Post-merge Codex review of the bend-curve PRs (#531/#532) surfaced edge cases:
- GP8 (#531 P2): bnv timing used rn.sustain, which is zeroed for notes <= 0.2s,
so short GP8 bends kept the scalar bn but lost bt/bnv. Use the beat duration
`dur` (matching the GP5 path) so the curve survives.
- 2D highway (#532 P2): bnvNormalizedPoints mapped x over the curve's own t-range
[first,last] instead of the note span, so curves not starting at 0 / ending at
sus were time-distorted. Now maps over [0, sus] (clamped), with a curve-span
fallback when sus<=0 (existing no-sus callers unaffected).
- 3D highway (#532 P3): the sustain ribbon + bend chevron were gated on bn>0, so a
note carrying an authoritative bnv with bn==0 drew no ribbon/marker. Both now
also fire on bnv presence; chevron steps derived from max(bn, bnv peak).
Codex-reviewed: clean (no findings). +1 JS test (sus-relative mapping + fallback).
JS 8/8, 250 core GP/song tests pass.
NB: GP8's short-bend path still lacks a dedicated synthetic-GPIF fixture (same gap
as the GP8 offset-prop-names P3) — _gpx_bend_shape units cover the function; the
fix is the one-line caller change.
Part of got-feedback/feedback#334.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
PR-B of the bend-shape feature (feedpak §6.2.1). Both highways drew a bend
from the scalar `bn` only; now they trace the authoritative `bnv` curve
([{t, v}]) when present and fall back to the `bn` arc/envelope otherwise.
2D (static/highway.js drawNote): when a note carries `bnv`, draw the real
shape as a contour above the gem (round-trip rises then falls, pre-bend
starts high, release descends — `bt` is implicit in the point shape), with
an arrowhead only when the gesture ends rising. `bnvNormalizedPoints` maps
{t,v} to a 0..1 x span. The scalar-arrow path is preserved unchanged as the
fallback; the peak label is unchanged.
3D (plugins/highway_3d/screen.js): `bnvSampleAt` linearly interpolates the
curve (clamped to its endpoints) and `bendSemisAtTime` samples it when
present, else keeps the synthetic rise→hold→release envelope from `bn`. The
chevron count still comes from the peak. Fixed a stale-scratch hazard: the
reused `_scrChordNote` now resets `bnv`/`bt` (omit-when-default) after
Object.assign, mirroring the existing `fhm` reset, so a chord note without a
curve can't inherit the previous note's contour.
Render-only — no wire/schema change. Pure helpers covered by
tests/js/highway_bend_curve.test.js (interp, clamping, round-trip,
degenerate/empty); node --check passes on both files; full tests/js green.
Part of got-feedback/feedback#334
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The 2D highway drew pitched slides (sl) but ignored unpitched slides (slu) —
drawNote read only opts.sl. The 3D highway already renders both (slideTrailEnd).
drawNote now draws slu as a dashed diagonal with no arrowhead (no definite
target pitch), keeping the solid arrow+arrowhead for pitched sl. The two are
mutually exclusive in the data. Chord notes flow through the same drawNote, so
chord-note unpitched slides are covered too.
Also fixes a latent pre-existing bug flagged in review: `opts?.sl || -1`
discarded a pitched slide-to-open (sl: 0); now `?? -1` preserves fret-0 targets
and keeps pitched precedence.
Codex-reviewed: no P1/P2; dash state reset on all paths, no pitched-slide
regression. node --check clean.
Closes#336. Part of got-feedback/feedback#334.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Both the v3 index.html and live-guitar-tone-source.js had an extra 'd'
in the brand name ('feed[dB]ack' instead of 'fee[dB]ack') in the guitar
tone source selector labels and help text.
Rename the user-visible label of the HOME-group nav entry (and the matching
"Open Shop →" button on the progress page) from "Shop" to "Unlockables".
Internal identifiers (nav key 'shop', screen id v3-shop, window.v3Shop) are
left unchanged so wiring/state are unaffected.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Collapse the per-plugin sidebar list down to the single "Plugins" entry
(the gallery is the one entry point for general plugins) and give two
bundled plugins their own first-class sidebar slots instead:
- SlopScale (manifest label "SlopScale - Practice"), directly under FeedBarcade
- Rig Builder, directly after the Library group
Both are driven by a PROMOTED_PLUGINS table: each slot is anchored after
a nav key and filled by renderPromotedNav() only when the plugin is
present in /api/plugins, so an absent bundle shows nothing rather than a
dead entry that bounces to the Plugins screen. The visible label uses the
plugin's own manifest nav.label (escaped), falling back to the static NAV
label.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* sloppak: read cover without unpacking + serialize/cap zip unpacks
Album art for a zip-form sloppak was served by resolve_source_dir(), which
unpacks the ENTIRE archive (stems included, ~30 MB) to disk just to read
cover.jpg. On the library grid that meant a full extraction per card on scroll.
- read_cover_bytes(): opens only the cover member from the zip (or reads the
file for dir-form), with zip-slip guarding. ~4 ms vs a full unpack.
- resolve_source_dir(): per-file lock + bounded global semaphore so concurrent
callers don't rmtree + re-extract the same dest at once (a race), and a burst
can't saturate disk/CPU. 8 concurrent calls now dedupe to 1 unpack.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* server: serve sloppak art via read_cover_bytes + cache album-art responses
- get_song_art() sloppak branch now reads the cover directly (no full unpack),
off-thread via asyncio.to_thread.
- All art responses carry Cache-Control: public, max-age=86400. URLs are already
cache-busted with ?v=<mtime>, so the browser stops re-fetching every cover on
scroll-back; day bound self-heals any URL missing ?v.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* v3 library: lazy-load + async-decode card cover images
The grid (24 cards/page) and artist-row thumbnails emitted plain <img> with no
loading hint, so a whole page of covers fetched + decoded at once on each
scroll batch. Add loading="lazy" decoding="async" to defer off-screen fetches
and keep image decode off the main thread.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* address Codex review: zip cover normalization + correct art revalidation
Findings from the preflight Codex passes:
- sloppak.read_cover_bytes (zip form) read the raw manifest cover string via
zf.read(), so a non-canonical name like './cover.jpg' or 'art/../cover.jpg'
404'd. Normalize via safe_join → relative member; reject escape and the
degenerate root-collapse case ('.', 'subdir/..') like _unpack_zip does.
- Album-art caching is correctness-first: Cache-Control: no-cache plus a strong
validator, with real conditional handling (Starlette FileResponse emits an
ETag but doesn't evaluate If-None-Match). All three art paths route through
_art_conditional/_file_art_response → bodyless 304 on a matching validator.
A long immutable max-age was rejected because the frontend ?v=<mtime> buster
is only second-resolution and would pin a same-second rewrite.
- The sloppak cover is validated by CONTENT (sha1 of the bytes), not a stat:
a dir-form sloppak edited in place changes the cover file's mtime but not the
directory's, so a dir-stat ETag could emit a stale 304. Content hashing is
correct for both dir- and zip-form. get_song_art gained an optional request
(internal get_art caller passes none — safe).
Adds tests/test_sloppak_cover_art.py pinning read_cover_bytes (canonical,
non-canonical, degenerate/escape, dir/zip, webp) and the endpoint's 304 contract
incl. the dir-form in-place-edit no-stale-304 regression.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The highway showed two overlapping note-detection scoreboards at once: the
core v3 live-performance HUD (#v3-live-performance-hud) and the note_detect
plugin's own HUD (.nd-hud). Both auto-render off the same note:hit/note:miss
events and neither suppressed the other.
Add a Settings → Visualization "Scoreboard" selector (Streak / Detailed /
Off, default Streak) backed by a single source of truth on
<html data-scoreboard>. CSS shows exactly one:
core (default) → core HUD; hide .nd-hud
detailed → .nd-hud; hide the core HUD
off → hide both
CSS-based suppression keys the default off ":not(detailed):not(off)", so the
correct HUD is right even before the pref script runs (no flash) and it
robustly hides any .nd-hud regardless of how many note_detect instances load.
Detection itself is untouched — only the duplicate scoreboard panel is hidden.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* style(tour): align tour engine + Shepherd bubbles to the v3 fb-* palette
The tour/help engine shipped its own indigo/blue dark palette (#181830 / #4080e0)
that predates the v3 fee[dB]ack tokens, and the spotlight bubbles themselves used
the vendored Shepherd LIGHT default (white card, black text) — both clashed with
the navy/sky v3 UI behind them.
- Recolor the "?" menu button, popover and first-visit toast to the fb-* tokens
(card #1e293b, primary #0ea5e9, border #334155, text #f8fafc/#94a3b8, gold
#e8c040 unchanged).
- Add a dark .shepherd-* override block (loads after the vendored shepherd.css,
which is left pristine for upgrades): dark bubble + arrow, fb-primary Next/Done
button, slate secondary button, fb text scale, and bump the modal dim to 0.6 to
match the onboarding overlay.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(tour-engine): let client/core tours register into the consolidated menu
The tour engine only listed server-discovered plugins (those with a tour.json,
populated from /api/plugins) in the "?" menu, and always prompted unseen relevant
tours via the toast + button pulse. Generalize register() so a core/client-owned
tour can participate:
- `name` registers the tour into the menu catalog (_tourPlugins) so it shows in
the "?" menu even without a server plugin; never clobbers a real plugin entry.
- `autoPrompt:false` opts the tour OUT of the unseen toast + pulse (for tours
driven programmatically by their owner), while still listing + running on
demand. _unseenRelevant honours it.
Both options are additive and default to the prior behaviour.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* chore(v3): add stable tour anchors to home cards + instrument badge
Give the first-run home tour stable spotlight targets: #v3-hero on the hero
panel and data-tour="continue" on the three continue/pick/browse card variants
(dashboard.js), and #v3-instrument-wrap on the topbar instrument selector
(badges.js, mirroring the existing #v3-tuner-wrap). The other targets (audio
routing, tuner, profile, sidebar nav) already had stable ids.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(onboarding): first-run home tour (spotlight coach marks)
After a genuine onboarding completion, dim the home page and spotlight one card
at a time with an explanatory bubble + Next, reusing the shared tour engine
(Shepherd). 7 stops: Hero/Start Playing → Continue/Pick → Instrument selector →
Tuner → Audio Routing → Profile → Sidebar nav. Auto-runs once; replayable
forever from the "?" tour menu as "Welcome tour".
- New static/v3/onboarding-tour.js: registers the spotlight tour (screens:
['v3-home'], name "Welcome tour", autoPrompt:false) and exposes startFirstRun(),
gated on the engine's seen/dismissed state so it never repeats; loaded after
tour-engine.js + dashboard.js.
- profile.js finish(): trigger startFirstRun() only on a real onboarding
completion (!editing) — a later profile edit must not relaunch it.
Verified headlessly (native core + Playwright): all 7 anchors resolve, the
spotlight advances one bubble at a time in the v3 dark theme, completion marks
seen, startFirstRun is once-only, and the "?" menu lists "Welcome tour".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(capabilities): clear the handler timeout timer once the race settles
Codex round-6: _withTimeout raced the handler promise against a bare setTimeout
but never cleared it, so a handler that resolves first leaves the timer alive
until it fires. Harmless at 250ms, but the new 15s MIDI permission-command
overrides (discover/open-source) kept the event loop alive ~15s after every
successful call (and the test process hung that long) and could accumulate
delayed callbacks across repeated scans. Capture the timer and clearTimeout it in
a .finally on the race. (Domain/capabilities tests now finish in ~0.1s, not 15s.)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(midi-input): give the built-in Web-MIDI provider a distinct participant id
Codex round-7: the built-in Web-MIDI provider registered with participantId
'core.midi-input' — the same id as the domain owner. unregisterProvider()
unregisters the provider's participant, so a provider swap/hot-reload would tear
down the domain OWNER too, leaving midi-input with no owner for later commands.
Register the provider as 'core.midi-input.web-midi'.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(onboarding): don't start the home tour when launching the diagnostic
Codex round-7: on the final onboarding step, "Play it now" calls finish() (which
started the home tour) and THEN playSong(target). startFirstRun() navigated to
v3-home and scheduled the tour, then playSong switched to the player — so the
tour spotlighted hidden home elements / stole focus from the diagnostic. Gate the
tour on a launchingSong flag (passed by the "Play it now" path); the Skip path
stays on home, so the tour still runs there.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(capabilities): add core-owned midi-input control-plane domain (#873, #880)
The MIDI analog of audio-input: a core-owned provider-coordinator over MIDI
device discovery, selection, and shared open/close sessions. Separate from
audio-input (whose source/open contract is audio-frame-centric) and not owned
by any feature plugin, so the device-access boundary outlives the input-setup
wizard. `discover` is the Web-MIDI permission boundary; selection persists by
redaction-safe logicalSourceKey; diagnostics redact device labels and never
carry raw MIDI messages.
- static/capabilities/midi-input.js + load-order wiring in both shells
- spec 012 + capability-domains/safety-matrix entries; midi-control narrowed
to mappings-only (split)
- 9 domain tests against the real runtime
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(input_setup): bundled plugin owning input-calibration + Web-MIDI provider (#872)
Bundled core plugin that supplies the Web-MIDI source provider to the core
midi-input domain, owns the input-calibration workflow domain (run/status/
inspect), and renders the per-instrument wizard (guitar/bass -> audio-input +
note_detect; keys/drums -> midi-input live note/pad test). Idempotent
hydration; redaction-safe. .gitignore allowlists the in-tree plugin.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(onboarding): input-device setup step between paths and calibration (#874)
After instrument-path selection and before the note-detect calibration
challenge, dispatch input-calibration `run` (fire-and-launch) and await the
`calibration-done` event. Fail-soft: a non-handled outcome (plugin/runtime
absent) advances immediately so onboarding can never be stranded.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(midi-input): ship a built-in Web-MIDI provider in the core domain
Move the Web-MIDI source provider out of input_setup and into the core
midi-input domain so every consumer (piano, drums, input_setup) gets MIDI
devices from the domain without depending on any one plugin being loaded.
input_setup is now a pure midi-input requester (manifest role updated).
Prepares piano/drums full consumption (#876/#877). +1 domain test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(input_setup): Settings panel to re-run input setup (#878)
Adds a settings.html with a "Set up input devices" button (window
._inputSetupRelaunch) that re-runs the wizard for the player's selected
instrument paths (from /api/progression; falls back to all instruments). Makes
the calibration wizard re-launchable outside first-run onboarding.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(midi-control): formalize the midi-input/midi-control split (#882)
Narrow the reserved midi-control domain to mappings ONLY (CC/pitchbend/note →
action routing), consuming the delivered midi-input domain for device access.
Adds spec 013 defining the contract + intended consumers (feedback-plugin-midi,
drums learn-mode), updates the safety-matrix row, and cross-references it from
capability-domains. Per governance, midi-control stays RESERVED (no runtime
domain) until a concrete mapping consumer + tests exist.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(onboarding): wait for input_setup before the calibration step (#874)
The input-setup wizard is a mandatory onboarding step, but plugins load
asynchronously — in the desktop app (40+ plugins) the user can reach path
selection and click Next before input_setup has registered its
input-calibration owner. The dispatch then got a no-owner outcome and
onboarding fell through to the calibration challenge, silently skipping the
wizard. Now wait (bounded, 8s) for the plugin's public global before
dispatching; fall through only if it never appears. Race-verified.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(onboarding): add Song directory step after name+avatar (#874)
New first-run step (now step 2 of 4: name+avatar → song directory → paths →
calibration challenge) where the player sets their songs folder, fixing the
"folder not configured" error on a fresh install. Saves to settings (dlc_dir)
and kicks a library scan; persists to config.json so it survives restart. A
native folder picker is offered on desktop (window.slopsmithDesktop
.pickDirectory); web users type/paste the path. "Skip for now" leaves it
unconfigured (settable later in Settings).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(input_setup): filter MIDI entries out of the guitar audio-input picker (#876)
Other plugins export pseudonymized MIDI sources ('midi-input-N') into the
audio-input domain; they aren't audio inputs and the cryptic labels confused
the guitar/bass device dropdown. Filter them out so only real audio inputs show.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(input_setup): de-dupe audio input picker entries (#876)
The desktop audio engine enumerates the same device under multiple driver
types, so the guitar audio-input dropdown showed repeated entries. De-dupe by
display label (paired with the desktop fix that surfaces real device names).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(midi-input): drop vanished devices on re-discovery; reset setup confirm on switch
Codex preflight findings:
- midi-input domain `_discover()` only upserted enumerated sources, so an
unplugged device (statechange re-discovery) lingered in list-sources and later
open/select hit stale state. Reconcile each provider's sources against the
fresh enumeration (close any live session, keep the selectedKey preference).
- input_setup MIDI panel left "Continue" enabled (and the instrument marked
done) after switching the device selection following a prior hit. Reset the
waiting state + disable Continue on every selection change, and discard a
stale open if the selection changed mid-await. +1 reconciliation test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(midi-input): coalesce concurrent opens; commit shown audio source pre-calibration
Codex re-review (round 2):
- midi-input domain: two concurrent open-source calls for the same source both
passed the `sessions.get` guard and each called provider.open(), which for the
built-in Web-MIDI provider overwrites the shared input.onmidimessage handler
and orphans the earlier session — leaving the device silent. Coalesce in-flight
opens onto one provider session (await the pending open, adopt its session;
re-check after open and release a redundant handle if another open won). +test.
- input_setup: the guitar/bass audio <select> shows its first option by default
but fires no `change`, so on a first run with nothing selected, audio-input was
never told before launchCalibration(). Commit the shown option on render.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(midi-input): longer timeout for MIDI permission commands; stale-open guard in wizard
Codex re-review (round 3):
- The advertised command surface ran `discover`/`open-source` through the 250 ms
default handler timeout, but those front a real Web-MIDI permission prompt /
device open that commonly takes longer, so dispatch returned `failed` while the
operation was still completing. Add per-(capability,command) timeout overrides
(15 s for those two), folding the existing audio-mix special-case into the same
table so both the command() and dispatch() paths honor it.
- input_setup MIDI panel: openSelected() compared the mutable shared `activeKey`
after its awaits, so a device switch mid-open could bind the old device's
listener / close the wrong session. Capture the requested key in a local and
use a generation guard to discard a superseded open.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(onboarding): detect 200-with-error song-dir saves; close MIDI session on skip
Codex re-review (round 4):
- /api/settings reports an invalid folder as a 200 response with an `error` body
(a bare dict return, not a non-2xx status), so saveSongDir's res.ok-only check
treated the failure as success and advanced onboarding without saving. Parse
the body and throw on `error` too.
- input_setup: the opened MIDI test session was only closed on the Continue
button, so using the generic "Skip for now" after scanning leaked the listener
and kept the Web-MIDI input live. Run teardown on every panel exit via a
per-panel cleanup hook invoked by advance().
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(input_setup): don't hard-code Web MIDI in the device wizard
Codex re-review (round 5): the MIDI panel gated availability on
navigator.requestMIDIAccess and filtered sources to providerId === 'web-midi',
which defeats the midi-input domain's provider-coordinator abstraction — a
native/desktop MIDI adapter registered with the domain would be reported
unavailable and hidden from the picker. Gate availability on the domain
(window.slopsmith.midiInput) and show every source it surfaces.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The first-run calibration prompt (profile onboarding step 3) and the
Progress-screen calibration card both told the user to play the
"Slopsmith Diagnostic". Update the visible copy to "fee[dB]ack
Diagnostic". Text-only; the diagnostic is matched functionally by the
is_diagnostic flag + filename, not by this label, so no behavior change.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: topkoa <topkoa@gmail.com>
Reword comments/docstrings/strings and rename identifiers that referenced
the external game and its file formats:
- format-id "psarc" -> "archive"; local vars psarc_path -> song_path,
psarc_base -> tone_base
- lyrics provenance value "sng" -> "notechart" (legacy "sng" still accepted)
- highway_3d fret-ghost scope value "rocksmith" -> "chords" (invalid/legacy
values fall back to the default, preserving behaviour)
- neutralise references in prose, test names/data, .gitattributes and docs
No functional change beyond the renamed identifiers; all Python compiles.
