa-0.9`.
The realtime-fallback log line now says what happens on each platform. Following the default device gets its `Deviation:` line: 1.12.1 initialises sound once (`0x7a4330`).
Built on macOS, with Linux and Windows linted.
Fixes#820🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit d93c231d8ebb27934791b726a0575f4c9b709446)
# Conflicts:
# Cargo.toml
# crates/benilla-app/examples/extended_launcher.rs
# docs/DEVIATIONS.md
Absorbed into the octo fork. Cargo.toml unions both patch entries (our mlua-sys traceback patch and upstream's alsa 0.9 lock-only stand-in). Cargo.lock is regenerated locally rather than taken from upstream, with cargo update -p cpal@0.17.1 --precise 0.17.3: cargo reaches the same resolution (alsa 0.9.1 from the stand-in, alsa 0.11 + alsa-sys 0.4 for the built cpal, windows-* dropped), so it is our own lock, resolved here. examples/extended_launcher.rs is upstream-only and stays absent. cargo check -p benilla-app clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SgstFWtn1RxEJ5kVrpgYwn
This checkout had no git history at all, so this one commit carries the whole
of the Everwood work as a single import on top of chairmanpow/benilla-octo's
`main` (706cd71). The two trees have genuinely diverged — each has files the
other does not — so this is a fork point, not a catch-up.
Today's work, which is the reason a baseline exists at all:
- Booth previews no longer cast world shadows (`NO_BOOTH_SHADOW`). Portrait
characters were entering the sun's shadow cascades, so every skinned-rig
shadow specialization compiled live mid-frame; the hitch detector logged
"main thread blocked 1316 ms" beside a `compiled LIVE` warning. This was
the Fashion/Tmog stutter.
- Per-pane dress-up: each `DressUpModel` pane assembles and stages its own
tried-on look instead of every pane mirroring the live player.
- Item template cache is append-only JSONL written on the IO pool, replacing
an 11.6 MB whole-map JSON rewrite on the main thread every 10 s.
- `WOW_SYSTEM_TIMING` rebuilt per-thread: it took 2 mutexes per system enter
and 4 per exit, costing 15 ms a frame (590 fps -> 60) and corrupting the
measurements it was taken to produce.
- The visibility-churn sweep moved to its own `WOW_FLASH_CHURN` switch, so a
frame-cost reading no longer pays 2.7 ms a frame for it.
Local state (`benilla-config/`, `logs/`, `target/`) is excluded by the
existing .gitignore; the account credential it holds is not in this commit.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SgstFWtn1RxEJ5kVrpgYwn
Same fork policy: upstream structure wins, local-unique re-instated.
Cargo.toml unions both measured dev-profile philosophies (codegen-units=1 +
lto=off from samwhosung, debug=line-tables-only + split-debuginfo from the
fork's disk measurements).
glue_booth.rs takes samwhosung's typed GlueScene (token/of_race/from_model_path)
and re-adds the fork's TurtleWoW races (Goblin=9, BloodElf=10) as variants;
their glue fog stays on Human's row, exactly as the fork's scene_fog fallback did.
ui script takes samwhosung's mouseover/hyperlink rework (world_mouseover,
minimap_blip, hyperlink.rs) over the fork's inline hover_link/span_at/pointer
path — the rework covers the same OnHyperlinkEnter/Leave behaviour.
game_plugins.rs keeps the fork's re-measured ceilings (466 / 5,509); mail.rs
keeps the explicit 13-tuple with a float daysLeft; MAP.md takes the superset.
Fork policy (docs/METHOD.md): upstream structure wins; local-unique work
re-applied on top. This commit is the upstream half of the reconciliation.
Upstream is the same program at a rewritten history (884dff77 == 4e0e89cb
tree), so the merge base was replaced to keep a real three-way diff. The
large conflicts are upstream moving first: the UI manifest (benilla.toc ->
the binary LAYER_MANIFEST/layer.toc split), the CVar table (cvars.rs ->
cvars/{mod,table}.rs, +24 video rows), the editbox flush model, the
host-side GetTime, the Options* -> BenillaOptions* rename, expanded
ui_unit.rs, and a rebuilt volumetric_light.rs.
Local-unique pieces re-instated on the upstream structure:
- OptionsFrame.xml: the 1.12 OptionsFrameSliders table restored under the
Benilla rename, with the TurtleWoW deviation note (patch-4 replaces
OptionsFrame.lua).
- layer.toc: CharacterFrameAdapters.xml (the sixth TurtleWoW character tab).
- ui script: wipe/sizeof, BenillaGetMerchantItemStats, and the attach/mount
teardown re-walk all survive.
- game_plugins.rs keeps UPDATE_ACTIONABLE_CEILING=5_509 (re-measure pending).
The LFT/Turtle_TransmogUI rows benilla.toc named off the player's install
now come from the chain's own FrameXML.toc via the new core loader.
An incremental rebuild of `benilla-app` spent most of its time in ThinLTO: above opt-level 0 rustc runs it over a crate's own codegen units by default, and incremental compilation redoes it whole. `cargo play` already turns it off, which left its dependencies without cross-unit inlining.
`[profile.dev]` now sets `lto = "off"`, and dev and play build dependencies as one codegen unit each.
Measured on an M4 Max, one-constant edit in `benilla-app`, cpu_ms at the leg pin:
| | before | after |
|---|---|---|
| `cargo build -p benilla` | 24.4 s, 5.2 GB | 5.6 s, 2.7 GB |
| `cargo test -p benilla-app --lib --no-run` | 25.8 s, 7.0 GB | 8.3 s, 4.0 GB |
| cold dev build | 5:19 | 4:06 |
| dev cpu_ms | 11.06 | 11.71 |
| play cpu_ms (release: 10.1) | 11.74 | 10.63 |
Release and ship are unchanged.
Fixes#852.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Apply the formatter baseline needed by the generated CI gate.
Verified the full treefmt check with vendored sources and generated MAP.md excluded. Parsed formatted Cargo manifests and compared their values with HEAD; all are identical.
DWARF is ~97% of what rustc writes here: the largest `benilla-app` codegen unit measured 145 MB,
of which 61 MB was `.debug_str` and 13 MB `.debug_info` against 1.5 MB of `.text`. That carried a
1.5 GB `libbenilla_app` rlib, a 2.9 GB test binary and an 82 GB `target/debug`. `.debug_line`
survives, so a backtrace and `gdb bt` still name a file and a line; `split-debuginfo` leaves the
rest in the object files rather than copying it into the linked binary.
An ambience change read and decoded the whole bed on the main thread, about 40 ms a swap. 1.12.1 opens the bed as a looping FMOD stream that reads the archive through its file callbacks (`0x460b54` → `0x7a54d0`, `FSOUND_Stream_Open` at `0x7a5525`).
Now:
- MPQ files open for in-place reads, one sector at a time.
- The bed streams from the archive, opened on the IO pool. A newer swap drops an open still in flight.
- kira moves to 0.12.4 on symphonia 0.6. 0.5 sized PCM frames from `nBlockAlign`, which 98 stereo WAVs store as 2, so a streamed bed fell silent halfway.
- A fork patch stops a stream at its last packet instead of spinning.
- A sector that inflates short is refused.
Tests cover a bed's true length, a seamless loop, an off-thread open, the stop at the last packet and the refusal.
Fixes#700Fixes#8🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The 0.2.0 release: the workspace version is 0.2.0, `rust-toolchain.toml` pins the stable the release was checked with, `1.98.1`, so the tag builds with that compiler after stable moves on, and the README says each release is a tag on the Releases page.
Checked on this tree: the land gates, the live smoke run (its build line reads `benilla build 0.2.0`), the Linux and Windows compile of `benilla-app`, and the client on Vulkan (lavapipe) through its four capture scenes.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The 0.2.0 release: the workspace version is 0.2.0, `rust-toolchain.toml` pins the stable the release was checked with, `1.98.1`, so the tag builds with that compiler after stable moves on, and the README says each release is a tag on the Releases page.
Checked on this tree: the land gates, the live smoke run (its build line reads `benilla build 0.2.0`), the Linux and Windows compile of `benilla-app`, and the client on Vulkan (lavapipe) through its four capture scenes.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
One line per dependency where the reason is not obvious, measurements
kept as the number and the unit. Parsed with tomllib before and after,
every manifest is identical.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
One line per dependency where the reason is not obvious, measurements
kept as the number and the unit. Parsed with tomllib before and after,
every manifest is identical.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
What a clone lacked: the gates went green with a thousand skipped tests and said nothing, every
login script assumed a worktree-pool slot and its probe account, the client defaulted to the
maintainer's account, a docker path and a home directory were in the code, the temp files broke
on GNU coreutils, and two generators could only run against a sibling repo.
Now the test gates print how many data-gated tests skipped and why; a scripted run logs in as
the account its checkout declares in `.probe-identity` or as the three variables name, never a
default, and the client refuses a scripted login on any other account; the event and message
tables regenerate from vendored inputs, byte for byte; the docs state the real prerequisites.
Out of the tree, into the maintainer's own tooling: the worktree pool, its hooks, the perf leg
runner, the cross-platform compile and the Vulkan container leg — a way of working, and setups
not every machine has. One hook stays: no writes into the install.
Two tests that could never run are gone with their subject: the quest-timer geometry diff
compared a copy of the window that no longer exists (and joined a chain path onto a folder, so
it always skipped), and the FrameXML flag gate now reads the stock files off the player's own
patch chain instead of an extracted folder only one machine had.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
What a clone lacked: the gates went green with a thousand skipped tests and said nothing, every
login script assumed a worktree-pool slot and its probe account, the client defaulted to the
maintainer's account, a docker path and a home directory were in the code, the temp files broke
on GNU coreutils, and two generators could only run against a sibling repo.
Now the test gates print how many data-gated tests skipped and why; a scripted run logs in as
the account its checkout declares in `.probe-identity` or as the three variables name, never a
default, and the client refuses a scripted login on any other account; the event and message
tables regenerate from vendored inputs, byte for byte; the docs state the real prerequisites.
Out of the tree, into the maintainer's own tooling: the worktree pool, its hooks, the perf leg
runner, the cross-platform compile and the Vulkan container leg — a way of working, and setups
not every machine has. One hook stays: no writes into the install.
Two tests that could never run are gone with their subject: the quest-timer geometry diff
compared a copy of the window that no longer exists (and joined a chain path onto a folder, so
it always skipped), and the FrameXML flag gate now reads the stock files off the player's own
patch chain instead of an extracted folder only one machine had.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- the spell owns its lifecycle state: the in-flight cast, the local cancel, the cooldowns and the spell-modifier tables live in one module (2328)
- a malformed MPQ, DBC, M2 or WDT file is refused instead of crashing the client, nine read sites bounded
- a test that needs the install or the addon corpus fails instead of skipping when the data is there; the Bagnon render suite runs again (2329)
- the dev crash injector is armed on every platform, and no instrument writes a path relative to the working directory
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- the spell owns its lifecycle state: the in-flight cast, the local cancel, the cooldowns and the spell-modifier tables live in one module (2328)
- a malformed MPQ, DBC, M2 or WDT file is refused instead of crashing the client, nine read sites bounded
- a test that needs the install or the addon corpus fails instead of skipping when the data is there; the Bagnon render suite runs again (2329)
- the dev crash injector is armed on every platform, and no instrument writes a path relative to the working directory
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- 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>
- 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>
- five idle minutes sit you down and flag you AFK, thirty log you out through the ordinary countdown; holding a key is idle, moving the mouse is not, and a swimmer stays afloat (2092)
- a knockdown no longer wedges: the landing froze the clip that held the base-animation lock, leaving a knocked-down player on their back for the session (2098); the perf cost pill starts hidden (2099)
- Lua is exactly 1.12's 5.0 — no vararg expression, no #, no %, the remembered table size — so Cartographer loads without errors and Mik's Scrolling Battle Text shows its text, in its own shipped font now that addon TTFs load (2101, 2102, 2103)
- a widget name is a Lua global for SetPoint and its kin, so Bartender2's bars spread out; a button's NormalText disowns its justify, the raise renumbers hidden frames too, and a missing toc file warns instead of erroring (2105, 2100, 2104, 2107, 2106)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- five idle minutes sit you down and flag you AFK, thirty log you out through the ordinary countdown; holding a key is idle, moving the mouse is not, and a swimmer stays afloat (2092)
- a knockdown no longer wedges: the landing froze the clip that held the base-animation lock, leaving a knocked-down player on their back for the session (2098); the perf cost pill starts hidden (2099)
- Lua is exactly 1.12's 5.0 — no vararg expression, no #, no %, the remembered table size — so Cartographer loads without errors and Mik's Scrolling Battle Text shows its text, in its own shipped font now that addon TTFs load (2101, 2102, 2103)
- a widget name is a Lua global for SetPoint and its kin, so Bartender2's bars spread out; a button's NormalText disowns its justify, the raise renumbers hidden frames too, and a missing toc file warns instead of erroring (2105, 2100, 2104, 2107, 2106)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- the quest giver's dialog and the quest log are the reference's own `QuestFrame.xml` and `QuestLogFrame.xml`, run off the player's chain at the reference's own toc seats; our 1366-line and 1999-line copies are deleted, and the verbs the stock files call are built — the four `GameTooltip` quest-item and reward-spell setters, the reward spell itself (parsed on all three packets and dropped by the model until now), and two verbs; the detail pane could never scroll because `UpdateScrollChildRect` measured a frame late where the reference's measure is synchronous, so it measures through the installed font engine now; and ten things the reference does that ours did not, each found by a test that had enshrined ours — a shift-click never selects, an empty log hides the whole detail pane and says "No Active Quests", the abandon confirm lives in the popup registry (1944)
- the quest bindings' law, read off the binary and folded back: no 1.12 packet carries a quest background material — `GetQuestBackgroundMaterial` (`0x502230`) forks on the dialog's SOURCE object, an item answering its template's PageMaterial, a GameObject its template data, a creature nil (FrameXML's "Parchment"), the id naming a `PageTextMaterial.dbc` row whose six words are nowhere in the binary; its twin `ItemTextGetMaterial` has the same body, so the reader window already held every piece and the giver's feed runs the same resolve; `QuestChooseRewardError` is game-error row `0x98`, and the row plays `igQuestFailed` first, which the verb now queues; the reward-spell getters' third return is `Spell.dbc` Attributes bit 5 (1161 of 22357 rows set — craft-cast spells set, `Pattern:` teaching spells clear), read off the catalog now (1946)
- the "20 ms frames" of a raid, chased to their number: the probe's worst frames now carry their churn (entities, assets, first-play sound decodes, live particles, the GPU reading) and in a Blackrock raid with sound on every one of those read flat; then the same window on the PACED clock — under vsync with pipelined rendering the raw delta is mis-split into long/short pairs that sum to exact refresh multiples, so an alone Stormwind leg read 115 of 300 frames over budget with cpu and gpu both at their means, while the paced clock read 0 skipped frames in every leg; what remains real in the raid is `cpu_p99` at 22–23 ms with sound on; and every submesh the streamer spawns now carries why it is an entity, counted per reason — 5.8k resident ADT-doodad submeshes at the Stormwind auction house, 5.3k of them hidden, three prop families the retained pass declines by rule, each riding every per-`Mesh3d` sweep for ~0.4 ms of the alone frame (1947)
- a raid's dozen voices cost the mixer ~1.5 ms of every 60 Hz frame: upstream kira spatializes every frame — normalizes, the attenuation curve and a dB→amplitude `powf` per frame per spatial track, two more `powf`s for the volume tween, ~1.7 M `powf`s a second at 48 kHz × 12 tracks for gains that change imperceptibly across a chunk — so kira 0.12.1 is vendored with one patch: the spatial gains and the volume × fade amplitude are evaluated at the chunk's two ends and interpolated across its ~10.7 ms, the granularity FMOD 3, the reference's own mixer, updates them at; nothing a listener can name changes (1945); and rig poses are evaluated in parallel over rigs — each reads its shared clips and writes only its own pose, and a raid's ~400 live rigs were 0.4 ms of one thread walked serially on the main thread's critical path (1947)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- the quest giver's dialog and the quest log are the reference's own `QuestFrame.xml` and `QuestLogFrame.xml`, run off the player's chain at the reference's own toc seats; our 1366-line and 1999-line copies are deleted, and the verbs the stock files call are built — the four `GameTooltip` quest-item and reward-spell setters, the reward spell itself (parsed on all three packets and dropped by the model until now), and two verbs; the detail pane could never scroll because `UpdateScrollChildRect` measured a frame late where the reference's measure is synchronous, so it measures through the installed font engine now; and ten things the reference does that ours did not, each found by a test that had enshrined ours — a shift-click never selects, an empty log hides the whole detail pane and says "No Active Quests", the abandon confirm lives in the popup registry (1944)
- the quest bindings' law, read off the binary and folded back: no 1.12 packet carries a quest background material — `GetQuestBackgroundMaterial` (`0x502230`) forks on the dialog's SOURCE object, an item answering its template's PageMaterial, a GameObject its template data, a creature nil (FrameXML's "Parchment"), the id naming a `PageTextMaterial.dbc` row whose six words are nowhere in the binary; its twin `ItemTextGetMaterial` has the same body, so the reader window already held every piece and the giver's feed runs the same resolve; `QuestChooseRewardError` is game-error row `0x98`, and the row plays `igQuestFailed` first, which the verb now queues; the reward-spell getters' third return is `Spell.dbc` Attributes bit 5 (1161 of 22357 rows set — craft-cast spells set, `Pattern:` teaching spells clear), read off the catalog now (1946)
- the "20 ms frames" of a raid, chased to their number: the probe's worst frames now carry their churn (entities, assets, first-play sound decodes, live particles, the GPU reading) and in a Blackrock raid with sound on every one of those read flat; then the same window on the PACED clock — under vsync with pipelined rendering the raw delta is mis-split into long/short pairs that sum to exact refresh multiples, so an alone Stormwind leg read 115 of 300 frames over budget with cpu and gpu both at their means, while the paced clock read 0 skipped frames in every leg; what remains real in the raid is `cpu_p99` at 22–23 ms with sound on; and every submesh the streamer spawns now carries why it is an entity, counted per reason — 5.8k resident ADT-doodad submeshes at the Stormwind auction house, 5.3k of them hidden, three prop families the retained pass declines by rule, each riding every per-`Mesh3d` sweep for ~0.4 ms of the alone frame (1947)
- a raid's dozen voices cost the mixer ~1.5 ms of every 60 Hz frame: upstream kira spatializes every frame — normalizes, the attenuation curve and a dB→amplitude `powf` per frame per spatial track, two more `powf`s for the volume tween, ~1.7 M `powf`s a second at 48 kHz × 12 tracks for gains that change imperceptibly across a chunk — so kira 0.12.1 is vendored with one patch: the spatial gains and the volume × fade amplitude are evaluated at the chunk's two ends and interpolated across its ~10.7 ms, the granularity FMOD 3, the reference's own mixer, updates them at; nothing a listener can name changes (1945); and rig poses are evaluated in parallel over rigs — each reads its shared clips and writes only its own pose, and a raid's ~400 live rigs were 0.4 ms of one thread walked serially on the main thread's critical path (1947)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- a 40-man raid cost +5 to +8 ms of CPU per frame — a population, not a hot function — and three cuts landed where it was measured: the rig palette upload sends one `write_buffer` per run, bridging the gaps parked bodies leave between live rigs (calls 93 → 3); a built material becomes an asset when something visible binds it, not when a spawner builds it — 9.9k material assets were alive for 36 drawn batches, so a material is realized only once bound and visible, readers that clone before binding doing it themselves through one new function in the engine's wall (172 → 173): a 17× cut in live GPU objects; and a body's batches spawn as material groups — a geared body stood as 12–27 mesh parts, its skin composite drawn as five separate batches, and two visible batches now join when nothing downstream can tell them apart: parts per body 17 → 12, the Stormwind raid leg 17.0–21.0 → 15.65–15.82 `cpu_ms` (1929, 1940)
- the instruments that priced it, all on the probe line: wgpu's own resource census, because a sampled profile put roughly half of the pass encoder in usage trackers sized to the device's live buffer and texture counts — a term that scales with residency, not with what is drawn, and nothing printed the number; the active camera count, since every view is its own pass set; the UI's live texture-identity runs, one 2D draw each, which is the number a UI change moves, not the quad count; and parts and materials per dressed body on the dress census, the ceiling on merging a body's batches read off a real body instead of guessed from a model's batch table (1940)
- the coin kit retires: every remaining money display — quest progress, the mail purse and postage, the trade recipient, the tooltip, the quest log, the trainer's purse and cost — moves onto the reference's `MoneyFrameTemplate`, each seat already matching ours; the progress panel's required-money flow is ported in full — `QuestFrame.lua:168-190` colours the FRAME red/white AND the LABEL black/dark-grey when the quest costs more than you hold, where ours had neither colour; the mail postage takes the reference's name and label, trade and mail issue no explicit update because the frames repaint on their own events, and the trainer's cost label is shown or hidden with its `local COSTS_LABEL` shadow deleted; with the trainer converted the hand-built coin trios have no consumer, so their templates and the `BenillaMoney_*` helpers are deleted and `MoneyFrame.xml` is the only money kit (1937)
- the crowd-control ladder's aura exemption exists, and it is a join between two `Spell.dbc` records — each arm asks whether the spell being cast is itself an immunity covering the aura that blocks it (Ice Block cast while stunned), the school arm shifting because `School` is an index and `EffectMiscValue` a mask; what the scan mostly decides is WHICH message a refusal carries — a blocking aura the cast does not counter renames the reason to "Can't do that while %s" (1941); bevy_audio leaves the build — its plugin had been disabled since 0530, but disabling a plugin does not remove a crate, and rodio, a second cpal and an Ogg/Vorbis decoder were still linked into a client that mixes through kira (1932); and cpal has four stream errors, not two — a routine ALSA underrun was treated as "the device is gone", tearing the stream down on every hiccup; now only a lost or invalidated device rebuilds (1939)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- a 40-man raid cost +5 to +8 ms of CPU per frame — a population, not a hot function — and three cuts landed where it was measured: the rig palette upload sends one `write_buffer` per run, bridging the gaps parked bodies leave between live rigs (calls 93 → 3); a built material becomes an asset when something visible binds it, not when a spawner builds it — 9.9k material assets were alive for 36 drawn batches, so a material is realized only once bound and visible, readers that clone before binding doing it themselves through one new function in the engine's wall (172 → 173): a 17× cut in live GPU objects; and a body's batches spawn as material groups — a geared body stood as 12–27 mesh parts, its skin composite drawn as five separate batches, and two visible batches now join when nothing downstream can tell them apart: parts per body 17 → 12, the Stormwind raid leg 17.0–21.0 → 15.65–15.82 `cpu_ms` (1929, 1940)
- the instruments that priced it, all on the probe line: wgpu's own resource census, because a sampled profile put roughly half of the pass encoder in usage trackers sized to the device's live buffer and texture counts — a term that scales with residency, not with what is drawn, and nothing printed the number; the active camera count, since every view is its own pass set; the UI's live texture-identity runs, one 2D draw each, which is the number a UI change moves, not the quad count; and parts and materials per dressed body on the dress census, the ceiling on merging a body's batches read off a real body instead of guessed from a model's batch table (1940)
- the coin kit retires: every remaining money display — quest progress, the mail purse and postage, the trade recipient, the tooltip, the quest log, the trainer's purse and cost — moves onto the reference's `MoneyFrameTemplate`, each seat already matching ours; the progress panel's required-money flow is ported in full — `QuestFrame.lua:168-190` colours the FRAME red/white AND the LABEL black/dark-grey when the quest costs more than you hold, where ours had neither colour; the mail postage takes the reference's name and label, trade and mail issue no explicit update because the frames repaint on their own events, and the trainer's cost label is shown or hidden with its `local COSTS_LABEL` shadow deleted; with the trainer converted the hand-built coin trios have no consumer, so their templates and the `BenillaMoney_*` helpers are deleted and `MoneyFrame.xml` is the only money kit (1937)
- the crowd-control ladder's aura exemption exists, and it is a join between two `Spell.dbc` records — each arm asks whether the spell being cast is itself an immunity covering the aura that blocks it (Ice Block cast while stunned), the school arm shifting because `School` is an index and `EffectMiscValue` a mask; what the scan mostly decides is WHICH message a refusal carries — a blocking aura the cast does not counter renames the reason to "Can't do that while %s" (1941); bevy_audio leaves the build — its plugin had been disabled since 0530, but disabling a plugin does not remove a crate, and rodio, a second cpal and an Ogg/Vorbis decoder were still linked into a client that mixes through kira (1932); and cpal has four stream errors, not two — a routine ALSA underrun was treated as "the device is gone", tearing the stream down on every hiccup; now only a lost or invalidated device rebuilds (1939)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- cinematics play: `SMSG_TRIGGER_CINEMATIC` resolves through `CinematicSequences.dbc` → `CinematicCamera.dbc` → a `Cameras\*.m2` whose camera record is an eye path, a look-at path and a roll — cubic tracks with the full step / linear / Bézier / Hermite dispatch the bone loops collapse to step-or-linear (every shipped fly-by is Bézier), planted at the row's origin and facing (`Rz(+facing)`, checked against vmangos's own waypoint table: 8.5 yd mean on the human path, 154 with the other sign); FOV is per shot (the Undead intro authors 90°, the rest 45°), roll is authored around whole turns, the flight ends at the sequence band's end, a trigger arriving before the loading cover lifts is latched and started when it does, ESC skips through `CinematicFrame`'s own `OnKeyDown` (`StopCinematic` has zero native callers), and every path acks — natural end, ESC, or an unresolvable trigger immediately, as before; `.debug play cinematic <id>` replays any of them
- a cinematic hides the HUD — `UIParent:Hide()` reaches nothing here because the HUD frames are not its children, so the ALT-Z draw-level hide is used and the letterbox moves out of FrameXML onto Bevy UI nodes where that hide cannot reach it, recomputed per frame (bars only wider than 4:3, the picture cropped to 2:1) — and the restore only un-hides a UI this module hid
- a lit particle's normal is world up, one constant per draw, not camera-facing: Onyxia's always-on floor haze read about four times brighter from a level camera than from above because the shader dotted a to-eye normal against a fixed light; the reference uploads row 2 of the view matrix — world +Z carried into eye space — on every vertex and carries the light into the same frame, so the rotation cancels; 400 lit emitters (the waterfalls, chimney and campfire smoke, Westfall dust, Dun Morogh snow) stop brightening and dimming as you turn
- the `play` profile drops ThinLTO — rustc #86049 was intermittently poisoning the incremental link with `_anon.<hash>.llvm.<id>` symbols promoted by an earlier compile (48 unresolved across 7 codegen units; 1065 such symbols → 0 after) — and `benilla-formats` stops taking `image` with default features, which had been compiling a complete AV1 encoder into a crate that writes two PNGs: 499 → 466 workspace crates
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- cinematics play: `SMSG_TRIGGER_CINEMATIC` resolves through `CinematicSequences.dbc` → `CinematicCamera.dbc` → a `Cameras\*.m2` whose camera record is an eye path, a look-at path and a roll — cubic tracks with the full step / linear / Bézier / Hermite dispatch the bone loops collapse to step-or-linear (every shipped fly-by is Bézier), planted at the row's origin and facing (`Rz(+facing)`, checked against vmangos's own waypoint table: 8.5 yd mean on the human path, 154 with the other sign); FOV is per shot (the Undead intro authors 90°, the rest 45°), roll is authored around whole turns, the flight ends at the sequence band's end, a trigger arriving before the loading cover lifts is latched and started when it does, ESC skips through `CinematicFrame`'s own `OnKeyDown` (`StopCinematic` has zero native callers), and every path acks — natural end, ESC, or an unresolvable trigger immediately, as before; `.debug play cinematic <id>` replays any of them
- a cinematic hides the HUD — `UIParent:Hide()` reaches nothing here because the HUD frames are not its children, so the ALT-Z draw-level hide is used and the letterbox moves out of FrameXML onto Bevy UI nodes where that hide cannot reach it, recomputed per frame (bars only wider than 4:3, the picture cropped to 2:1) — and the restore only un-hides a UI this module hid
- a lit particle's normal is world up, one constant per draw, not camera-facing: Onyxia's always-on floor haze read about four times brighter from a level camera than from above because the shader dotted a to-eye normal against a fixed light; the reference uploads row 2 of the view matrix — world +Z carried into eye space — on every vertex and carries the light into the same frame, so the rotation cancels; 400 lit emitters (the waterfalls, chimney and campfire smoke, Westfall dust, Dun Morogh snow) stop brightening and dimming as you turn
- the `play` profile drops ThinLTO — rustc #86049 was intermittently poisoning the incremental link with `_anon.<hash>.llvm.<id>` symbols promoted by an earlier compile (48 unresolved across 7 codegen units; 1065 such symbols → 0 after) — and `benilla-formats` stops taking `image` with default features, which had been compiling a complete AV1 encoder into a crate that writes two PNGs: 499 → 466 workspace crates
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- water and lava keep time in the shader: the 24 Hz CPU tick that re-uploaded every liquid material is gone, −0.28 ms at the Stormwind gates, and lava scrolls smoothly instead of stepping
- static transform tracking is pinned on — the adaptive check spent two full scans a frame to say yes in a world where movers are 1–3% of the rows
- WOW_MEGA_STATIC merges static geometry per material and 133-yard cell — an experiment lever, priced at −1.15 ms CPU for +0.3 ms GPU at the Stormwind gates
- WOW_GPU_MS times the whole frame on the GPU and reads 2.2 ms against 17.4 CPU at the Stormwind gates: the frame is CPU-bound, so the work ahead is on the CPU side
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- water and lava keep time in the shader: the 24 Hz CPU tick that re-uploaded every liquid material is gone, −0.28 ms at the Stormwind gates, and lava scrolls smoothly instead of stepping
- static transform tracking is pinned on — the adaptive check spent two full scans a frame to say yes in a world where movers are 1–3% of the rows
- WOW_MEGA_STATIC merges static geometry per material and 133-yard cell — an experiment lever, priced at −1.15 ms CPU for +0.3 ms GPU at the Stormwind gates
- WOW_GPU_MS times the whole frame on the GPU and reads 2.2 ms against 17.4 CPU at the Stormwind gates: the frame is CPU-bound, so the work ahead is on the CPU side
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- an animated doodad is culled by its authored box, not its bind pose — the one-yard bird flying a 64-yard circuit stops vanishing with the camera angle — and the fix survives the engine's own bound recompute
- the logon proof answers the server's version challenge instead of sending zeros, so a strict realmd stops rejecting benilla as a modified client
- two instruments land: a frame-by-frame visibility trace for one model, and a probe that classifies whether a realmd enforces the version check
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- an animated doodad is culled by its authored box, not its bind pose — the one-yard bird flying a 64-yard circuit stops vanishing with the camera angle — and the fix survives the engine's own bound recompute
- the logon proof answers the server's version challenge instead of sending zeros, so a strict realmd stops rejecting benilla as a modified client
- two instruments land: a frame-by-frame visibility trace for one model, and a probe that classifies whether a realmd enforces the version check
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Lua 5.0's generic-for and nested long strings are restored in a forked VM — 60 corpus addons stopped dying on one deleted opcode
- a day of demand-driven carving takes the 218-addon corpus from 24 addons surviving a session start to 114, with 144 loading without a single error
- Bagnon draws a real bag and answers real gestures — hover tooltip, pickup, use, drag — because our frames wear the reference's names, and the reference's own ContainerFrame.lua runs off the player's install instead of being transcribed
- the physics broad phase leaves: nothing in the client reads a contact, and it was measured at up to 41% of main-thread samples in real sessions
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Lua 5.0's generic-for and nested long strings are restored in a forked VM — 60 corpus addons stopped dying on one deleted opcode
- a day of demand-driven carving takes the 218-addon corpus from 24 addons surviving a session start to 114, with 144 loading without a single error
- Bagnon draws a real bag and answers real gestures — hover tooltip, pickup, use, drag — because our frames wear the reference's names, and the reference's own ContainerFrame.lua runs off the player's install instead of being transcribed
- the physics broad phase leaves: nothing in the client reads a contact, and it was measured at up to 41% of main-thread samples in real sessions
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- timed quests get their countdown — a Quest Timers box under the minimap and a Time Remaining line in the quest log, kept against the server's own clock
- interiors go honest: footsteps stop crunching the terrain's snow through a building's floor, and tavern voices are authored dry, so the fake cave echo is gone
- snow is the reference's again — point-sprite flakes at the era's pixel size, and a spawn slab that leans into your run so a runner no longer outruns the field
- Stormwind stops drawing Elwynn through its walls, a looted quest crate's lid settles, cold zones puff visible breath, and the collect-quest toast fires only on a real advance
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- timed quests get their countdown — a Quest Timers box under the minimap and a Time Remaining line in the quest log, kept against the server's own clock
- interiors go honest: footsteps stop crunching the terrain's snow through a building's floor, and tavern voices are authored dry, so the fake cave echo is gone
- snow is the reference's again — point-sprite flakes at the era's pixel size, and a spawn slab that leans into your run so a runner no longer outruns the field
- Stormwind stops drawing Elwynn through its walls, a looted quest crate's lid settles, cold zones puff visible breath, and the collect-quest toast fires only on a real advance
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Entering the world used to show an uncovered, movable world before the screen
came up — seconds of it on a slower machine ("like the Stormwind intro
video", as one report with video put it). The screen also cleared onto bare
terrain with the buildings still streaming in, and could hang forever if you
arrived swimming or flying. It has a real lifecycle now: it rises at the
entry edges (character pick, transfer, teleport), clears only when the scene
is actually presentable — terrain, buildings, props and their collision all
resident — and its physics hold releases in every movement mode, not just
walking. The progress bar counts buildings too.
- A far teleport flashed a frame of the destination before the screen rose.
The "swap never flashes" guarantee existed only in comments — a streamer
rework had silently dropped the scheduling constraint that enforced it; it
is structural now. And logout got its missing cover: the original's plain
black cut, not a frame of torn-down world.
- Animation parking now governs everything with a rig. ~350 animated objects
(braziers, doors, chests) were sampling their loops off-screen forever, and
a patrolling parked mob kept writing pose data nothing could read — ~620 rig
refreshes a frame became 7, and creatures in rooms the camera can't reach
through any portal park too.
- The engine's own tax got measured: bare Bevy spends ~3.9 CPU ms a frame on
this machine drawing nothing at all, which resets what "fast enough"
honestly means for everything built on top of it. A shipping build profile
(fat LTO) landed alongside — worth about −1 ms at the pin, kept out of
everyday builds.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
gromandClaude Opus 5