mirror of
https://github.com/got-feedBack/feedBack.git
synced 2026-07-25 22:31:48 +00:00
188bdaa837
364 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
topkoa
|
188bdaa837 |
feat(panes)!: move the real element, instead of rebuilding it
The first cut of this got the model wrong. A pane was a SECOND
implementation of the plugin's panel — its own sliders, its own styling,
driven over a cross-realm bridge (ctx, a state store, capability RPC,
mirrorGlobal, a stream sampler). Popping out gave you something that
resembled the panel you popped, and every feature it did not reimplement
(presets, tabs, EQ, language) was simply gone.
What a user wants from "pop this out" is the thing they popped out.
So: MOVE THE REAL ELEMENT. Same-origin windows can adopt each other's
nodes, and an adopted node keeps its event listeners and its closures.
The panel goes on running the plugin's own code, against the plugin's own
state, in the plugin's own realm — it is merely being DISPLAYED in another
window. Copy the app's stylesheets into that window and it looks identical
too, because it is identical.
The plugin's side collapses to two lines:
feedBack.panes.register({ id, title, element: () => panelEl });
feedBack.panes.attachChip(panelEl, id);
and everything comes along: the CSS, the listeners, the presets, the
state. Nothing to keep in step, because there is no second copy.
Deleted, all of it now pointless: pane-bridge (ctx + transports), pane-hub
(the cross-realm server), pane-runtime (the pane realm's boot), pane-streams
(the rAF sampler that existed because an AnalyserNode can't cross a window),
pane-mirror (mirrorGlobal), pane-plugins + the manifest `panes[]` key and its
server-side validation, panes.state(), and both built-in demo panes. ~1200
lines. None of it was wrong — it was all correct machinery for the wrong
problem.
Consequences worth knowing:
- The window MUST be opened by the renderer with window.open(), not by the
desktop's main process: a window we did not open gives this realm no handle
to its document, and without the handle there is nothing to adopt into.
Electron turns the same-origin window.open() into a real BrowserWindow
anyway (setWindowOpenHandler → 'allow'), so we get the OS window AND the
live DOM link. The desktop side finds it by frame name.
- `.fb-paned` neutralises PLACEMENT only (position/inset/width/z-index/shadow).
A plugin panel is nearly always a fixed overlay pinned to a corner of the
app; alone in a 380px window that positioning is nonsense. Colours, borders,
padding, fonts and the panel's own internal layout are untouched — the whole
promise is that what you popped out is what you get.
- The element is returned to its EXACT home on dock: same parent, same position
among its siblings.
- The plugin's code still runs in the main window. So a document.body
.appendChild() inside a panel (a tooltip, a popover) lands in the main
window, not the pane — anchor to the panel instead. And a continuously
animating panel may run slowly while the main window is backgrounded, since
its rAF lives there. Both documented.
Signed-off-by: topkoa <topkoa@gmail.com>
|
||
|
topkoa
|
330995588c |
feat(panes): panes.state(id) — let a plugin apply its own pane's values
mirrorGlobal covers the case where a pane drives a plain global that some renderer reads each frame. It does not cover the far more common one: a plugin whose MAIN-realm code is the authority — it clamps, it persists, it emits events, it owns the audio graph or the camera rig — and which must therefore APPLY the pane's values itself rather than have core splat them somewhere. Camera Director is the case that forced this. Its brain is the sole writer of the camera store, the sole broadcaster on splitscreen's channel, and the only thing that clamps an axis to its legal range. A pane cannot write window.__h3dCamCtl behind its back without desynchronising its presets, its persistence, and the panel's own sliders — and running the brain inside the pane realm would make it a SECOND store writer and a second broadcaster, racing the real one. So: `panes.state(id)` hands the main realm the open pane's store (get/set/all/subscribe). A plugin seeds it on `panes:opened`, subscribes, and applies what comes back through its own API. The pane stays realm-agnostic — it only ever touches ctx.state — and the plugin stays the single source of truth. For that to work, the hub now broadcasts EVERY change to the store, not just the ones a pane asked for: it subscribes to the store on connect rather than echoing pane-originated writes by hand. A value the plugin clamps or corrects therefore reaches the pane window immediately, and there is exactly one path by which state arrives in a pane — so it cannot drift. Signed-off-by: topkoa <topkoa@gmail.com> |
||
|
topkoa
|
254e26bb3a |
feat(panes): mirrorGlobal, manifest-declared panes, and the plugin docs
Three things a plugin needs before it can actually use panes.
## mirrorGlobal — the camera-director problem
The 3D highways read their free camera from a plain global,
`window.__h3dCamCtl` (highway_3d/FREECAM_BRIDGE.md), once per frame in
_resolveFreeCam(). A camera panel in the main window just writes that
object and the camera moves. A panel in a POP-OUT window cannot:
window.__h3dCamCtl there is a different object in a different realm, and
writing it moves nothing.
So a pane declares one field — `mirrorGlobal: '__h3dCamCtl'` — and
pane-mirror.js (main realm, where the renderers live) copies that pane's
state onto the global whenever it changes. highway_3d, keys_highway_3d
and drum_highway_3d are NOT modified and do not know panes exist.
The rule that makes it work: MUTATE THE OBJECT, NEVER REPLACE IT. A
renderer may be holding the reference, and swapping in a new object would
leave it reading an orphan. Keys the pane doesn't set are left alone
rather than deleted — the global may carry a renderer's own bookkeeping.
Closing the pane deliberately leaves the global as-is: closing the camera
panel should not snap the camera back to a default, which is exactly what
happens today (nobody clears __h3dCamCtl).
## Manifest-declared panes
"panes": [{ "id": "camera_director", "title": "Camera Director",
"script": "panes/camera.js", "mirrorGlobal": "__h3dCamCtl" }]
Declaring a pane beats calling panes.register() from screen.js because it
becomes openable FROM THE RAIL OR THE TRAY WITHOUT THE PLUGIN'S SCREEN
EVER HAVING BEEN VISITED — core registers a stub from the manifest and
fetches the script only when the user opens it. A pane you can only reach
by first navigating to the screen it was meant to replace is not much of a
pane.
The script sets `window.feedBackPane_<id> = { mount, unmount }`, mirroring
the existing window.feedBackViz_<id> convention, and the SAME file is what
a pop-out window loads in its own realm.
`script` is validated as a relpath under the plugin's src/ and served
through the sandboxed /api/plugins/<id>/src/ route — the containment rule
`styles` already has for assets/. Traversal, absolute paths, drive letters,
backslashes and non-.js are rejected; a bad entry is dropped with a warning
rather than failing the whole plugin, because one malformed pane should not
cost the user a working plugin.
Note the projection is written TWICE — _nav_entry() and the /api/plugins
route re-project independently — so panes had to be added to both, plus the
pending branch (a pane can be opened while its plugin is still installing
deps; the script is fetched on open, not at discovery).
## docs/plugin-panes.md
The contract, and the one rule it all hangs on: mount(root, ctx) runs in a
realm that may not have the app in it. Everything comes through ctx, or the
pane works docked and silently dies popped out.
Verified: manifest validation rejects ../.., C:\, non-.js, dupes and
missing fields while passing a good entry; /api/plugins projects panes[] for
all 20 plugins. mirrorGlobal mutates the global IN PLACE — a reference held
the way _resolveFreeCam holds it sees the change, and a renderer's own field
on that object survives — both for a local write and for a write arriving
over the channel from a pop-out realm.
pytest: 2401 passed, 8 failed — all 8 reproduce on a clean main (including
the one in tests/test_plugins.py) and are unrelated.
Signed-off-by: topkoa <topkoa@gmail.com>
|
||
|
topkoa
|
d508380532 |
feat(panes): desktop host — real windows and the system tray
Registers a `desktop` pane host at priority 20, above the browser pop-up host (10) and the dock (0), whenever the Electron bridge exposes feedBackDesktop.panes. A popped-out pane then gets a real BrowserWindow: it remembers where you put it, can float above everything, minimizes to the system tray, and appears in the tray's menu. In a plain browser — or on an older desktop build that predates the bridge — this file registers nothing and the browser pop-up host handles detach exactly as before. Nothing else in the pane system changes. That is what the host registry is for. Two things only this realm can decide, so it owns them: - The user closed a pane window (or it crashed). Close the pane, or the dialog its pop-out chip hid never comes back and the user is left with no way to reach their own UI. - The tray asked to toggle a pane it has no window for. Main cannot know what opening one means — the pane might belong in the dock — so it asks. Unlike a browser pop-up, this host needs no user gesture, so it sets autoRestore: true — a pane you left popped out comes back popped out, where you left it, on the next launch. Pairs with got-feedback/feedBack-desktop#103. Signed-off-by: topkoa <topkoa@gmail.com> |
||
|
topkoa
|
fefb9051a4 |
feat(panes): pop-out windows — the pane realm, hub, and remote transport
A pane can now leave the main window entirely. Same `mount(root, ctx)`,
same file, different JS realm — which is what the ctx-only contract in the
previous commit was for.
## A purpose-built document, not the app shell with a flag on it
`GET /pane` serves static/panes/pane.html: the bridge, the runtime, and
the pane's own script. No highway, no library, no v3 shell, no <audio>,
no Tailwind.
The splitscreen follower takes the other road — it reloads the whole app
at `/?ssFollower=1` and hides what it doesn't want — and pays for it with
an anti-flash block that must run before any script parses (index.html),
bail-outs in app.js and shell.js, and ~40 lines of CSS hiding core
elements by id. It loads the entire app to throw it away. A pane window
has nothing to throw away, so it boots in milliseconds and there is
nothing to flash.
The cost is that `window.feedBack` in a pane realm is a deliberate,
documented SUBSET. The runtime installs exactly what a pane is promised —
`panes.register`, and the no-op chip/dock calls a shared script may make
at load — so a pane reaching for something it was never given fails
loudly at authoring time instead of subtly at runtime.
## The channel
BroadcastChannel('feedback-panes'), same origin. This works only because
Electron's setWindowOpenHandler returns `action: 'allow'` for same-origin
URLs: `deny` would push the window to the system browser, a different
Chromium instance, where BroadcastChannel cannot reach it and the pane
would silently never sync. That flag is load-bearing.
hello -> snapshot resync-on-open, always. The snapshot is the only way
the pane realm learns anything.
state main is authoritative. A pane's write is a REQUEST;
main applies it and echoes to every realm, so a
losing write self-corrects instead of splitting brain.
rpc / rpc:reply ctx.call() -> the capability bus, with a 10s deadline.
Without one, a main window that died mid-call leaves
the pane's promise pending forever.
event allowlisted bus events, JSON-safe. A CustomEvent
carrying a DOM node (highway:canvas-replaced does)
would throw on postMessage and take the channel down
for everyone, so detail is round-tripped through JSON.
stream one coalesced message per pane per frame, OVERWRITING
anything not yet flushed. Queueing would build a
backlog: Chromium throttles a backgrounded window, and
the main window is exactly what's backgrounded while
the user looks at the pane.
sub / unsub refcounts the main-realm sampler.
bye both directions.
## The follower clock
The pane extrapolates between broadcasts: anchor + observedRate * elapsed,
capped at 2s. observedRate is learned from the broadcasts themselves
(dt/dwall) so it tracks the speed slider without being told about it, and
seeks/pauses are excluded from the fit — a jump is not a tempo. Capping it
means a dead main window decays into a frozen clock rather than one that
confidently runs away. This is splitscreen's hard-won trick, generalized:
panes just call ctx.playhead().
## Failure modes, all of them
- Main window closes -> `bye {main-closed}` and the pane says so plainly,
rather than showing a frozen playhead that looks live. The host also
closes its windows outright; a pane that cannot be fed should not be on
screen.
- Pane window X'd or crashed -> a `closed` poll reaps it (a crashed
renderer never sends `bye`), the pane closes, and the chip's dialog comes
back. Without this the user's dialog stays hidden with no way back.
- Popup blocked -> a toast, and we bail BEFORE the manager records
anything, so the caller's dialog stays exactly where it was.
- Nobody answers `hello` in 5s -> the window says so instead of spinning.
- A pane with no `script` is a closure in this realm and cannot honestly
cross a window boundary. The window host declines it (canHost) and the
router falls back to the dock.
- A browser blocks window.open() outside a user gesture, so a popped-out
pane cannot be auto-restored on page load — it would only ever produce a
"blocked" toast. Such a pane comes back in the DOCK, and the chip pops it
out again on the next click. (autoRestore: false. The desktop host will
set it true.)
Hosts may now declare `remote: true`, meaning the pane's mount() runs in
another realm: the manager then owns only the authoritative state store and
never calls mount() itself. That is the seam the Electron BrowserWindow +
tray host drops into next, with no change here.
Verified: popped Now Playing and Mixer into real windows. The pane realm has
no window.highway, no capability bus and no <audio>, yet the Mixer renders
its faders via ctx.call('audio-mix','list-faders') across the channel — and
dragging that fader IN THE PANE WINDOW moved the main window's song volume
to 55 and persisted it. Closing the pane window un-hid the mixer dialog,
removed the stub and restored the chip, while the other pane window stayed
open.
Signed-off-by: topkoa <topkoa@gmail.com>
|
||
|
topkoa
|
e5cbea2e9f |
feat(panes): core detachable pane system + pop-out chip
The option-heavy player UIs (mixer, camera director, viz, audio routing)
all live in the rail popovers, which are exclusive: openPopFor() closes
the last one before opening the next. You cannot watch the mixer while
riding the camera, and both vanish the moment you look at the highway.
Add `window.feedBack.panes` — a core registry for live UI that is
authored once as `mount(root, ctx)` and hosted anywhere. Panes are
non-exclusive, and they survive song switches structurally: the dock is a
body child outside every .screen, so the per-song teardown never sees it.
The adoption cost for a plugin is two calls:
feedBack.panes.register({ id, title, icon, mount, unmount });
feedBack.panes.attachChip(myExistingDialogEl, id);
attachChip injects THE standard pop-out chip. Clicking it opens the pane
and hides the plugin's dialog, leaving a stub to bring it back. Core owns
the hide/restore, so every plugin's pop-out looks and behaves the same —
which is the point. It hides via a dedicated .fb-pane-detached class, not
.hidden/[hidden], because the dialogs we attach to already toggle those.
Everything a pane may touch arrives through `ctx` — never a global. That
is what will let the same mount() run inside a pop-out window, a separate
JS realm with no window.feedBack, no window.highway and no audio graph:
ctx.call(domain, cmd, payload) -> the capability bus
ctx.on(event, fn) -> the feedBack bus (allowlisted)
ctx.subscribe(stream, fn) -> playhead / meters
ctx.state.get/set -> persisted, main realm is the only writer
ctx.playhead(), ctx.song(), ctx.toast(), ctx.close()
ctx tracks every subscription it hands out and drops them on unmount, so
a pane cannot leak listeners across a dock/undock cycle.
Streams exist because an AnalyserNode cannot cross a window boundary:
levels are reduced to numbers in the realm that owns the audio graph.
One shared rAF loop, refcounted against live subscriptions, dirty-checked
before fan-out, and stopped dead when the last pane closes.
Hosts register themselves with the manager rather than being imported by
it — the dock lands at priority 0 (the floor, always available), so the
OS pane window can drop in later without this code changing.
Ships two built-in panes: Now Playing (the reference pane — reads the bus,
a stream, and levels, and touches no globals) and Mixer (the same faders
as the rail, via ctx.call('audio-mix', ...), with the chip attached to the
real #mixer-control). Plus a "Panes" rail popover to open panes that have
no dialog of their own; the system tray will mirror that list.
Note the dock sits at z-index 110, not on the docs/plugin-v3-ui.md ladder
(transport 20, rail 30, popovers 40) — those live INSIDE #player's
stacking context, and #player is itself fixed at z-index 100. A dock below
100 is invisible on the one screen panes exist for. Body-level ladder:
#player 100 < dock 110 < toasts 120 < modals 200.
Pop-out windows, the system tray, manifest-declared panes and mirrorGlobal
(the window.__h3dCamCtl proxy the camera director needs) follow.
Signed-off-by: topkoa <topkoa@gmail.com>
|
||
|
Byron Gamatos
|
ea0ca94742
|
feat(career): career plugin — stars, venue tiers, pack downloads (career mode 2/3) (#907)
* feat(career): career plugin — stars from song_stats, venue tiers, pack downloads (career mode PR2) Bundled plugin: per-song stars from best_accuracy (60/75/85% → 1/2/3★), cumulative stars unlock bar → club → arena (data-driven venues.json). Venue packs (UE-rendered crowd loops) download on demand to CONFIG_DIR/plugin_uploads/career/ on a background thread with sha256 + zip-slip validation, served via FileResponse. Career screen (promoted sidebar entry) shows progress and pushes the active venue's manifest into the crowd video layer (v3VenueCrowd, PR1) — degrades cleanly when either side is absent. Pack URLs land in venues.json in PR3. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(career): keep manifest cleanup path alive on delete; badge only for installed venues Codex preflight: nulling _appliedManifestVenue on delete skipped pushCrowdManifest's setManifest(null) cleanup, leaving the crowd layer on a deleted pack; and the 'playing here' badge showed for an override venue whose pack was removed. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(career): generation-guard in-flight manifest fetches Codex preflight: a manifest fetch resolving after a newer refresh (pack deleted, venue switched) could re-apply a stale pack over the user's newer selection — fetches now carry a generation token and bail when superseded. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(career): exclude orphaned song_stats from star totals Codex preflight: scans hide rather than delete stats of removed songs, so stars now apply the same existing-song filter other stats surfaces use (filename IN (SELECT filename FROM songs)). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(career): 50/150 star thresholds + star collection overview Byron's progression tuning: club at 50★, arena at 150★. /state now returns star_detail rows (title/artist joined from the library, stars, best accuracy, next-star threshold) sorted closest-to-next-star first, and the career screen renders a collection panel: tier summary plus a per-song list with a 'N% to next star' practice hint. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(career): venue select/unselect UX, intro manifest support, fullmatch guards - 'Play here' now also defaults the visualization to Venue (remembering the prior viz); active venues show 'Leave venue' which restores it and sets the '__none__' override so no installed venue silently reapplies. - Pack manifests may ship an intro block (flyover video + ambience mp3); files validate like loops/stingers, .mp3 added to the serving whitelist. - Codex preflight: whitelist regexes use fullmatch (trailing-newline names could validate but 500 on serving). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(career): let pushCrowdManifest clear the manifest on Leave venue Codex preflight: nulling _appliedManifestVenue before refresh skipped the setManifest(null) cleanup branch, leaving the crowd playing. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(career): refresh tailwind output --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Byron Gamatos
|
e779c72396
|
feat(venue): reactive crowd video layer behind the 3D highway (career mode 1/3) (#905)
* feat(venue): reactive crowd video layer behind the 3D highway (career mode PR1)
Two crossfading video backdrop planes in the highway_3d venue background
style, driven by a new venue-crowd.js state machine that maps
v3:live-performance-state to crowd states (bored/neutral/engaged/ecstatic)
with 3s stability + 8s dwell hysteresis, plus one-shot reaction stingers
on streak milestones and end-of-song accuracy. Inert without a venue pack
manifest (career plugin, PR2) or the feedBack-venue-crowd-dev flag — the
static bg plate behaves exactly as before.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): retry renderer binding + preserve mid-stinger transitions
Codex preflight P2s: (1) videos created before highway_3d registered its
globals never reached the backdrop planes — binding is now idempotent and
retried from start/perf-event/re-activation paths; (2) a crowd-state
switch committing while a stinger played was dropped because the machine
had already advanced — it is now deferred and played when the stinger ends.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): per-video load tokens + unbind renderer on stop
Codex preflight round 2: (1) the global load token let a stinger cancel a
committed loop load on the other layer — tokens are now per-element, and a
stinger preempting an in-flight loop on its own layer requeues that loop
for when the stinger ends; (2) setManifest(null)/deactivate left the last
crowd frame bound and visible over the static plate — stop() now unbinds
both layers from the renderer and zeroes the mix.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): flush deferred loop on stinger failure, source accuracy from perf events
Codex preflight round 3: (1) a failed/timed-out stinger left a deferred
loop switch queued forever; the failure path now flushes it. (2)
stats:recorded only carries {filename, arrangement}, so the end-of-song
reaction now uses the accuracyPct from the song's last
v3:live-performance-state event (a real percentage) instead of a field
that never existed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): requeue mid-crossfade loops preempted by stingers; hard-stop on manifest swap
Codex preflight round 4: (1) idleLayer() still points at the fading-in
layer during a crossfade, so a stinger firing mid-fade overwrote the new
loop with nothing requeued — the fading loop is now tracked and requeued
like an in-flight load; (2) swapping venue packs while active now goes
through stop() so _stopGen invalidates the old manifest's in-flight loads.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): generation-gate stinger handlers; recrop on video size change
Codex preflight round 5: (1) an ended/timeout handler orphaned by stop()
could fire into a later stinger's lifecycle on the reused element — handlers
now detach unconditionally and carry a generation token; (2) the renderer
only re-applied cover-crop on camera aspect changes, so a src swap with a
different intrinsic size kept stale repeat/offset — it now recrops when
videoWidth/Height change.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): bail loop-fade completion when a stinger preempted the layer
Codex preflight round 6: the loop crossfade's completion callback could
still run between a stinger's start and its canplaythrough, promoting the
stinger's layer to active and pausing the real loop — it now bails when
the fading loop was preempted.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): keep rear video layer opaque during crossfades
Two half-transparent layers let the static bg plate bleed through (~25%
at mid-fade) — visible as a flash of the old still image on every state
transition. The crossfade is now always the front layer fading over an
opaque rear layer.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): reset active layer with mix on stop
Codex preflight: stop() zeroed the mix but left _activeLayer at 1, so a
restart flashed layer 0's stale frame until the new loop loaded.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): reset crowd mood to neutral on song load
Codex preflight: a song ending in ecstatic/bored left the next song's
crowd stuck in that mood until the hysteresis window passed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): cancel in-flight fade when a stinger preempts it
Codex preflight: the orphaned ramp kept pushing the mix toward the layer
whose src the stinger had just replaced.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): don't let null accuracy resets wipe the end-of-song value
Codex preflight: Number(null) is 0, so idle HUD resets overwrote
_lastAccuracyPct before stats:recorded consumed it, suppressing the
end-of-song stinger.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): abort stale stinger state on song load
Codex preflight: a stinger straddling a song change could fade back into
the previous song's layer or flush its pending loop.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): always detach load listeners, gate only the callback
Codex preflight: superseded loads left canplaythrough/error listeners
attached to the persistent video elements — unbounded growth over a
session.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(venue-crowd): flyover intro with crowd-ambience ducking
On song:loaded, an optional pack intro plays once: a camera flyover video
(idle layer, one-shot) with bar-crowd ambience audio that ducks out on
song:play, near the flyover's landing, or at handoff — whichever first.
Machine commits and stingers defer during the intro; stop()/song-change
abort it. Packs without an intro behave as before.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(venue-crowd): fall back to the loop when the intro fails to load
Codex preflight: a failed/timed-out intro left the song with no crowd
loop at all.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
8d3db5f42c
|
fix(nav): nobody may monkey-patch window.showScreen — add screen:changing, make the shell listen (#924) (#925)
Some checks are pending
ship-ci / ci (push) Waiting to run
* fix(nav): the library sometimes showed the legacy screen — map 'home' inside showScreen
Testers: "randomly, when moving to the library from another menu option, the library shows the
old interface — never when a song ends."
━━━ WHAT WAS ACTUALLY HAPPENING ━━━
#home is the PRE-V3 library screen. The v3 shell replaced it with #v3-songs, and the mapping DID
exist — but only inside WRAPPERS on window.showScreen, and only for callers that go through
`window`. THREE independent parties monkey-patch it, each capturing whatever happens to be there
at the time:
app.js publishes the raw function
-> shell.js wraps it, adding the home -> v3-songs mapping
-> the stems plugin wraps it AGAIN (src/main.js:1029), capturing the current value
Plugins load ASYNCHRONOUSLY. The chain links up in whatever order the race settles, and any
capture taken before shell.js installs — or any re-assignment after it — silently drops the
mapping. Hence "randomly".
AND THE INTERNAL CALLERS NEVER TOUCHED window.showScreen AT ALL. closeCurrentSong and the
Esc-from-settings shortcut call the IMPORTED showScreen, which no wrapper ever sees. Reproduced
in a browser: the unwrapped function with 'home' lands on the dead legacy screen EVERY time.
"Never when a song ends" is the tell, and it is what identified the mechanism: closeCurrentSong
resolves its target through _resolvePlayerOrigin(), which ALREADY applies this mapping. That one
path was fine — which is exactly why the bug looked random rather than total.
PRE-EXISTING, not a regression from the module carve: the onclick="showScreen('home')" links and
the wrapper-only mapping both date to 2026-06-22.
━━━ THE FIX ━━━
The guard lives inside showScreen now: ONE place, in the function every caller routes through,
instead of a chain of monkey-patches that must each remember. Wrapper order stops mattering, and
the module-internal callers are covered for the first time.
Verified in a browser: the raw, unwrapped showScreen('home') — which reproduced as #home — now
lands on #v3-songs, and cannot be undone by any wrapper order.
━━━ AND A [P1] I INTRODUCED, WHICH CODEX CAUGHT ━━━
My first cut mapped BOTH 'home' and 'v3-home', copied straight from _resolvePlayerOrigin.
That is correct THERE and wrong HERE. _resolvePlayerOrigin computes where to RETURN TO after a
song, and landing on the Songs list from the dashboard is the right behaviour. But #v3-home is
the v3 DASHBOARD — a real screen that the shell's Home nav, the onboarding tour and the dashboard
re-render listener all target. Redirecting it would have made Home unreachable.
A LEGACY ALIAS IS NOT THE SAME THING AS A RETURN TARGET. Only 'home' is mapped now, and a test
pins that: re-adding 'v3-home' to the guard fails it.
4 tests, bite-tested both ways.
node 1049, pytest 2425, ESLint 0, Codex 0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(nav): nobody may monkey-patch window.showScreen — add screen:changing, make the shell listen (#924)
window.showScreen was wrapped by THREE independent parties, each capturing whatever happened to be
there at the time:
app.js publishes the raw function
-> static/v3/shell.js wrapped it (to call syncActive, and to map home -> v3-songs)
-> the stems plugin wrapped it AGAIN (to tear down on leaving the player)
Plugins load ASYNCHRONOUSLY, so the chain linked up in whatever order the race settled. A capture
taken before shell.js installed silently dropped the mapping it carried — and the library opened on
the dead legacy #home screen. Testers saw that as "randomly, the library shows the old interface"
(#923).
#923 fixed the symptom by moving the mapping inside showScreen. This removes the CAUSE: neither
wrapper ever needed to be one.
━━━ TWO EVENTS, AND THE DISTINCTION IS THE WHOLE POINT ━━━
screen:changing emitted BEFORE anything happens. "I am leaving `from`." Teardown/cancel here.
screen:changed emitted after the DOM and data settle. "I am on `id`." Now carries `from`.
screen:changing is new, and it exists because Codex caught me collapsing the two. The stems plugin
tore down its audio graph BEFORE showScreen did anything; screen:changed fires at the very END,
after core awaits library and provider loads — so moving the plugin onto it would have delayed
teardown behind a slow fetch, or skipped it entirely if that fetch threw, and stems would keep
playing on a non-player screen. A test pins the ordering: screen:changing must precede the first
await.
shell.js is a plain screen:changed listener now, like app.js, audio-mixer.js and tour-engine.js
already were. window.showScreen is an unwrapped function again, and tests/js/
no_showscreen_monkeypatch.test.js fails CI if anything in static/ ever assigns to it again — so the
hazard is structurally impossible rather than merely avoided.
━━━ AND A FALLBACK THAT COULD NEVER FIRE ━━━
My retry-if-the-bus-is-late path listened for `slopsmith:capabilities:ready`. Core dispatches
`feedBack:capabilities:ready` (capabilities.js:1536) — the slopsmith: name is the PRE-DMCA event
and nothing has emitted it since the rename. Codex caught it. A guard that cannot fire is worse
than no guard: it reads as protection and is decoration.
(The same dead-event bug turned out to be sitting in THREE of the stems plugin's fallbacks, where
it has silently disabled its lifecycle wiring whenever the bus was late. Fixed in
feedback-plugin-stems#38.)
VERIFIED. A/B against origin/main: the nav highlight and topbar title follow IDENTICALLY with
shell.js as a listener; screen:changing -> screen:changed fire in order with the right {id, from};
window.showScreen is unwrapped; and showScreen('home') still lands on v3-songs.
node 1053, pytest 2425, ESLint 0, Codex 0.
Closes #924
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
f27d4f623c
|
fix(nav): the library sometimes showed the legacy screen — map 'home' inside showScreen (#923)
Testers: "randomly, when moving to the library from another menu option, the library shows the
old interface — never when a song ends."
━━━ WHAT WAS ACTUALLY HAPPENING ━━━
#home is the PRE-V3 library screen. The v3 shell replaced it with #v3-songs, and the mapping DID
exist — but only inside WRAPPERS on window.showScreen, and only for callers that go through
`window`. THREE independent parties monkey-patch it, each capturing whatever happens to be there
at the time:
app.js publishes the raw function
-> shell.js wraps it, adding the home -> v3-songs mapping
-> the stems plugin wraps it AGAIN (src/main.js:1029), capturing the current value
Plugins load ASYNCHRONOUSLY. The chain links up in whatever order the race settles, and any
capture taken before shell.js installs — or any re-assignment after it — silently drops the
mapping. Hence "randomly".
AND THE INTERNAL CALLERS NEVER TOUCHED window.showScreen AT ALL. closeCurrentSong and the
Esc-from-settings shortcut call the IMPORTED showScreen, which no wrapper ever sees. Reproduced
in a browser: the unwrapped function with 'home' lands on the dead legacy screen EVERY time.
"Never when a song ends" is the tell, and it is what identified the mechanism: closeCurrentSong
resolves its target through _resolvePlayerOrigin(), which ALREADY applies this mapping. That one
path was fine — which is exactly why the bug looked random rather than total.
PRE-EXISTING, not a regression from the module carve: the onclick="showScreen('home')" links and
the wrapper-only mapping both date to 2026-06-22.
━━━ THE FIX ━━━
The guard lives inside showScreen now: ONE place, in the function every caller routes through,
instead of a chain of monkey-patches that must each remember. Wrapper order stops mattering, and
the module-internal callers are covered for the first time.
Verified in a browser: the raw, unwrapped showScreen('home') — which reproduced as #home — now
lands on #v3-songs, and cannot be undone by any wrapper order.
━━━ AND A [P1] I INTRODUCED, WHICH CODEX CAUGHT ━━━
My first cut mapped BOTH 'home' and 'v3-home', copied straight from _resolvePlayerOrigin.
That is correct THERE and wrong HERE. _resolvePlayerOrigin computes where to RETURN TO after a
song, and landing on the Songs list from the dashboard is the right behaviour. But #v3-home is
the v3 DASHBOARD — a real screen that the shell's Home nav, the onboarding tour and the dashboard
re-render listener all target. Redirecting it would have made Home unreachable.
A LEGACY ALIAS IS NOT THE SAME THING AS A RETURN TARGET. Only 'home' is mapped now, and a test
pins that: re-adding 'v3-home' to the guard fails it.
4 tests, bite-tested both ways.
node 1049, pytest 2425, ESLint 0, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
57e7db5c2a
|
refactor(app): carve the keyboard-shortcuts subsystem into static/js/shortcuts.js (R3d) (#922)
19 declarations + 23 TOP-LEVEL STATEMENTS. 922 lines. app.js 3,243 -> 2,325 (-28%). The panel registry, both global keydown dispatchers, the library arrow-nav, and the whole plugin-facing shortcut API. ━━━ MOST OF THIS SUBSYSTEM WAS NOT DECLARATIONS ━━━ A declaration-seeded dependency closure reports this cluster as 10 names, 246 lines. It is 42 statements and 922. window.registerShortcut, createShortcutPanel, getAllShortcuts, unregisterShortcut, clearWindowShortcuts, the panel registry, and BOTH global keydown dispatchers are bare TOP-LEVEL STATEMENTS at app.js's top level. A call-graph scan sees NONE of them. That blind spot has now cost three times: * it nearly shipped a dead library A-Z rail (#896) — 43 of library.js's exports were referenced only from app.js's window contract; * it threw "Assignment to constant variable" in the session carve (#921), where the autoplay gate's top-level statements wrote state that had just become a read-only import; * and here it under-reported the slice by 3x. The extractor takes them by construction now — any top-level statement that TOUCHES a moved binding comes along — and the SEED is closed to a FIXED POINT, because those statements have their own dependencies (_modifiersMatch, _isShortcutActive, _handleLibArrowNav, _gridColumns…) that the declaration closure never walked. Seed -> pull the statements -> the statements need more names -> re-seed. Iterate until it stops growing. ━━━ syncLibrarySong GOES ACROSS THE SEAM, NOT THROUGH AN IMPORT ━━━ The library arrow-nav calls it on Enter. It cannot be imported: syncLibrarySong reaches showScreen/playSong, and a module importing app.js closes a cycle. It is the ONE name here that had to stay behind, so it comes across the host seam — which is exactly what the seam is for. host.js throws loudly if the wiring is ever dropped, and tests/js/host_contract.test.js fails in CI if the hook drifts. VERIFIED. A/B against origin/main in two browsers, IDENTICAL, zero page errors — and driven for real, not merely present: the plugin API (register / unregister / getAll / panels), THE GLOBAL KEYDOWN DISPATCHER actually firing a registered shortcut, that same shortcut correctly SUPPRESSED while typing in a text input, and `?` opening the help modal. node 1045, pytest 2425, ESLint 0 (no-cycle clean), host contract 2/2, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
545e569ad6
|
refactor(app): carve the song session out of app.js — playSong, showScreen, closeCurrentSong (R3d) (#921)
36 declarations + the 4 autoplay/auto-exit gate statements. 359 lines. app.js 3,772 -> 3,242. Bodies VERBATIM. ━━━ THIS WAS "THE UNCUTTABLE HEART", AND IT IS 359 LINES ━━━ At the start of this epic, seeding a dependency closure from count-in, from loops, from section-practice, or from the JUCE seek shim all returned the SAME 178-function, 3,360-line set. playSong and showScreen called each other; everything called them; nothing could be cut anywhere. The conclusion — correct at the time — was that NO closure-based carve could touch it at any seed, and the answer was a host seam. That was true THEN. Every slice taken out since (transport, loops, count-in, section-practice, the library, the edit modal, settings) removed edges, and the strongly-connected component DISSOLVED. This closure is 36 declarations with an interface width of FOUR. The lesson is not that the seam was wrong — the seam is what MADE this possible, by letting the carves proceed against a cyclic core instead of stalling on it. The lesson is to RE-MEASURE. An SCC is a fact about a graph at a moment, not a property of the code. ━━━ THE BUG NO SCAN COULD SEE, AND THE A/B DID ━━━ First cut passed every gate — no-undef clean, no-cycle clean, 1045/1045, pytest green — and THREW IN THE BROWSER: "Assignment to constant variable." window.feedBack.holdAutoplay / holdAutoExit and their two event handlers are TOP-LEVEL STATEMENTS, not declarations. They WRITE this cluster's state (_autoplayHeld, _autoExitTimer, …), and an imported binding is READ-ONLY — so left behind in app.js, every one threw the instant the module existed. A dependency scan that walks DECLARATIONS cannot see them. Mine didn't. This is the same blind spot that nearly shipped a dead library A-Z rail (#896): app.js keeps its public API in top-level statements, and a call-graph is blind to every one of them. The extractor now finds them by construction — any top-level statement that WRITES a moved binding comes with the carve — and the gate statements live beside the machinery they drive, which is where they belonged anyway. ━━━ ZERO OUTSIDE WRITES, BY MOVING THE BOUNDARY RATHER THAN BUILDING MACHINERY ━━━ The autoplay scalars and the wake-lock state were written from outside the cluster, which would have forced a setter or a state container. But the writers — _releaseAutoplay, _acquireWakeLock — plainly belong here. Pulling them in left ZERO outside writes, so every export is a plain import. Same move as settings (#920): measure the writers before you reach for a container. VERIFIED. A/B against origin/main in two browsers, IDENTICAL, zero page errors — including the autoplay gate driven end to end: a plugin HOLDS autoplay, the song loads but does not start, the RELEASE fires it, and a stale release is a no-op. That is the exact machinery that was throwing. node 1045, pytest 2425, ESLint 0 (no-cycle clean), host contract 2/2, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
84fe29688c
|
refactor(app): carve settings into static/js/settings.js (R3d) (#920)
22 declarations, 446 lines. app.js 4,218 -> 3,772. Bodies VERBATIM. Settings load/save, the AV-offset nudge, the default-arrangement pin, the instrument pathway, and the app-update channel. ━━━ INTERFACE WIDTH 1, AND IT GOT THERE BY DRAWING THE BOUNDARY IN THE RIGHT PLACE ━━━ app.js calls loadSettings() and nothing else. The first cut was NOT clean: _defaultArrangement was written from OUTSIDE the cluster, and an imported binding is READ-ONLY, so that one write would have forced a setter or a state container — as it did for the player (player-state.js) and the library (library-state.js). But the writers were saveSettings and pinCurrentArrangementDefault, which ARE settings functions. Widening the slice to include them left ZERO outside writes. Every export is now a plain read-only import and no container is needed. Worth naming, because I reached for a container twice before: the fix for "this binding is written from outside" is sometimes a container, and sometimes it just means the boundary is in the wrong place. Measure the writers before you build machinery. ━━━ handleSliderInput STAYS A HOST HOOK, DELIBERATELY ━━━ It lives in settings now (it is a settings control), but player-controls.js must NOT import it: this module already imports player-controls (_applyMastery, _autoplayExitEnabled, …), so a direct back-import would close a cycle. player-controls keeps reading it through the host seam, and app.js — the root, which imports both — wires it. That is exactly what the seam is for, and the contract test proves the wiring survived. VERIFIED. A/B against origin/main in two browsers: the window contract, the settings screen rendering, the AV-offset and default-arrangement controls present, and a real `input` event dispatched on a slider — which is the path that goes through the host seam. IDENTICAL, zero page errors. node 1045, pytest 2425, ESLint 0 (no-cycle clean), host contract 2/2, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
69aac32278
|
refactor(app): carve the edit-song modal into static/js/edit-modal.js (R3d) (#919)
4 functions, 234 lines. app.js 4,452 -> 4,218. Bodies VERBATIM. INTERFACE WIDTH ZERO — nothing in app.js calls into this cluster. app.js needs only the names on the window contract, so the markup's onclick= handlers resolve. That is what makes it the cleanest slice left. AND IT ONLY BECAME CLEAN BECAUSE THE LIBRARY CAME OUT FIRST (#896). Every dependency the modal has is a module now: it reads six bindings out of ./library.js (loadLibrary, loadFavorites, loadTreeView, _removeLibCardsForFilename, libView, _lastLibSelected) plus dom.js and the L container. Before that carve, extracting this would have dragged the whole library with it. Checked, and it matters: the modal never WRITES any of those six. An imported binding is READ-ONLY, so a single write would have forced a setter or a state container. Every use is a read, so plain imports suffice. Acyclic: edit-modal -> { dom, library-state, library }, and library imports none of them back. VERIFIED. A/B against origin/main in two browsers: the window contract, the modal actually OPENING off a real library row, its title and year fields rendering, and the data-edit-save wiring (rather than an inline onclick embedding the filename — the fix this cluster's harness exists to guard). IDENTICAL, no new page errors. node 1045, pytest 2425, ESLint 0 (no-cycle clean), host contract 2/2, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
0a6e0309e5
|
fix(tailwind): stop the dev server rewriting a tracked file (#911) (#918)
The runtime stylesheet moves to CONFIG_DIR. static/tailwind.min.css is never written again.
━━━ TWO DIFFERENT THINGS WERE SHARING ONE PATH ━━━
static/tailwind.min.css a BUILD ARTEFACT. Committed, image-baked, generated by scanning
the in-tree plugins only. CI's tailwind-fresh check verifies it.
the RUNTIME sheet PER-INSTALL STATE. Additionally scans whatever the user installed
into FEEDBACK_PLUGINS_DIR, so it differs machine to machine.
Writing the second over the first meant that MERELY RUNNING THE DEV SERVER from a git checkout
silently modified a tracked file. `git add -A` then swept a 100KB reshuffle of minified CSS
into the commit and ci/tailwind-fresh went red with a diff that explains nothing — on a PR
whose real change touched no Tailwind classes at all. It also wrote app state into the app
directory, which is read-only in some deploys.
A new route serves the runtime sheet when there is one and falls back to the committed one
otherwise. It is registered BEFORE the /static mount, which would otherwise swallow the path.
━━━ A PERSISTED SHEET MUST NOT OUTLIVE ITS REASON (Codex [P2] x2) ━━━
1. THE USER REMOVES THEIR PLUGINS. Startup only rebuilds when user plugins exist, so nothing
would ever overwrite the stale sheet — and it still carries classes for plugins that are
gone. With no user plugins the COMMITTED sheet is complete by definition. Guarded.
2. THE APP IS UPGRADED, and my first guard for this was WRONG. I compared mtimes. Codex: that
is not a freshness signal across install methods — archives and container images routinely
PRESERVE SOURCE MTIMES, so a just-shipped stylesheet can carry an OLDER timestamp than a
runtime sheet a user built days ago. The mtime check then calls the stale one FRESH and it
masks the new core CSS indefinitely — permanently, if no Tailwind toolchain is present to
trigger a rebuild.
Freshness is decided by CONTENT now. Each runtime build stamps a sidecar with the sha256 of
the committed sheet it was made from. Core ships new CSS -> that file changes -> the hash
changes -> the runtime sheet is correctly judged stale. Timestamps only gesture at the
question that hashing answers.
Falling back to the committed sheet is always safe: at worst it lacks a just-installed plugin's
classes for the seconds until the async rebuild lands.
VERIFIED END TO END. Ran the real dev server with 3 plugins installed: it rebuilt Tailwind over
them (123,291 bytes), wrote the sheet + sidecar to CONFIG_DIR, still served /static/
tailwind.min.css at 200 — and `git diff` on the tracked file came back CLEAN.
8 tests. Bite-tested: reverting to the shared path fails 3, dropping the staleness guards fails
2 more.
pytest 2425, pyflakes 0, Codex 0.
Closes #911
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
36cf77dc44
|
refactor(highway): carve the 2D drawing layer into highway-draw.js (R3c) (#917)
18 functions, 1,245 lines. highway.js 3,972 -> 2,727 (-31%). The biggest R3c slice: notes,
sustains, chords, strum groups, unison bends and lyrics — everything the default renderer
paints each frame.
━━━ MUTABILITY, NOT LOCATION, DECIDES WHERE A THING BELONGS ━━━
Three per-instance caches came out with this slice, and they are why it needed care:
_frameMismatchWarned a warn-once Set of chord ids (feedBack#88)
_chordRenderInfo a WeakMap of chord -> chain info
_lyricMeasureCache Map<fontSize, Map<text, width>>
All three are MUTATED. Left at module scope they would be SHARED ACROSS PANELS — one
highway's lyric widths and chord chains stomping another's, silently, with nothing throwing.
createHighway() is a factory (the constitution publishes window.createHighway so a plugin can
build a second highway), so they are lifted onto hwState, which is exactly what hwState is for.
The shimmer LUT went the OTHER way — to MODULE scope in highway-geometry.js. It is a
deterministic xorshift table, byte-for-byte identical for every instance, so sharing it is not
merely safe but BETTER: built once for the page rather than once per panel.
Same slice, opposite directions, decided entirely by whether the thing mutates.
━━━ MY SCRIPT WAS WRONG TWICE. THE GATES CAUGHT BOTH. ━━━
1. HAND-LISTED THE MOVE SET. I listed 10 functions and missed six that drawChords needs
(_ensureChordRenderCache, bsearchChords, getChordTemplateInfo, _computeChordBox,
_updateFretLinePreview, _drawFretLineChordPreview). The no-undef gate named every one. The
set is now DERIVED from the dependency closure — 18, not 10.
2. JUDGED PURITY TOO EARLY, and this one is subtle. I classified _computeChordBox as pure
because its ORIGINAL body never mentions hwState. Then the call-site rewriter injected
`fretX(hwState, …)` INTO it — fretX takes hwState now (#916) — leaving a function that
references an hwState it was never given. Purity has to be judged from the body AS IT WILL
BE, so the classifier iterates to a fixed point: a function needs hwState if it mentions it,
OR calls anything that now takes it. That moved _computeChordBox to the stateful side.
VERIFIED. A/B against origin/main: IDENTICAL, zero page errors. The PLUGIN BUNDLE contract is
byte-identical (b.fretX arity 3, b.getNoteState arity 2, both stable references, both correct
under the old calling convention). PERF GATE PASSES AT 1.92ms against its 12ms budget — and
this is the slice that could really have cost something: the ENTIRE per-frame drawing path is
now cross-module. It costs nothing measurable.
TESTS. highway_teaching_marks follows strumGroupBuckets to the new module. The two source-shape
harnesses now read highway.js AND every static/js/highway-*.js, rather than being re-pinned at
whichever file currently holds a function — re-pinning breaks again next time, and a shape
assertion that silently stops finding its target is indistinguishable from one that passes.
node 1045, pytest 2416, ESLint 0, no-undef 0, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
12eb73aee9
|
refactor(highway): carve the STATEFUL primitives, threading hwState explicitly (R3c) (#916)
fretX, fillTextReadable, _noteState, _paintGemGlow -> static/js/highway-state-primitives.js.
50 call sites rewritten. highway.js 4,105 -> 3,965.
The first slice that changes signatures. Each of these four gains hwState as an explicit
FIRST PARAMETER.
━━━ hwState IS A PARAMETER, NOT AN IMPORT ━━━
createHighway() is a FACTORY. The constitution publishes window.createHighway so a plugin can
build a SECOND highway for its own panel, and highway.js says so itself. Import hwState as a
module singleton and two panels silently share one clock, one render scale, one string
palette — each driving the other. Nothing throws. The picture is just wrong, in a way no test
would catch.
(The exact opposite of the app.js carve, where player-state.js and library-state.js ARE
module singletons — correctly, because there is exactly one app. Same epic, same language,
opposite answer, decided entirely by whether the thing is a factory.)
━━━ THE PLUGIN BUNDLE NEARLY BROKE, SILENTLY ━━━
The renderer bundle hands two of these STRAIGHT TO PLUGINS:
b.fretX = fretX;
b.getNoteState = _noteState; // stable reference
highway_3d calls both EVERY FRAME, with the old arity. Handing out the new 3-arg versions
would have passed `note` where hwState belongs — no throw, no error, just wrong geometry and
wrong judgment state INSIDE A PLUGIN, which no core test would ever see. Green CI, broken 3D
highway.
So hwState is bound ONCE per instance, in the factory, and the bundle hands out those views.
A per-frame arrow would have fixed the arity and reintroduced exactly the per-frame allocation
the bundle's stable-reference contract (feedBack#254) exists to prevent. b.project needs none
of this — project() is pure and its arity never changed.
VERIFIED IN A BROWSER, against the real bundle, on both builds:
fretX arity 3 3 (NOT 4 — the bound view preserves it)
getNoteState arity 2 2
fretX(5,1,800) in 0..800 True True
getNoteState null w/o provider True True
getNoteState honours provider True True
fretX is a stable reference True True
IDENTICAL. Without the bound views fretX would have reported arity 4 and computed garbage.
Also caught on the way: my generated module imported STRING_BRIGHT_FALLBACK, a name
highway-constants.js does not export. ESLint does not flag that — but importing a name a
module does not export is a runtime SyntaxError that kills the WHOLE module. These four need
no constants at all; the import is gone.
TESTS. highway_note_state pins the signature AND the stable-reference contract — it caught the
bundle break. Retargeted at the module and the new arity; both contracts still asserted, and
the "no fresh arrow per frame" rule is now asserted explicitly rather than implied by
`getNoteState: _noteState`.
PERF GATE PASSES AT 1.94ms against its 12ms budget — fretX and _noteState are now CROSS-MODULE
calls, per note, per frame. It costs nothing measurable.
node 1045, pytest 2416, ESLint 0, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
1a386c272d
|
refactor(highway): carve the PURE geometry primitives into highway-geometry.js (R3c) (#915)
6 functions, 53 lines. highway.js 4,158 -> 4,105. NOT ONE CALL SITE CHANGES. project, roundRect, bnvNormalizedPoints, teachingFingerLabel, teachingDegreeLabel, chordHarmonyLabels — the shared primitives every drawing function leans on. ━━━ PURITY IS THE WHOLE POINT OF THIS SLICE ━━━ Every one of these is a pure function of its arguments. None touches hwState. None closes over the canvas context — roundRect() already took `ctx` explicitly, and the rest need nothing but numbers. project() reads only the module-level constants from #914. That matters because createHighway() is a FACTORY: a plugin can build a second highway for its own panel, so anything holding per-instance state must be PASSED hwState rather than importing it, or two panels silently share one clock and palette. These six hold no state at all, so they move VERBATIM — the module boundary is invisible to every caller. The asserts are mechanical and in the extractor: it REFUSES to move a function whose body mentions hwState, or that references `ctx` without taking it as a parameter. Purity is checked, not assumed. ━━━ WHAT IS DELIBERATELY LEFT BEHIND ━━━ The four primitives that DO need hwState — fretX, fillTextReadable, _noteState, _paintGemGlow — stay in the factory for now. They need an explicit hwState parameter threaded through 53 call sites, which is a real behavioural change and belongs in its own commit rather than smuggled in beside a provably-identical move. Separating the provable from the risky is the whole discipline of this epic. TESTS. Three harnesses brace-match these functions out of the source and run them in a sandbox; they now read static/js/highway-geometry.js. `export function x` still contains `function x`, so the extractor needed no change — only the path. VERIFIED. A/B against origin/main: 15 probes IDENTICAL, zero page errors. PERF GATE PASSES AT 1.91ms against its 12ms budget — and this is the one that could plausibly have cost something: project() runs for every visible note on every frame and is now a CROSS-MODULE call. It costs nothing measurable. That is the answer #910 was built to give. node 1045, pytest 2416, ESLint 0, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
8e89b39ad3
|
refactor(highway): carve the constants into static/js/highway-constants.js (R3c) (#914)
29 constants, 190 lines. highway.js 4,267 -> 4,158. The first real slice, and the one that
every later one imports.
━━━ WHY ONLY THE CONSTANTS MAY LIVE AT MODULE SCOPE ━━━
createHighway() is a FACTORY, not a singleton. The constitution publishes
window.createHighway precisely so a plugin can build a SECOND highway for its own panel, and
highway.js already says so at the top of the closure:
// R3c: per-instance mutable state in one object, so extracted renderer/ws
// modules can close over it as a factory arg without cross-panel sharing.
So hwState — all 79 mutable properties — must NEVER become a module-level singleton: two
highways would silently share it, and one panel would drive the other's clock, scale and
colour tables. Extracted functions will take it as an ARGUMENT.
That is the OPPOSITE of the app.js carve, where a single state container (player-state.js,
library-state.js) was exactly right, because there is exactly one app. Same epic, same
language, opposite answer — because one is a singleton and the other is a factory.
These 29 are pure literals: numbers, strings and colour tables, never reassigned, never
mutated. Sharing them across instances is not merely safe, it is what you want — one copy of
the shimmer LUT bounds and the string palettes rather than one per panel. Anything with a
runtime dependency (document, window, performance, localStorage) stays in the factory;
checked, and none of these has one.
ESLint now knows static/highway.js is a module. It could not have known before this commit:
the flip (#913) changed the SCRIPT TAG, but the file had no import/export yet, so it still
parsed as a script and lint stayed green. The first `import` is what makes the config wrong.
TESTS. Four source-shape harnesses asserted `const _AUTO_SCALE_MIN = …` etc. lived in
highway.js. They now read highway.js AND every static/js/highway-*.js — deliberately, rather
than being re-pinned at whichever file currently holds a constant. Re-pinning just breaks
again on the next carve, and a source-shape assertion that silently stops finding its target
is indistinguishable from one that passes. Bite-tested: renaming two constants away fails
them.
VERIFIED. A/B against origin/main: 15 probes IDENTICAL, zero page errors. AND THE PERF GATE
PASSES AT 1.97ms against its 12ms budget — which is the point of having built it (#910)
first: these constants moved from closure scope to module scope, and V8 does not treat those
identically. It does here. Now I know rather than hope.
node 1045, pytest 2416, ESLint 0, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
c6963fdf30
|
refactor(highway): flip highway.js to an ES module (R3c) (#913)
Two lines. index.html: defer -> type="module". highway.js: one explicit assignment.
highway.js can now `import`, which is the whole point — the carve can begin.
━━━ THE ONE THING THE FLIP ACTUALLY BREAKS: window.createHighway ━━━
A top-level `function createHighway()` in a CLASSIC script IMPLICITLY becomes
window.createHighway. In a module it does not — module declarations are module-scoped, and
the name vanishes from the global object the instant the tag grows type="module".
The constitution names window.createHighway as PUBLIC EXTENSION CONTRACT (alongside
window.playSong / showScreen / feedBack). NOTHING IN-TREE CALLS IT. That is exactly why this
would have shipped: the only consumers are third-party plugins rendering their own highway
panel, and I cannot grep those. Green CI, green tests, and a broken plugin API.
Verified by removing the assignment and reloading:
flip WITHOUT an explicit assignment: window.createHighway === undefined <-- gone
flip WITH it: window.createHighway === function
So it is assigned explicitly now — same object, same behaviour, no longer an accident of how
the file happens to be loaded.
━━━ AND A CORRECTION TO #912 ━━━
#912 (merged) rewrote 73 bare `highway.x` -> `window.highway.x` on the stated grounds that
the flip would turn every one of them into a ReferenceError. HAVING NOW ACTUALLY FLIPPED IT,
THAT WAS WRONG. highway.js already did `window.highway = highway`, which puts the name on the
GLOBAL OBJECT — and bare-identifier resolution falls back to the global object whether or not
a lexical global binding exists. Measured on both builds: bare `highway` resolves either way.
#912 is defensible as hygiene and it does not hurt, but it was not a precondition and it fixed
no latent bug. A correction is posted on the PR so its commit message does not mislead. The
real hazard was the factory, not the instance — same class of breakage, wrong name.
ORDERING is unchanged: classic-defer and non-async type="module" share ONE post-parse
execution queue, in document order, so highway.js keeps its position at index.html:1244.
VERIFIED. A/B against origin/main: 15 probes IDENTICAL, zero page errors — window.highway,
window.createHighway, the full API surface, a real song playing, the chart clock advancing,
and the seek->setTime sync. THE PERF GATE PASSES at 1.85ms against its 12ms budget (module
evaluation costs nothing at render time), which is exactly what #910 was built to tell me.
node 1045, pytest 2416, ESLint 0, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
d9fa6d3f55
|
refactor(highway): make the highway global explicit before the module flip (R3c) (#912)
73 bare `highway.x` references -> `window.highway.x`, across app.js and 10 other files. Provably a NO-OP today. It is the precondition for flipping highway.js to a module. ━━━ WHY THIS HAS TO LAND FIRST ━━━ highway.js is a CLASSIC script. Its top-level `const highway = createHighway()` therefore creates a GLOBAL LEXICAL BINDING — visible as a bare name to every other classic script AND to every ES module. 73 call sites quietly rely on that. The moment highway.js becomes a module, that binding is gone. `const` in a module is module-scoped, not global. Every one of those 73 sites becomes a ReferenceError, and the flip is impossible until they say what they mean. `window.highway = highway` is already set, to the same object, on the same line. So this is an identity rewrite — verified in the browser below. ━━━ THE REWRITE BIT ME THREE TIMES. REGEX IS NOT ENOUGH FOR THIS. ━━━ 1. A SHADOWED LOCAL. capabilities/note-detection.js does `const highway = window.highway`. Its 9 bare uses are LOCAL and already correct; a blind rewrite would have emitted `const window.highway = window.highway`. Excluded. 2. HALF-CONVERTED GUARDS — the dangerous one. Six sites read `typeof highway !== 'undefined' && highway && typeof highway.setTime === 'function'`. The regex converted the CONSEQUENT and left the TEST, which is WORSE than not touching them: after the flip `typeof highway` is 'undefined', so each guard is PERMANENTLY FALSE and the code behind it silently never runs. transport.js's was the seek->setTime sync: the chart clock would have quietly desynced after every seek, with nothing failing. All six now test window.highway. 3. TWO MORE BARE REFERENCES, found by Codex [P2] and confirmed by an AST scan: app.js:3114 and :3176 use `highway && typeof window.highway.getSections === 'function'`. My grep searched for `typeof highway`, not `highway &&`. After the flip these throw, the catch swallows it, and the editor silently falls back to a ±4s edit window and arrangement 0. Regex missed a shadow, a half-conversion, and two bare reads. The final check is an AST pass that resolves scopes and reports every `highway` identifier not bound locally. It now reports ZERO. VERIFIED. A/B against origin/main in two browsers, 15 probes, IDENTICAL, zero page errors: window.highway is the same object as the bare global, the whole API surface resolves, a real song plays, the chart clock advances, getPerf().drawMs > 0 — and `seek syncs chart` passes, which is the exact guard I nearly broke in (2). node 1045, pytest 2416, ESLint 0, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
23ecddc721
|
test(highway): the R3c perf gate — measure the render loop before carving it (R3c) (#910)
highway.getPerf() (additive) + tests/browser/highway-perf-baseline.spec.ts. No behaviour change. This lands BEFORE highway.js is touched, because a perf-gated refactor without a perf gate is just a refactor. ━━━ FRAME RATE IS THE WRONG THING TO MEASURE ━━━ The highway AUTO-SCALES. When the smoothed draw cost passes _DRAW_BUDGET_HI_MS (12ms) it LOWERS THE RENDER RESOLUTION to protect the frame rate (#654). Exactly right for players — and it means a real perf regression does NOT show up as dropped frames. It shows up as a BLURRIER PICTURE at a perfectly healthy 60fps. Benchmark fps and you measure the feedback loop, not the renderer, and conclude nothing changed while the image quietly degrades. So the gate pins the scale (setRenderScale(1) + setMinRenderScale(1), which clamps autoScale to [1,1]) and measures drawMs — the renderer's own cost. None of that was reachable before: neither drawMs nor the effective scale escaped the closure. Hence getPerf(). The threshold is the app's OWN: _DRAW_BUDGET_HI_MS is the cost at which the highway itself starts sacrificing resolution in production. Exceeding it is not an arbitrary benchmark line — it is the renderer failing its own budget. Current cost ~2.2ms, so ~5x headroom: far more than headless-CI variance, far less than any regression worth shipping. ━━━ I WROTE THIS GATE WRONG THREE TIMES. EACH TIME IT PASSED. ━━━ 1. VACUOUS ASSERTION. First cut asserted "the auto-scaler wasn't forced to intervene", i.e. effectiveScale == 1. I injected a 10x regression (drawMs 2.4 -> 22.4ms, nearly DOUBLE the budget) and it PASSED. Of course it did: setMinRenderScale(1) sets the scaler's FLOOR to 1, so effectiveScale CANNOT drop below it. The very pinning that stops the scaler hiding a regression also stops it ever reporting one. A guard that cannot fail. 2. MEASURING AN IDLE RENDERER (Codex [P2]). playSong() takes ~3-4s to actually start — it is fetching and decoding stems. My "if not playing after 2s, togglePlay()" fired BEFORE autoplay, started playback, and then the app's own autoplay toggled it straight back to PAUSED. The renderer idled through the entire measurement. Now it WAITS for playback rather than racing it, and asserts the chart clock advanced DURING the sampling window — not merely at some point beforehand, which the first fix would have accepted. 3. UNENCODED FILENAME (Codex [P2]). playSong() decodes its argument before building the /ws/highway path, so every real caller passes encodeURIComponent(filename) (app.js:2879, 4137). Raw, a name containing # ? % or / yields an invalid WebSocket URL, the song never loads — and on those libraries the gate would have silently measured an idle renderer instead of failing. Every one of those bugs made the gate PASS. That is the whole hazard of a perf test: it fails safe in the wrong direction. BITE-TESTED, and this is the only reason I trust it: a 10x regression injected into the draw path FAILS the gate under live playback (22.0ms vs the 12ms budget) and the clean build passes at ~2.1ms with the chart clock advancing 5.6s across the sample. node 1045, pytest 2412, ESLint 0, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
79825af28e
|
fix(demo): the janitor re-entry guard actually works now (#902) (#909)
The guard in startup_events() read:
if getenv_compat("FEEDBACK_DEMO_MODE") or getenv_compat("FEEDBACK_DEMO_MODE") == "1" \
and not _DEMO_JANITOR_STARTED:
`and` binds tighter than `or`, so that is `A or (B and C)`. The not-already-started half
never ran when the env var was truthy — the only case that reaches it at all. A second
startup started a SECOND janitor thread, overwrote the handle, and shutdown then joined
only the last: the first leaked and kept firing registered hooks hourly, forever.
The guard now lives INSIDE start_janitor(). A caller cannot get operator precedence wrong
if there is nothing left for it to get wrong.
━━━ THREE WAYS TO WRITE THIS GUARD WRONG. I HIT ALL THREE. ━━━
1. NO GUARD — the original bug. Double-start, orphaned thread.
2. GUARD ON THE FLAG (`if _DEMO_JANITOR_STARTED: return`). Codex [P2]. stop_janitor()
DELIBERATELY leaves that flag True when a hook outruns its join timeout, so that a later
startup cannot spawn a janitor beside a live one. But the hook usually finishes a moment
later: the thread exits and the flag is stale. A flag-keyed guard then refuses to start a
replacement for the rest of the process — demo cleanup silently dead. (The original bug
accidentally MASKED this by always starting.)
3. GUARD ON LIVENESS ALONE (`if thread.is_alive(): return`). Codex [P2], second pass. A
timed-out stop leaves the old thread ALIVE BUT DOOMED — its stop event is set and it
exits as soon as its current hook returns. Treating that as a running janitor skips the
replacement, and we are back at (2) a second later.
So: a janitor counts as running only if its thread is alive AND it has not been told to stop.
━━━ AND EACH JANITOR NOW OWNS ITS STOP EVENT ━━━
start_janitor() used to `_DEMO_JANITOR_STOP.clear()` a single SHARED Event. Start a
replacement while a doomed thread is still finishing a hook and that clear RESURRECTS it: it
loops back to wait(), sees the flag cleared, and carries on. Two janitors — the exact bug we
started from. A fresh Event per janitor makes it impossible; the old thread waits on its own
event, which stays set, so it can only exit.
Env semantics UNCHANGED, verified across every value ("", "1", "0", "true", "false", "off"):
the old expression and demo_mode_enabled() agree on all of them. The only behavioural change
is the idempotency fix.
FOUR tests, and each of the three wrong guards fails a different subset:
no guard -> 2 fail (double start; orphaned thread)
guard on the flag -> 2 fail (never restarts after a timed-out stop)
liveness alone -> 1 fail (no replacement for a doomed janitor)
liveness + not-stopping -> all pass
pytest 2416, pyflakes 0, Codex 0.
Closes #902
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
OmikronApex
|
db3ca34fcb
|
Merge pull request #906 from got-feedBack/fix/loopback-raw-audio
Some checks are pending
ship-ci / ci (push) Waiting to run
fix(static): raw stereo loopback capture — no voice-call DSP (tin-can fix) |
||
|
OmikronApex
|
34215fbd32 |
fix(static): request raw stereo audio for the loopback capture track
Chromium treats a getDisplayMedia audio track as a voice call by default: echo cancellation, noise suppression, auto gain control and mono downmix. Music through that pipeline is the tester-reported "tin can" sound on ASIO/exclusive outputs. Request the raw path explicitly (EC/NS/AGC off, stereo, 48 kHz) — all constraints are best-effort so unsupported ones degrade silently instead of failing the capture. The [asio-diag] loopback line now dumps track.getSettings() so logs prove which processing actually applied. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> |
||
|
Byron Gamatos
|
f5d448af5c
|
refactor(server): carve demo mode into lib/demo_mode.py (R3b) (#903)
lib/demo_mode.py (342). server.py 1,870 -> 1,649.
The read-only request guard (its 96-entry blocked-route table + the middleware) and the
hourly session janitor (registry, hook runner, thread). Bodies VERBATIM.
THE MIDDLEWARE NEEDS `app`, SO THE MODULE TAKES IT. _demo_mode_guard is an
@app.middleware("http") and cannot exist without an app object. Rather than have a module
under lib/ reach for a global, it exposes install(app) and server.py — which owns the app —
hands it over. The janitor is symmetrical: start_janitor() / stop_janitor(), called from
server.py's startup and shutdown hooks, where the process lifecycle actually lives.
register_demo_janitor_hook IS PART OF THE PLUGIN CONTRACT. It is a key in plugin_context,
so plugins hold it as a LIVE REFERENCE from setup(). server.py imports this exact object
and puts it in the dict unchanged — identity preserved, and
tests/test_plugin_context_contract.py (#898, merged) fails if that ever stops being true.
This is the first carve that guard has actually protected.
━━━ stop_janitor()'s ORDER IS LOAD-BEARING ━━━
The obvious way to write it — clear the "started" flag, then join — is WRONG, and I wrote
it that way first. server.py's original deliberately returns EARLY, leaving
_DEMO_JANITOR_STARTED True and the thread handle intact, when the thread outlives the join:
# Leave _DEMO_JANITOR_STARTED True so a new janitor is not
# spawned by a subsequent startup while the old one is alive.
Clearing the flag first quietly reintroduces exactly the double-janitor leak the flag
exists to prevent. Preserved byte-for-byte, and the reason is now written down at the
function rather than only at its single call site.
━━━ A BUG MOVED VERBATIM, ON PURPOSE ━━━
if getenv_compat("FEEDBACK_DEMO_MODE") or getenv_compat("FEEDBACK_DEMO_MODE") == "1" \
and not _DEMO_JANITOR_STARTED:
`and` binds tighter than `or`, so this is `A or (B and C)` — the not-already-started
re-entry guard is DEAD whenever the env var is truthy, which is the only case that runs.
A second startup leaks a janitor thread (the handle is overwritten, so shutdown joins only
the last). Verified. Preserved exactly and filed as issue #902: a carve whose whole value
is being provably behaviour-neutral is not the place to change behaviour.
pyflakes caught three more missing imports on the way in (uuid, warnings x2). Five carves,
ten missing imports, every one a NameError on a live path.
pytest 2399, pyflakes 0, Codex 0.
Refs #48
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
8f014e6a30
|
refactor(server): carve the library scanner into lib/scan.py (R3b) (#901)
lib/scan.py (326). server.py 2,098 -> 1,870.
The background scan, its spawn ProcessPoolExecutor, and the kick/runner plumbing that
serialises passes. Bodies VERBATIM except the seam reads.
Everything shared is read LATE off appstate — the same contract every module in
lib/routers/ uses, and it is not cosmetic: tests monkeypatch CONFIG_DIR and swap meta_db,
so a value captured at import time pins the wrong one for the life of the process.
CONFIG_DIR -> appstate.config_dir
meta_db -> appstate.meta_db
_default_settings -> appstate.default_settings()
_stat_for_cache -> appstate.stat_for_cache()
━━━ THE SCAN STATUS IS REBOUND, NOT MUTATED ━━━
_background_scan does `global _scan_status; _scan_status = {**INIT, ...}` at every stage
transition. It REPLACES the dict; it never updates it in place. So nothing may hold that
dict by value — a reference captured once goes permanently stale at the first stage change
and would report "listing" forever while the scan ran to completion.
Hence `scan.status()`, a getter, and hence appstate publishes scan_status as a CALLABLE.
appstate.py already said so in a comment; this is the code that makes it true. (Same for
the plugin_context entry, which was already `lambda: dict(_scan_status)` — late-bound, so
it survives the move unchanged. The contract test from #898 covers it.)
━━━ appstate.server_root: A TRAP CLOSED PERMANENTLY ━━━
_background_scan seeds the builtin content, which needs the directory holding server.py.
`Path(__file__).resolve().parent` is correct in server.py and silently WRONG anywhere under
lib/ — it yields lib/, which holds no docs/ or data/ — and it fails by finding NOTHING
rather than by raising, so the seeds would just quietly never run.
lib/builtin_content.py (#900) closed that by taking the root as a parameter. This adds the
other half: server.py publishes it ONCE as appstate.server_root, so no module under lib/
ever has a reason to derive it. Documented at the slot.
pyflakes caught two more missing imports on the way in (loosefolder_mod, enrichment) —
each a NameError on a live scan path, and the suite would have handed them over one failure
at a time. It stays part of every server.py slice.
TESTS. The two scan fixtures (test_settings_api::scan_module,
test_feedpak_extension::scan_server) patched server._make_scan_executor to swap the spawn
pool for an in-process ThreadPool; they now patch it on lib/scan.py. Worth noting WHY that
still works: the fixtures re-import `server` per test, but `scan` stays cached in
sys.modules — and it picks up the fresh CONFIG_DIR anyway, because the appstate reads are
late-bound. The seam is doing exactly the job it was built for.
━━━ TEST ISOLATION: A REGRESSION THE CARVE ITSELF CREATED (Codex [P2]) ━━━
background_scan() deliberately NEVER sets running=False — ownership of that flag lives in
_scan_runner, so a kick_scan() racing the terminal write cannot observe a stale False and
start a second runner. Correct in production.
But the scan fixtures call background_scan() DIRECTLY, skipping the runner. That was
harmless while the state lived on `server`, which the fixtures RE-IMPORT per test. It is
NOT harmless now: `scan` stays cached in sys.modules across sys.modules.pop("server"), so
the status dict OUTLIVES the test. One direct call leaves the shared scanner marked
"running" forever, and every later scan or rescan returns "already in progress" and quietly
does nothing.
Verified: after a direct call, kick_scan() returns False and starts no scan at all.
The suite passed anyway, on ordering luck — which is exactly how this class of bug ships.
tests/conftest.py::reset_scan_state now snapshots and restores lib/scan.py's module state
around the two fixtures that drive it directly.
pytest 2398, pyflakes 0, Codex 0.
Refs #48
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
6b8f79dd9a
|
fix(server): a raising tuning provider no longer takes down get_merged() for everyone (#899) (#904)
One word. server.py's TuningProviderRegistry.get_merged():
except Exception:
- logger.exception("tuning provider %r raised during get_merged()", provider_id)
+ log.exception("tuning provider %r raised during get_merged()", provider_id)
There is no `logger` in server.py — the module logger is `log`. So the handler written to
swallow-and-report a bad provider instead raised NameError from inside the except, and that
NameError propagated out of get_merged().
The effect was the exact OPPOSITE of what the handler is for: one misbehaving plugin took
the whole merged-tunings call down for every other provider, AND the traceback named the
wrong problem ("name 'logger' is not defined" rather than the provider that actually blew
up). Doubly silent: nothing was ever logged either, because the logging call was the thing
that crashed.
Found by pyflakes while carving server.py (R3b). It survived because NOTHING exercised the
failure path — no test ever had a provider raise. That is the whole reason this class of
bug is invisible: it lives only on error paths, so the suite is green and the feature is
broken exactly when it matters.
tests/test_tuning_provider_isolation.py is that path:
* a raising provider must not lose the HEALTHY providers' tunings, nor the defaults
* and the failure must actually be LOGGED — swallowing is only acceptable if it reports
Bite-tested: restoring `logger` fails both.
(The log assertion attaches caplog's handler to the feedBack logger directly. It sets
propagate=False, so pytest's root capture sees nothing from it — test_plugins.py has a
capture_logger() for this, but it is not importable here: pyproject pins pythonpath to
[".", "lib"], so `tests` is not a package. Three lines beat churning 21 call sites in an
unrelated file to convert that helper into a fixture.)
pytest 2410, pyflakes 0 undefined names in server.py, Codex 0.
Closes #899
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
70dbe45e27
|
refactor(server): carve builtin-content seeding into lib/builtin_content.py (R3b) (#900)
lib/builtin_content.py (321 lines moved). server.py 2,418 -> 2,098.
The calibration/diagnostic sloppaks and the starter library: _copy_builtin_packs,
_write_builtin_pack, the two seed helpers, their source tables, and the seed marker.
━━━ THE ONE SIGNATURE CHANGE, AND WHY THE CARVE IS UNSAFE WITHOUT IT ━━━
server.py has:
def _feedBack_server_root() -> Path:
return Path(__file__).resolve().parent
That is correct IN server.py: the repo root in dev, resources/feedBack when bundled — the
tree that actually holds docs/ and data/.
Move that body into lib/ unchanged and it keeps working, silently, and returns lib/. There
is no docs/diagnostics under lib/, so every seed would find nothing, log "source missing"
at debug, and return. Nothing raises. Nothing fails. The starter library simply never
appears, and the calibration sloppak is never seeded — on a fresh install, in the field.
A verbatim move whose MEANING changed because __file__ did.
So this module cannot compute a root: `server_root` is a PARAMETER, and server.py — the
only place that legitimately knows where it lives — passes it in. The trap is now
structurally impossible rather than merely avoided. (_copy_builtin_packs already took the
root that way; the two seed helpers now do too.)
Everything else is byte-identical. CONFIG_DIR is read late as appstate.config_dir and the
DLC root through dlc_paths._get_dlc_dir — the same seam every router in lib/routers/ uses,
late-bound because tests monkeypatch it.
━━━ PYFLAKES FOUND THREE MISSING IMPORTS THE TESTS WOULD HAVE FOUND ONE AT A TIME ━━━
The moved code uses `secrets`, `stat` and `tempfile`; none was in my import block. Each is
a NameError on a live path. `python3 -m pyflakes` names all three in one shot — this is the
Python twin of the no-undef gate that guarded every frontend carve, and it should run on
every server.py slice from here.
It also flagged a PRE-EXISTING one I deliberately did not touch: server.py's
TuningProviderRegistry.get_merged() calls `logger.exception(...)` in an except handler and
there is no `logger` in the module (it is `log`). So a raising tuning provider takes down
the merged-tunings call for everyone, with a NameError naming the wrong problem. Filed as
issue #899 rather than smuggled into a carve whose whole value is being behaviour-neutral.
The constants lost their underscore prefix: they cross a module boundary now (the seed
tests read them), so `_BUILTIN_STARTER_SOURCES` was a lie.
pytest 2397, pyflakes 0, Codex 0. Guarded by the plugin_context contract test (#898).
Refs #48
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
1cef01d02c
|
test(plugins): pin the plugin_context contract before carving server.py (R3b) (#898)
tests/test_plugin_context_contract.py (3 tests). No production code changes. server.py is about to be carved apart around startup_events(), and `plugin_context` — the 20-key dict handed to every plugin's setup() — is built inline inside it. Issue #48 flagged this while planning the split and asked for exactly this guard: "Plugin context[...] are passed as live references into already-loaded plugins. Refactoring must preserve the exact callables — moving them to a new module is fine, but renaming or wrapping them breaks third-party plugins. We'd want a 'plugin context unchanged' assertion in CI." It never got written. Writing it FIRST, because a key silently dropped or renamed by a move is invisible to every other test in the suite — nothing in-tree reads most of these — and would break plugins at runtime, in the field. This is the backend's version of the window contract, and the frontend carve just taught me what that costs: 43 of library.js's exports were referenced ONLY from app.js's top-level window block, invisible to any call-graph scan, and trusting the scan would have shipped a dead A-Z rail with CI fully green. A contract only external code reads has to be pinned BY NAME, before the move, not after. THE SURFACE IS BIGGER THAN server.py's DICT. Shipped plugins read `log` and `load_sibling`, and neither is in it — plugins/__init__.py layers them on per-plugin. A test pinning only server.py's 18 keys would have missed both. ━━━ CODEX CAUGHT ME WRITING A VACUOUS ASSERTION ━━━ My first identity test built a dict locally and called setup() on it — asserting `dict(x)['k'] is x['k']`, which is trivially true and blind to everything the loader does. [P2], and correct. It now drives the REAL plugins.load_plugins() with a probe plugin, which matters: the loader DOES deliberately wrap one key (register_library_provider is scoped per-plugin so a plugin cannot forge owner attribution and impersonate another). The test pins that single intentional exception so it cannot quietly become two. Codex then caught [P2] number two: my hand-rolled teardown restored only PLUGINS_DIR and LOADED_PLUGINS, while load_plugins() also mutates sys.path, sys.modules and PENDING_PLUGINS — order- and environment-dependent. tests/test_plugins.py already had a fixture that does this properly, so `reset_plugin_state` moved to tests/conftest.py: ONE copy, shared, rather than a second that will drift. BITE-TESTED IN FIVE DIRECTIONS — drop a key, rename a key, drop a per-plugin key, wrap extract_meta in the loader (all key names intact, identity broken), and remove the register_library_provider scoping (the impersonation guard). Each fails. pytest 2399, Codex 0. Refs #48 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
756588678b
|
fix(plugins): make a module plugin actually re-evaluate on reload (#879) (#897)
A plugin reload silently did nothing for scriptType:"module" plugins. ES modules are evaluated ONCE PER URL PER DOCUMENT, so re-inserting a <script type="module"> whose src the module map has already seen fires `load` without re-running the body — and the loader then recorded the reload as applied. A no-op that reported success. THE ISSUE UNDERSTATES IT. #879 says "upgrades are fine — a new version yields a new URL". That is true of screen.js and FALSE of the plugin. I drove a real browser through install(1.0.0) -> upgrade(1.1.0) -> rollback(1.0.0), counting evaluations of src/main.js: ONE. Not three, not two. The upgrade re-runs the one-line screen.js shim at its new ?v= URL; the shim does `import './src/main.js'`; a relative specifier resolves against the base URL WITH THE QUERY DROPPED; that is the same URL as before; the module map hands back the already-evaluated v1.0.0 module. The plugin's own code never re-ran. Busting the entry point cannot fix this, whatever token you hang off it. So the token goes in the PATH: /api/plugins/<id>/g/<n>/screen.js. From there './src/main.js' resolves to /api/plugins/<id>/g/<n>/src/main.js — every relative import inherits it, at every depth, for free. No import-specifier rewriting (which could never see `import(expr)` anyway). Same browser drive after the fix: THREE evaluations. Keyed on the plugin ID, not id@version: EVERY re-load of a module plugin needs a fresh path, not just a rollback. First load keeps the stable ?v= URL, so the ETag/304 live-edit caching the R0 rails depend on is untouched. Classic-script plugins are not affected and never take a /g/ path. ━━━ A PATH REWRITE, NOT TWO MIRRORED ROUTES ━━━ Codex caught this, and it was right. The token shifts the BASE URL, so EVERYTHING the module graph resolves relatively moves with it — not only imports. `new URL('../assets/worklet.js', import.meta.url)` from /api/plugins/x/g/1/src/main.js resolves to /api/plugins/x/g/1/assets/worklet.js. Mirroring only screen.js and src/ would have fixed imports and 404'd every asset, worklet and wasm file the graph reaches — and would have broken again the next time someone added a plugin route. So the /g/<token> segment is STRIPPED BEFORE ROUTING. Every plugin route, present and future, works under the prefix with no extra wiring. The token is opaque and never joined into a filesystem path, so containment still rests entirely on the same safe_join. Codex then caught a [P3] in that: eagerly re-encoding raw_path with latin-1 raises UnicodeEncodeError on a valid plugin file like src/工具.js, 500ing a request the plain route serves fine. raw_path is informational and Starlette routes on scope["path"], so the mutation is simply gone — and leaving raw_path as the client sent it is more truthful for logs anyway. TESTS. tests/js/plugin_module_rollback.test.js (5) + 8 in test_plugin_src_route.py: identical bytes under the prefix, the whole graph one and two levels deep, ASSETS (the Codex [P2]), every plugin route, non-ASCII filenames (the [P3]), an opaque token, and containment asserted as PARITY with the un-prefixed route rather than a guessed 404 — `../screen.js` legitimately 200s on both, because the URL normalises before routing. All bite-tested: reverting the fix fails the rollback tests, disabling the rewrite fails the asset tests. Two harnesses re-anchored on `script.src = _pluginScriptUrl(` — the URL literal they keyed on now lives in the helper, further down the file, so their slice ran off the end. node 1045, pytest 2404, ESLint 0, Codex 0. Closes #879 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
bd830328f0
|
refactor(app): carve the library out of app.js (R3a) (#896)
static/js/library.js (1,988) + static/js/library-state.js (29) — bodies VERBATIM.
app.js 6,313 -> 4,451.
THE BIGGEST SLICE OF THE CARVE: 145 declarations, ~1,900 lines, 30% of what was left.
The grid, the artist tree, the A-Z rail, filters, pagination, selection, favourites, the
scan banner, and the library-provider plumbing.
A LOW module: it imports only leaves (./dom.js, ./format.js, ./library-state.js,
./tuning-display.js — all four import nothing themselves) and needs ZERO host hooks. It
calls nothing in app.js. That is not luck; it is why this cluster was picked. Two entry
points that WOULD have dragged the playback core in were left behind in app.js:
* syncLibrarySong reaches showScreen/playSong
* _handleLibArrowNav Enter on a selected row plays the song
Both are one hop from the library, and app.js is the root, so it imports from both sides
for free. Pulling them in swallows playSong, showScreen and the whole remaining core — I
measured it: the closure jumps from 145 declarations to 189.
library-state.js holds exactly FIVE fields. An imported binding is read-only, and of the
library's outward bindings only these five are genuinely WRITTEN from outside — by
showScreen, deleteSongFromModal and syncLibrarySong, none of which can move in. The other
23 are read-only from outside, so they stay plain exports (ES live bindings mean app.js
still sees every reassignment).
━━━ THE EXPORT LIST NEARLY SHIPPED A DEAD A-Z RAIL ━━━
59 exports — and 43 of them CANNOT be found by a call-graph scan. They are referenced only
from app.js's TOP-LEVEL statements: the Object.assign(window, {...}) contract and the
scattered window.X = X lines, which live outside every function, so a closure walk over
declarations never sees them. Among them are the four handler names app.js composes AT
RUNTIME into onclick="" strings — filterTreeLetter, filterFavTreeLetter, goTreePage,
goFavTreePage — the library A-Z rail and its pagination. No static tool can see those at
all. Had I trusted the call-graph, the rail would have died silently on click with nothing
failing in CI.
━━━ AND MY OWN SCANNER LIED ━━━
The cycle-risk pass reported "(none)" for this carve. It was wrong, and it could not have
been right: a dangling `else if` bound to an inner `if` instead of the outer chain, so its
`imported` map was ALWAYS empty and the check reported clean no matter what. A guard that
cannot fail is worse than no guard. Fixed, and it then found the real edges — dom.js,
format.js, tuning-display.js, library-state.js. All four are leaves, so the carve is
genuinely acyclic; I just now know it instead of assuming it.
(The AST rewriter had its own trap: `MAP[name]` with an object literal and name ===
'constructor' hits Object.prototype.constructor — truthy — and it happily rewrote
`constructor(id)` into `L.function Object() { [native code] }(id)`. Every identifier in
the file is looked up, so the lookup must not see the prototype chain. It is a Map now.)
TESTS. legacy_shim_hits SPLIT (loadLibraryProviders + setLibraryProvider -> the module;
syncLibrarySong stayed in app.js). v3_library_refresh now reads app.js AND the module,
rather than being re-pinned to whichever file happens to hold the emit this week.
VERIFIED. A/B against origin/main in two browsers: the whole window contract, cards render,
grid/tree/sort/filter/clear round-trip — and, specifically, the A-Z rail: 28 onclick
handlers composed at runtime, identical on both, and a real .click() on a letter works.
IDENTICAL on all 33 + 7 probes, no new page errors.
pytest 2396, node 1040/1040, host contract 2/2, ESLint 0 (no-cycle clean).
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
09f7e450a5
|
refactor(app): give formatTime a home — a leaf format.js, one fewer host hook (R3a) (#895)
static/js/format.js (17). One function. Retires the formatTime hook: 12 -> 11. WHY A MODULE FOR ONE FUNCTION. formatTime was a host hook — loops.js and section-practice.js both reached back through the seam for it. It is ALSO, by pure accident of who calls it, inside the dependency closure of the library carve that comes next. Leaving it there would have made loops.js and section-practice.js import the LIBRARY in order to format a timestamp — nonsense, and a cycle waiting to happen. Same rule as the transport carve: a hook is a cycle you agreed to live with; an import is a dependency you actually have. formatTime has a real owner. It just isn't app.js, and it certainly isn't the library. Give it a home and both consumers import it directly. A leaf on purpose. Anything else that turns out to be a shared pure formatter belongs here too; nothing does yet (I checked — formatBadge, _safeImageUrl and _fetchJsonOrThrow have no callers outside the library), so nothing else is here. node 1040/1040, host contract 2/2, ESLint 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
8bec8d2466
|
refactor(app): carve the playback transport out of app.js — and RETIRE 8 host hooks (R3a) (#894)
static/js/transport.js (377) — bodies VERBATIM. app.js 6,643 → 6,316.
THIS IS THE FIRST CARVE THAT SUBTRACTS HOOKS INSTEAD OF ADDING THEM.
Every carve before this one added host hooks: a module pulled out of app.js still had
to call back into it. But four modules were all reaching through the seam for the SAME
handful of names — _audioSeek, _audioTime, setPlayButtonState, _songEventPayload,
jucePlayer. Those names have an owner, and it isn't app.js. Give them one, and the
consumers import them directly:
count-in.js 5 hooks -> 0 (host import deleted)
juce-audio.js 4 hooks -> 0 (host import deleted)
loops.js 6 hooks -> 4
section-practice.js 10 hooks -> 7
----------------------------------------------------------
configureHost() 20 hooks -> 12
A hook is a cycle you agreed to live with. An import is a dependency you actually have.
Prefer the import whenever the name has a real owner.
_audioSeekGen now stays PRIVATE. It has exactly one writer — _resetAudioSeekState(),
which moved with it — so readers get audioSeekGen() and nobody outside can desync it.
Strictly better than the hook it replaces, which handed out a getter and left the writer
behind in app.js.
THE SCAN HAD A HOLE, AND IT BIT. Picking the carve by dependency closure over app.js's
own top-level decls said this cluster was downward-closed. It wasn't:
_currentPlaybackSnapshot reads loopA/loopB — which live in ./js/loops.js, and loops.js
imports transport. The scan saw nothing, because loopA STOPPED BEING an app.js decl the
moment loops.js was carved out. Any dependency scan of a partly-carved monolith has to
resolve the imports too, or it will confidently hand you a cycle. Added that pass; it
found exactly one back-edge, and _currentPlaybackSnapshot stays in app.js (as does
restartCurrentSong, which calls _cancelCountIn). app.js is the root — it imports both
sides for free.
TESTS. Four harnesses retargeted (play_button_reroute_guard, song_event_payload,
song_seek -> transport.js; playback_app_adapter SPLIT, since
_installPlaybackTransportAdapter stayed behind).
The two CENSUS tests — "≥8 song:* emit sites", "every seek callsite passes a reason" —
now scan app.js AND every static/js/*.js, not one file. Pointed at a single file, their
count silently shrinks as code leaves, which reads as "someone deleted an emit" or, worse,
passes while genuinely missing sites. Both bite-tested: stripping a _songEventPayload()
from an emit and adding a reason-less _audioSeek() each fail the suite.
VERIFIED. A/B against origin/main, real song, real playback: song:play payload is exactly
{audioT, chartT, perfNow, time}; song:seek carries reason "seek-by" with finite from/to;
all five song:* events fire; seekBy advances the clock; restartCurrentSong returns to zero;
the play button's aria-pressed tracks state. IDENTICAL on all 21 probes, zero page errors.
pytest 2396, node 1040/1040, host contract 2/2, ESLint 0 (no-cycle clean), Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
8d0e270345
|
refactor(app): carve the JUCE/desktop audio shims out of app.js (R3a) (#893)
* refactor(app): carve resume-session out of app.js (R3a)
static/js/resume-session.js (157) — the snapshot taken when you leave a song and the
pill that offers it back. Bodies VERBATIM. app.js 7,727 → 7,601.
Fifth slice out of the strongly-connected core. ONE hook (playSong) + a
currentFilename getter.
S.pendingResume JOINS THE CONTAINER — on demand, exactly as intended. app.js WRITES
it (playSong({ resume }) arms it; the song:ready listener consumes it) while this
module reads it, so it cannot be a plain export: an imported binding is read-only.
Same reason isPlaying is there. The container grows one field per carve that needs
it, never speculatively.
THE CONTRACT TEST CAUGHT THE MISSING HOOK, again on a path nothing executes:
"playSong is read by a module but never wired by app.js — it would throw at runtime".
Second time it has caught a real wiring gap the moment it appeared.
A REAL TRAP, worth remembering: I first did the S.pendingResume rewrite by feeding
acorn's identifier RANGES from node into python, and it corrupted the file
(`_pS.pendingResume null;`). **Acorn's offsets are UTF-16 code units; Python's string
indices are code points.** static/app.js contains emoji, so every offset past one
drifts. Do an AST-driven rewrite in the SAME language that produced the offsets.
`node --check` caught it; a silent version of that bug is very easy to imagine.
VERIFIED. A/B against origin/main in two browsers, real song: the window API
(resumeLastSession / _snapshotResumeSession / _readResumeSession /
_clearResumeSession), snapshot, read-back, and clear — IDENTICAL, zero page errors.
HONEST LIMIT: my probe never got the snapshot to actually PERSIST (there is a guard
beyond the 3s minimum position that a scripted playSong does not satisfy), so that
path is verified only as identical-to-main, not as observed-working. The real
coverage is tests/browser/resume-session.spec.ts, which drives the flow properly.
Zero harnesses broke. pytest 2396, node 1040/1040, ESLint 0 (no-cycle clean),
tailwind clean, Codex 0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(app): carve the JUCE/desktop audio shims out of app.js (R3a)
static/js/juce-audio.js (994) — bodies VERBATIM. app.js 7,603 → 6,643.
THE LARGEST SINGLE SLICE of the whole carve phase: 960 lines, ~13% of what was left.
Three self-installing IIFEs:
_installJuceEngineRoutingWatcher (444) routes a song to the JUCE engine or HTML5 as
the desktop output enters/leaves exclusive/ASIO
_installRendererBusFeeder (337) feeds the highway renderer bus from whichever
transport is actually running
_installJuceAudioElementShim (156) patches audio.play/pause so the rest of the app
keeps talking to the <audio> element while JUCE
owns the transport
They EXPORT NOTHING — all three publish through `window.*` (_juceMode,
_reevaluateJuceRouting, _reevaluateRendererBus, …). So app.js needs only a
side-effect import plus the one binding it actually uses
(_resetJuceAudioShimChain, which the shim IIFE assigns).
THE ORDERING QUESTION, CHECKED RATHER THAN ASSUMED. Importing this module runs the
IIFEs EARLIER than before: imports evaluate ahead of app.js's body, and therefore
ahead of configureHost(). A hook read at IIFE-execution time would THROW. So I walked
the AST at IIFE-body depth to see what they actually touch when they run: nothing but
listener registration, and `audio.play`/`audio.pause` patching — and `audio` is itself
an imported module now. Verified in the browser: both are patched on the carved build
exactly as on main, which proves the shim installs correctly at its new, earlier point.
(Had I got this wrong, host.js throws loudly rather than silently misbehaving — which
is the whole reason it has no no-op defaults.)
VERIFIED. A/B against origin/main in two browsers: the entire window.* surface the
IIFEs publish (_juceMode, _juceOutputIsExclusive, _reevaluateJuceRouting,
_reevaluateRendererBus, _clearJuceRerouteMemo), audio.play/pause patched, a real song
loading and togglePlay driving the public mirror — IDENTICAL, zero page errors.
Harnesses: juce_engine_reroute (19 tests) + renderer_bus_feeder (13) slice the IIFEs by
signature — retargeted, and each sandbox gains a `host` object routed at its EXISTING
stubs so every assertion holds unchanged. test_plugin_runtime_idempotence is SPLIT: 3 of
its 4 source-asserts stayed in app.js, the sm.emit('song:resume') one moved.
pytest 2396, node 1040/1040, ESLint 0 (no-cycle clean), tailwind clean, Codex 0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
dc429ecd16
|
refactor(app): carve the player controls out of app.js (R3a) (#891)
static/js/player-controls.js (229) — the speed + mastery sliders and the four
playback-preference reads (autoplay-exit, up-next, countdown-before-song,
confirm-exit). Bodies VERBATIM. app.js 7,914 → 7,727.
The fourth slice out of the strongly-connected core, and by far the easiest: ONE
hook (handleSliderInput) and NO shared mutable state. The three groups are the same
surface — the controls under the highway — and the preference reads are the
one-line localStorage lookups half of app.js consults before deciding whether to
auto-start, show the Up Next pill, run a count-in, or confirm on exit. They travel
with the controls that set them.
Zero missed members on the first build (the no-undef pass was clean), which is the
first time that has happened in this phase.
TWO HARNESSES ARE SPLIT, and both taught something:
* speed_reset spans BOTH files — playSong (app.js) resets the speed controls
(module). Its presence GUARDS still read `src.includes('function setSpeed')`
against app.js, so once the code moved they silently evaluated FALSE and the
helpers were quietly dropped from the sandbox. A guard that disables itself is
worse than no guard. Repointed at the file the code actually lives in.
* Its `host.handleSliderInput` stub had to route at the sandbox's EXISTING spy,
not a fresh `() => {}`. The test asserts the slider was actually refreshed
(`deepEqual(__sliderInputs, ['speed-slider'])`); a fresh stub swallows the call
and the assertion passes VACUOUSLY. Same failure mode as a no-op host default —
the thing this whole seam design exists to prevent.
* autoplay_exit is split too: _autoplayExitEnabled moved, but the auto-exit
machinery around it (_clearAutoExit, holdAutoExit, _resolvePlayerOrigin) stayed.
VERIFIED. A/B against origin/main in two browsers, real song: setSpeed(0.75) ->
playbackRate 0.75; applySpeedPreset(100) -> 1; the speed slider; setMastery;
setAutoplayExit / setCountdownBeforeSong / setShowUpNext — IDENTICAL, zero page errors.
pytest 2396, node 1040/1040, ESLint 0 (no-cycle clean), tailwind clean.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
11f8c36b61
|
refactor(app): carve count-in (and the song-credits overlay) out of app.js (R3a) (#890)
static/js/count-in.js (389) — bodies VERBATIM. app.js 8,223 → 7,913. The third slice out of the strongly-connected core, and the first that WRITES shared state rather than only reading it. #889's container is what makes it possible. imports: loops (setLoop/loopA/loopB — a count-in inside an A-B loop must begin at A), audio-el, player-state, host hooks : _audioSeek, setPlayButtonState, _songEventPayload, togglePlay + a jucePlayer getter Nothing imports count-in back — app.js and section-practice both reach it through the seam — so the graph stays acyclic. app.js's autoplay path used to reach IN and set this module's credits timers itself (_creditsTimer, _creditsHideOnPlay) and read _countingIn. It cannot now, and should not have to, so the module exports the OPERATIONS instead — armCreditsHideOnPlay(), scheduleCreditsHide(), holdCreditsThen(start), isCountingIn() — and owns its own timer invariants. Third time this has happened (section-practice's resetSelection, loops' state) and each time the constraint produced better code than was there before: the module keeps its own promises instead of trusting a caller 6,000 lines away to zero the right fields. THE no-undef GATE FOUND FIVE MISSED MEMBERS, one at a time: showSongCreditsOverlay and startSongCountIn (my name regex matched startCountIn, not startSongCountIn), then _creditLineLabel, _CREDITS_MAX_MS, and _CREDIT_ROLE_VERBS. A call-graph closure does not see a const table; only the undefined-symbol pass does. AND A REAL TRAP: I computed _CREDIT_ROLE_VERBS's span against the ALREADY-MODIFIED app.js and applied it to the clean one — the line numbers had drifted, so the slice would have cut somewhere else entirely. Recomputed every span from the clean file with acorn. Never carry line numbers across an edit. VERIFIED. A/B against origin/main in two browsers with a real song: playback state, the public feedBack.isPlaying mirror, audio position, cancel-count-in — IDENTICAL, zero page errors. Unit coverage moved with the code: loop_restart's count-in cancellation-token test and the 5 song_credits_overlay tests now read count-in.js; loop_restart's sandbox gains a `host` object routed at its EXISTING stubs, so every assertion is unchanged. HONEST LIMIT: I could not make the count-in OVERLAY actually render headlessly — its autoplay path needs a fresh-load _pendingAutostart that a scripted playSong() never arms. Behaviour is identical to main on every probe and the unit tests cover the logic, but the on-screen 1-2-3-4 and the credits card want a human look. pytest 2396, node 1040/1040, ESLint 0 (no-cycle clean), tailwind clean, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
5fb28d5c5a
|
refactor(app): lift the shared player state onto a container (R3a) (#889)
static/js/player-state.js — one exported object, two fields. app.js's 70 reference
sites rewritten. Provably a no-op; nothing shrinks.
WHY NOW. Every slice carved out of app.js so far only ever READ the state it shared
(loopA/loopB, _audioSeekGen, currentFilename), so a read-only getter hook was enough
and no container was needed — twice I checked and twice I got away with it. That
runs out at count-in: it genuinely WRITES `isPlaying` (it starts and stops playback,
4 sites) and `lastAudioTime` (2). `import { isPlaying }` then `isPlaying = true`
THROWS — an imported binding cannot be assigned to. So the state has to live on an
object: `S.isPlaying = true` is a property write, and works from any module holding
the same S. Same shape stems, studio, and editor all converged on.
DELIBERATELY SMALL. app.js has ~104 top-level `let` scalars; lifting all of them is
a ~977-site rewrite for no benefit, because most are private to one cluster and
travel with it. Only what a carved module must WRITE goes here. Add on demand.
THE REWRITE IS AST-DRIVEN, NOT TEXTUAL — and that is not fussiness. Of 100 textual
occurrences of these two names, only 70 resolve to the module binding:
* 22 are member accesses (`someObj.isPlaying`, `window.feedBack.isPlaying`)
* 4 are the LOCAL PARAMETER of `function setPlayButtonState(isPlaying)` — a blind
replace yields `function setPlayButtonState(S.isPlaying)`
* 1 is an object key
* 2 are shorthand properties `{ isPlaying }`, which must become
`{ isPlaying: S.isPlaying }` — and acorn gives a shorthand's key and value the
SAME range, so rewriting both produced `isPlaying: S.isPlaying: S.isPlaying`
until I deduped by range
A find-and-replace corrupts all 29. The rewrite walks the AST, skips shadows, member
properties and keys, and replaces identifier RANGES.
`window.feedBack.isPlaying` — the PUBLIC mirror — is a different thing and is
untouched. Two test sandboxes stub it; those were left alone deliberately.
VERIFIED WITH REAL PLAYBACK. A/B against origin/main in two browsers, real song:
togglePlay -> the public mirror goes true -> false -> true across two toggles, the
audio element's paused state follows, seekBy works — IDENTICAL, zero page errors.
Harnesses: 8 vm-sandbox suites slice playback code out of app.js and now see
S.isPlaying — juce_engine_reroute, loop_restart, play_button_reroute_guard,
playback_app_adapter, song_restart, song_seek, speed_reset, and the python
idempotence source-assert. Each gets the same container in its sandbox; every
assertion is unchanged.
pytest 2396, node 1040/1040, ESLint 0, tailwind clean, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
cb236e6c04
|
refactor(app): carve the A–B loop out of app.js (R3a) (#888)
static/js/loops.js (261) — bodies VERBATIM. app.js 8,421 → 8,224.
The second slice out of the strongly-connected core.
It OWNS the loop state — loopA, loopB, _loopMutationGen. Nothing outside writes
them: restartCurrentSong() looked like it did, but it declares its own local `let
loopA/loopB` shadows, so the module-level bindings only ever change in setLoop /
setLoopStart / setLoopEnd / clearLoop. All four move here. No state container.
DIRECTION IS THE WHOLE DESIGN. loops and section-practice are mutually dependent —
the SCC in miniature. clearLoop() must drop section-practice's selection, and
practiceSection() must call setLoop(). Both edges cannot be imports or no-cycle
(rightly) rejects it. So:
section-practice -> reaches loops through the HOST SEAM (host.setLoop, …)
loops -> imports section-practice DIRECTLY
section-practice is the higher-level feature — a consumer of loops, not the reverse
— so it is the one that takes the indirection. app.js hands the loop module's
exports across into the seam for it. Graph stays acyclic; no-cycle passes.
THE CONTRACT TEST EARNED ITS KEEP IMMEDIATELY. It failed on the first build with
"these hooks are wired by app.js but no module reads them: playSong". My dependency
scan had counted a mention of playSong() inside a COMMENT in loops.js as a real
call. Wired but unused is precisely the "fossil of a rename" case the test exists
for — and it caught it on a path no test executes.
VERIFIED BY DRIVING BOTH SIDES OF THE SEAM. A/B against origin/main in two browsers,
real song loaded:
* setLoop(5,12) -> true; getLoop() -> 5,12 — IDENTICAL
* clearLoop() (loops -> section-practice, a direct import) -> getLoop() ->
null,null — IDENTICAL
* onPhraseNext() (section-practice -> loops, ACROSS THE SEAM) -> ok — IDENTICAL
* loadSavedLoop / saveCurrentLoop / deleteSelectedLoop on window — IDENTICAL
* zero page errors either side. An unwired hook throws, so a live app is itself
proof the seam is wired.
Harness: loop_api extracts the loop helpers by signature — retargeted to loops.js,
`export` stripped for the vm sandbox, and the sandbox's existing _audioSeek /
_audioTime / formatTime spies are now routed through a `host` object so every
assertion holds unchanged, just through the indirection the real code uses. It is
SPLIT: one test still reads app.js for the window.feedBack API surface, which stayed.
pytest 2396, node 1040/1040, ESLint 0 (no-cycle clean), tailwind clean.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
64f04565e2
|
refactor(app): the host seam + carve section practice out of app.js (R3a) (#887)
static/js/host.js (99) + static/js/section-practice.js (1,214).
app.js 9,461 → 8,409.
THE FIRST SLICE OUT OF THE STRONGLY-CONNECTED CORE. What is left in app.js is not
a tree, it is a cycle: seeding a dependency closure from section-practice, from
loops, from count-in, or from the JUCE seek shim all return the SAME 178-function
set, and setLoop() and practiceSection() call each other directly. No closure-based
carve can cut it at any seed. So it is cut BY NAME, and the calls back into app.js
go through a host seam.
61 functions + its own 24 _sectionPractice*/_sectionParents* scalars (read nowhere
else) move out. 11 hooks come back in. 4 of those are read-only GETTERS —
loopA/loopB/_audioSeekGen/_loopMutationGen are only ever READ here, never written,
so app.js keeps owning them and NO state container is needed (a 977-site lift
avoided).
app.js used to reach IN and reset the module's state by hand (clearLoop() zeroed the
selection; changeArrangement() invalidated the parent count). It cannot now — an
imported binding is read-only — so those are exported as resetSelection() and
invalidateParentCount(). Strictly better: the module owns its own invariants instead
of trusting two callers on the far side of the file to zero the right three fields.
═══ THE SILENT-NO-OP PROBLEM, SOLVED ═══
The obvious host seam is an object of no-op defaults. That is a TRAP and we walked
into it once: the plugin loader's seam defaulted populateVizPicker to `() => {}`, so
a dropped wiring line would have left the viz picker quietly not refreshing with NO
test, boot check, or bot noticing. Two layers stop it here:
1. RUNTIME — host.js is a Proxy with NO defaults and NO stubs. Reading an unwired
hook THROWS. An unwired hook cannot degrade into a no-op because there is
nothing to degrade INTO. configureHost() also rejects a non-function at WIRE
time, and refuses to run twice.
2. STATIC — tests/js/host_contract.test.js asserts the hooks the modules USE are
exactly the hooks app.js WIRES. This is the layer that matters: a runtime throw
only fires if the broken path executes, and the whole danger of a seam is the
paths that never run in a smoke test. VERIFIED TO BITE in all three drift
directions: drop a hook from configureHost -> fails; rename host.setLoop in the
module -> fails; wire a hook nobody uses -> fails.
Writing that guard took three tries and each failure is instructive: (a) the
configureHost regex anchored `});` at column 0, ran past the indented close, and
swallowed app.js's 66-name window contract — 77 "hooks"; (b) an import-stripping
regex with `[\s\S]*?` ate 14,000 characters INCLUDING the drift the bite test was
meant to catch — a guard with a hole is worse than no guard, because you trust it;
(c) `host.js'` in the import path backtracked from `js` to a "hook" called `j`.
The bite tests are what surfaced all three.
CODEX FOUND A REAL RACE [P2]. configureHost() was inside the async boot function,
after several awaits — but the window handlers (onPhraseNext, …) go live during
app.js's SYNCHRONOUS module evaluation. A user clicking one in that window would hit
"[host] … was read before configureHost() ran". It is now a bare top-level statement
sitting immediately before the window contract, so the seam is always wired before a
handler can be reached. Verified live: invoking a handler 1.2s in — well before the
boot awaits settle — works.
VERIFIED. A/B against origin/main in two browsers with a REAL song loaded: popover
toggle, practice-mode change, phrase-next, and clearLoop (all of which cross the seam
— setLoop/clearLoop/_audioTime/loopA/loopB) — IDENTICAL, zero page errors. Since an
unwired hook throws, a live app is itself proof the seam is wired.
Harnesses: section_practice_dismiss retargeted; loop_api's clearLoop sandbox gains a
resetSelection SPY (not a stub) and ASSERTS it fires — the guarantee is still tested,
just through the seam.
pytest 2396, node 1040/1040, ESLint 0 (no-cycle clean), tailwind clean, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
f53d566dbc
|
refactor(app): give the <audio> element a module of its own (R3a) (#886)
static/js/audio-el.js — one exported const. app.js's diff is 5 lines.
This is a HINGE, not a carve: nothing shrinks, but almost everything left in
app.js is blocked behind it.
WHY. `audio` is `document.getElementById('audio')` with 162 references in app.js
and 173 outside any one cluster. Every remaining cluster measured — settings,
app-updates, count-in (208 fns), exit-confirm (212), library-render (220) — lists
`audio` among its inbound symbols, because they all touch playback and playback
reaches for the element directly. A module that needs it cannot import app.js to
get it (that closes a cycle and fails import-x/no-cycle), so today the only way to
carve any of them would be a host seam — the exact thing #878 had to build and
#880 had to tear out.
WHY IT'S SAFE. `audio` is a `const` and is NEVER reassigned anywhere in core, so a
read-only import binding is exactly right and no state container is needed. The
162 call sites are untouched — the binding keeps its name, it is just imported
instead of declared. (Contrast the reassigned scalars — isPlaying, _avOffsetMs —
which CANNOT be shared this way: an imported binding cannot be written to. Those
still need containers, and that is the next problem, not this one.)
TIMING. app.js is <script type="module">, so it evaluates after the HTML is parsed
and its imports evaluate just before its body — the same moment app.js used to run
this exact lookup. If the element had not been in the document, `audio` would be
null and app.js's top-level `audio.addEventListener(...)` calls would throw and
kill the module. They don't.
VERIFIED WITH REAL PLAYBACK, not a boot check. A/B against origin/main in two
browsers: app alive with zero page errors (which is itself the proof the import
resolved), #audio is an AUDIO element, togglePlay/seekBy live, and playSong() on a
real library song sets audio.src and the element reports a duration — IDENTICAL on
both sides.
pytest 2396, node 1038/1038, ESLint 0 (no-cycle clean), tailwind clean, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
b5dd585d25
|
refactor(app): carve settings backup + plugin updates out of app.js (R3a) (#885)
Two leaves, one PR. app.js 9,651 → 9,457.
static/js/settings-io.js (155) — exportSettings + importSettings, the Settings
backup bundle. Imports nothing. The two-phase rationale comment (server first and
atomic; then a best-effort localStorage merge) is the contract and moved with the
code.
plugin-updates → INTO static/js/plugin-loader.js, not a module of its own.
checkPluginUpdates + updatePlugin are plugin MANAGEMENT; they belong with the code
that loads plugins. A new file for 50 lines would have been a file for its own
sake.
All four are inline handlers on the Settings screen and already in app.js's window
contract, so app.js re-exposes the imported bindings unchanged.
VERIFIED BY DRIVING BOTH FLOWS. A/B against origin/main in two browsers:
* checkPluginUpdates() -> hits the API and settles the button back to
"Check for Updates" — IDENTICAL
* exportSettings() -> POSTs /api/settings/export and writes
"Exported feedBack-settings…" to #backup-status — IDENTICAL (fetch intercepted
so the assertion is on the real call, not a stub)
* all four resolve on window — IDENTICAL
* zero console/page errors either side
Zero harnesses broke. pytest 2396, node 1038/1038, ESLint 0, tailwind clean, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
d47883c5e5
|
refactor(app): carve the tuning-display helpers out of app.js (R3a) (#884)
static/js/tuning-display.js (228 lines) — bodies VERBATIM. app.js 9,838 → 9,650.
A LEAF: imports nothing.
Tuning NAME resolution (Drop D / Eb Standard / raw-offset fallback), bass
detection, effective string count, and the target FREQUENCIES + note names the
tuner checks against. Pure functions over a small MIDI/note-name table; the 3
_TUNING_* tables are read nowhere else and move in.
NOT A SLICE — a node-level extract. The span 2309-2535 INTERLEAVES the functions
with the `window.*` / `window.feedBack.*` assignments that publish them, and one
of those is `window.feedBack = window.feedBack || {}` — the BUS BOOTSTRAP, not
tuning code at all. Every ExpressionStatement stays exactly where it was; only the
16 functions and 3 tables move. app.js re-exposes the imported bindings from the
same lines, so the public surface and its ordering are untouched (constitution II
names window.feedBack).
app.js -> { plugin-loader, viz, diagnostics-export, dom, highway-colors,
tuning-display }
HARNESSES — 4 broke, and 3 of them broke in the SAME informative way: they sliced
app.js from `function isBassArrangement(` UP TO the marker
`window.feedBack.parseRawTuningOffsets = parseRawTuningOffsets;` — an end-marker
that (correctly) stayed behind in app.js. The module is now nothing BUT the tuning
helpers, so there is no block to slice: they read it whole and strip `export ` so
the vm sandbox still evaluates it as a script.
tuner_auto_open is SPLIT — its autoplay-gate test still reads app.js, so it keeps
APP_JS and gains TUNING_JS. Retargeting its path wholesale (my first attempt)
silently pointed the autoplay test at the wrong file.
VERIFIED BY DRIVING THE CONTRACT. A/B against origin/main in two browsers, through
the real window surface: displayTuningName -> "E Standard" / "Drop D" /
"Eb Standard", parseRawTuningOffsets('-2,0,0,0,0,0') -> [-2,0,0,0,0,0],
isBassArrangement, effectiveStringCount, displayTuningTargets, and
window.feedBack.displayTuningName / .songTuningContext — IDENTICAL on both, zero
console/page errors either side.
pytest 2396, node 1038/1038, ESLint 0, tailwind-fresh clean, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
ebbfc8da6f
|
refactor(app): carve the highway string-colours out of app.js (R3a) (#883)
static/js/highway-colors.js (601 lines) — bodies VERBATIM. app.js 10,415 → 9,837. Under 10k. DECOMPOSED, not sliced. The "settings" blob measured 47 fns / 19 inbound and was not carvable as-is. Seeding the closure from a FUNCTION (initHighwayColors) found only 18 fns and left 4 HWC_* constants used outside it — i.e. the seed was wrong, not the cluster. Re-seeding from the STATE (every function touching HWC_*/`_hwc*`) found the true cluster: 45 top-level nodes, lines 2751-3331, CONTIGUOUS, with ZERO foreign nodes inside the span. INBOUND: 0. EXPORTS: 2 (initHighwayColors, hwcInitSettingsUI). The other 43 symbols — the HWC_* tables, the 12 presets, the theme store, the share codec, the picker handlers, the window.feedBack.highwayColors facade — are used nowhere else in core and stay private. No inline on*= handlers here (the Settings buttons are wired by addEventListener inside hwcInitSettingsUI), so nothing needed re-exposing on window. The three bus listeners register inside initHighwayColors, which app.js calls — not at module top level — so no ordering change. THE no-undef GATE EARNED ITS KEEP. My closure said INBOUND=0; the module actually uses `uiPrompt` (the "name this theme" prompt). It was missed because uiPrompt is no longer an app.js DECLARATION — it's an IMPORT BINDING (from #882's dom.js), and I was collecting declarations only. `no-undef` with typeof:true caught it. => Lesson for the next carve: seed `tops` from ImportDeclaration bindings too. => And it VALIDATES carving dom.js early: this module just imports uiPrompt from it. Had dom.js still been stranded in app.js, this carve would have needed a host seam. app.js -> { plugin-loader, viz, diagnostics-export, dom, highway-colors } plugin-loader -> viz highway-colors -> dom viz, diagnostics-export, dom -> (leaves) VERIFIED BY DRIVING THE FACADE. A/B against origin/main in two browsers: window.feedBack.highwayColors installed, identical method surface, 12 presets, identical default slot colours, and a share-code encode→decode round-trip returning #112233 — IDENTICAL on both, zero console/page errors either side. Harnesses: highway_colors_facade + highway_string_colors retargeted (both brace-extract blocks out of the source by signature). pytest 2396, node 1038/1038, ESLint 0, tailwind-fresh clean, Codex 0. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Byron Gamatos
|
14b4058bc6
|
refactor(app): carve the DOM/modal primitives out of app.js (R3a) (#882)
static/js/dom.js (203 lines) — esc, _escAttr, _isElementVisible, _trapFocusInModal,
_confirmDialog, uiPrompt. Bodies VERBATIM. app.js 10,593 → 10,414.
A GATHER, not a slice — the six lived in six different places (108, 635, 659,
2617, 2623, 8892). They belong together because they are the BOTTOM of the UI
stack: `esc` alone has 25 call sites and `_escAttr` 23, and every later carve that
renders HTML will need them.
That is the actual point of doing this one now. Give them a home and the next
carve imports them; leave them in app.js and the next carve that renders HTML has
to invent a host seam to reach back into app.js — exactly the trap the
plugin-loader carve had to work around until the viz layer became a module. This
is the cheapest possible way to stop that recurring.
app.js -> { plugin-loader, viz, diagnostics-export, dom }
plugin-loader -> viz
viz, diagnostics-export, dom -> (nothing)
Zero imports. Six exports (every one is used outside the cluster).
VERIFIED BY DRIVING THE MODALS, not just booting — they are interactive, so a
green suite says little. A/B against origin/main in two browsers:
* window.uiPrompt / _confirmDialog / _trapFocusInModal all resolve
* uiPrompt() mounts its modal, accepts typed input, and resolves with the typed
value ('typed') — IDENTICAL on both
* _confirmDialog() mounts and resolves true on confirm — IDENTICAL
* zero console/page errors either side
pytest 2396, node 1038/1038, ESLint 0, tailwind-fresh clean, Codex 0.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
bfb31a8b89
|
refactor(app): carve the diagnostics-bundle export out of app.js (R3a) (#881)
static/js/diagnostics-export.js (280 lines) — bodies VERBATIM.
app.js 10,858 → 10,592.
Chosen BY MEASUREMENT, not by eye. Ran the transitive closure over four candidate
clusters and took the one with the smallest interface:
diagnostics 7 fns 235 lines span 4110-4378 imports 1 exports 2
shortcuts-modal 5 fns 235 lines span 104-9897 imports 4 exports 5
settings+updates 47 fns 1012 lines span 1409-7636 imports 19 exports 30
library-render 220 fns 4064 lines span 20-10536 imports 126 exports 117
diagnostics is contiguous and nearly closed; its one inbound symbol
(_DIAG_FILE_LABELS) lives inside the region and is read only by _renderDiagPreview,
so it moves in and the module ends up a LEAF — imports nothing.
app.js -> { plugin-loader, viz, diagnostics-export }
plugin-loader -> viz
viz, diagnostics-export -> (nothing)
Exports exactly 2: previewDiagnostics + exportDiagnostics, both already in app.js's
window contract (they're inline handlers in the Settings screen) — so app.js keeps
re-exposing them, now as imported bindings. The preview renderer, the file-label
table, and the byte/HTML formatters are used NOWHERE else in core and stay private.
VERIFIED BY DRIVING IT, not just booting. Zero harnesses broke — because the
diagnostics export flow had NO source-level test at all, which is exactly why a
green suite proves nothing here. So the flow was exercised for real: A/B against
origin/main in two browsers, window.previewDiagnostics() invoked, the preview
container rendered identical content on both, both entry points resolve on window,
zero console/page errors either side.
pytest 2396, node 1038/1038, ESLint 0, tailwind-fresh clean, Codex 0.
NOTE for the next carve: library-render is NOT a cluster — 220 functions and 126
inbound symbols is most of app.js entangled together. It cannot be carved as a
unit; it needs decomposing from the inside first.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
a222b45c02
|
refactor(app): carve the viz layer out of app.js — and delete the loader seam (R3a) (#880)
static/js/viz.js (770 lines) — the viz picker, renderer selection, Auto-match, the WebGL2 probe, the 3D-promotion nag, the notation hints. Bodies VERBATIM. app.js 11,603 → 10,857. THE SEAM IS GONE. #878's plugin-loader needed configurePluginLoader({ populateVizPicker }) purely because _populateVizPicker lived in app.js and importing app.js would have closed a cycle. viz.js is a LEAF — it imports NOTHING — so plugin-loader now imports _populateVizPicker straight from it. The _host object, the configure function, its loud-default guard, and the wiring line in app.js are all deleted. The second carve simplifies the first. app.js -> { plugin-loader, viz } plugin-loader -> viz viz -> (nothing) NOT A PURE MOVE — one listener block had to be SPLIT. app.js had a single top-level `if (window.feedBack) { … }` registering four handlers, and only two were viz. song:loaded / arrangement:changed / song:ready (the mastery slider) stay in app.js and now call the imported _autoMatchViz / _maybeShowNotationViewHint. The viz:reverted handler MOVES, because it REASSIGNS _cancelPendingAutoLabel and an imported binding is read-only — `_cancelPendingAutoLabel = null` would throw if the listener stayed behind while the state moved. ORDER CHECKED, NOT ASSUMED: viz.js's song:ready listener now registers BEFORE app.js's own (imports evaluate first). Safe — _pendingPromotionNag is only ever set inside _populateVizPicker, which runs at boot/plugin-refresh, never from inside the other song:ready handler, so the two are independent. VERIFIED — the listeners are the risk here, so they were DRIVEN, not just booted. A/B against origin/main in two browsers: * viz picker: 6 options (auto|default|venue|drum_highway_3d|keys_highway_3d| highway_3d), selected highway_3d, Auto label — IDENTICAL. This alone proves plugin-loader's direct import of viz.js works. * emit('viz:reverted') -> picker resets to default, localStorage resets to default, the warning logs — IDENTICAL. The MOVED listener fires. * emit('song:ready') -> mastery slider enables, no throw — IDENTICAL. The SPLIT listener still does both halves. * plugin screens, module injections, 37 capability participants — IDENTICAL. * zero console/page errors on both. pytest 2396, node 1038/1038, ESLint 0, tailwind-fresh clean. no-cycle re-bitten on the 3-module graph (viz -> plugin-loader fails). Codex preflight raised a [P2] claiming viz.js's top-level bus guards would be false because "app.js only creates the event bus later" — FALSE POSITIVE. app.js does not create the bus; capabilities.js does, from its own <script type="module"> at index.html:122, and module scripts execute in document order, so the bus exists long before app.js's import graph evaluates. Instrumented the setter: by viz.js's turn `window.feedBack.on` is already a function, and the viz:reverted listener is provably attached (firing it resets the picker). The ordering is also enforced by test_app_shell_loads_capability_registry_before_app_runtime. Harnesses: 5 tests retargeted to viz.js across legacy_shim_hits, venue_scene_3d, venue_viz (each SPLIT — their non-viz tests still read app.js). Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
OmikronApex
|
5b904706d0
|
feat(audio): loopback feeder mode + static no-cache — all app audio under exclusive/ASIO (#877)
* feat(audio): route feedpak full-mix natively under exclusive output Song playback runs through the renderer, which WASAPI-exclusive (and ASIO) output silences. Route single-mix feedpaks (stem-less original_audio packs AND single-stem packs) onto the engine's backing transport when the output device type is exclusive-style, and migrate back to HTML5 when it isn't. Extends /api/audio-local-path to resolve /api/sloppak/.../file/... URLs via the same containment guards as serve_sloppak_file. Multi-stem packs stay on the WebAudio path (Phase 2). Includes [feedpak-route] transition-gated diagnostics logging. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(audio): renderer-bus feeder — mix renderer song audio into engine output (Phase 2) Under exclusive-style output the native backing transport (Phase 1, #824) carries loose /audio/ songs and feedpak full-mixes, but not the stems plugin's multi-stem WebAudio graph or tracks JUCE rejected. The feeder taps the renderer-side master with an AudioWorklet, re-points the owning AudioContext at a null sink so it keeps rendering without a device, and pushes ~10 ms chunks over IPC into the desktop engine's renderer bus (feedBack-desktop#90 follow-up). Inert in the Docker sphere and in shared mode. Validated by the fix12 tester spike: null-sink rendering works, clocks hold, no overflow. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(diag): --debug ASIO routing diagnostics in static bundle Gated on window.feedBackDesktop.audio.debugEnabled() (desktop --debug); inert in the Docker sphere and normal desktop runs. - [asio-diag] getCurrentDevice= full device object on outputType change (catches ASIO drivers reporting a non-'ASIO' type name) - [asio-diag] renderer-bus: full feeder decision vector, change-gated (running/exclusive/stems/juceMode/elementSong/want/mode) - [asio-diag] setSink: every sink flip with ctx state + rate Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(audio): loopback feeder mode — all app audio under exclusive/ASIO Tester-confirmed (2026-07-11 log): song previews and other plugin-private audio bypass the per-surface feeder taps and leak to the default WASAPI device under ASIO output. Also confirmed: the element capture path poisons itself when highway_3d already owns #audio's one-shot MediaElementSource (InvalidStateError with _elCtx assigned pre-throw → TypeError every later tick). - New preferred mode 'loopback': one getDisplayMedia frame-audio capture (desktop main answers with the app's own frame) covers song, previews, and UI sounds for the whole exclusive session — engages even with no song loaded. Local playback silenced via suppressLocalAudioPlayback, page-mute IPC fallback otherwise. - Sticky fallback to the existing stems/element surface modes when capture is unavailable (old desktop main, denied, Docker sphere). - Element capture: assign module state only after the whole chain succeeds; close the context on failure — collision now retries clean. - Failed engage now disables the bus and tears down loopback (no more bus-enabled-with-no-producer stranding). - Tests: 12 (5 new — loopback engage/preference/mute-fallback/sticky fallback, collision retry). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(audio): loopback feeder mode — all app audio under exclusive/ASIO Tester-confirmed (2026-07-11 log): song previews and other plugin-private audio bypass the per-surface feeder taps and leak to the default WASAPI device under ASIO output. Also confirmed: the element capture path poisons itself when highway_3d already owns #audio's one-shot MediaElementSource (InvalidStateError with _elCtx assigned pre-throw → TypeError every later tick). - New preferred mode 'loopback': one getDisplayMedia frame-audio capture (desktop main answers with the app's own frame) covers song, previews, and UI sounds for the whole exclusive session — engages even with no song loaded. Local playback silenced via suppressLocalAudioPlayback, page-mute IPC fallback otherwise. - Sticky fallback to the existing stems/element surface modes when capture is unavailable (old desktop main, denied, Docker sphere). - Element capture: assign module state only after the whole chain succeeds; close the context on failure — collision now retries clean. - Failed engage now disables the bus and tears down loopback (no more bus-enabled-with-no-producer stranding). - Tests: 12 (5 new — loopback engage/preference/mute-fallback/sticky fallback, collision retry). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(audio): close loopback capture context on teardown (release tap worklet) The loopback context was reused across engages (_lbCtx || new), but teardown only stopped the stream + deactivated the tap — never closing the context or detaching the worklet node. Each exclusive<->shared switch orphaned a live tap worklet on the long-lived context. Use a fresh context per session and close it on disengage. Adds a test asserting the context is closed on teardown. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(diag): install-time + uncaught-error diagnostics for the reroute chain 2026-07-11 tester log showed the routing watcher and renderer-bus feeder never installed (zero [feedpak-route]/[renderer-bus] lines) plus an uncaught SyntaxError with no source location — nothing in the log said why. New: - global error/unhandledrejection tap logging message + filename:line:col (error events carry the location even for parse errors in other scripts) - explicit install / NOT-installed lines for watcher and feeder (incl. loopback capability probe) - DOMException detail (name/message/stack head) in the feeder retry warn — the console-message forward stringified it to [object DOMException] Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(static): force conditional revalidation on /static (Cache-Control: no-cache) Without Cache-Control Chromium's heuristic freshness (10% of file age) serves /static/app.js from disk cache for hours-to-days without revalidating. Desktop consequence: a new build's window ran the previous build's app.js — the 2026-07-11 ASIO investigation traced 'routing watcher never installed' + a stems module-plugin SyntaxError to exactly this (stale loader predating scriptType support). no-cache keeps caching but revalidates via ETag — unchanged files still cost only a 304. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(diag): gate install-time + uncaught-error [asio-diag] lines on --debug The error tap and install lines from the previous diag commit were unconditional. Now: error/rejection taps check _asioDiagEnabled() at event time; install lines log deferred once the async debugEnabled() resolves true. The NOT-installed anomaly lines stay bridge-gated (window.feedBackDesktop present) instead — a broken bridge can't deliver the debug flag, they fire at most once, and only in the broken state they exist to witness. Docker sphere: fully silent. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: Byron Gamatos <xasiklas@gmail.com> |
||
|
Byron Gamatos
|
38772f604a
|
refactor(app): carve the plugin loader out of app.js into static/js/ (R3a) (#878)
The first carve, and deliberately the riskiest: app.js IS the plugin loader (the
R0 host rails), so it goes first while the module graph is still one edge deep.
static/js/plugin-loader.js (829 lines) — bodies VERBATIM. app.js 12,217 → 11,439.
Core's first `static/js/` module, exactly as constitution II anticipates.
CLOSURE (measured with acorn, not regex — brace-matching stripped source drifted):
the block at app.js:11246-12031 is contiguous and self-contained. It needs only
TWO things from the rest of app.js, and exports only TWO:
exports: loadPlugins (the window contract), bootstrapPluginsAndUi (boot)
inbound: window.showScreen — already the public host contract (constitution II),
so it is called through `window`, not re-coupled as an import
_populateVizPicker — injected via configurePluginLoader()
WHY A SEAM, NOT AN IMPORT. plugin-loader must not import app.js: app.js imports
it, so that would close a cycle. I checked whether _populateVizPicker could just
move into the module instead (which would delete the seam entirely) — it drags 9
further symbols (_canRun3D, _autoMatchViz, _showPromotionNag, …), i.e. a whole
viz cluster. That is its own carve, so the seam stays.
THE SEAM'S DEFAULT IS LOUD, ON PURPOSE. A no-op stub is the classic silent
failure for this pattern (see the editor's setHostHooks trap, hit twice): drop the
wiring call and the loader keeps working while the viz picker quietly stops
refreshing — no test, no boot check says a word. The default now console.errors,
so the smoke harness catches it. VERIFIED BY BITE TEST: removing
configurePluginLoader() from app.js surfaces
"[plugin-loader] host seam not configured" at boot. The seam IS exercised on the
plugin-startup path, so an unwired hook cannot pass silently.
no-cycle is now LIVE on core's own graph for the first time. eslint.config.js
gains `static/app.js` + `static/js/**` to the module block — app.js now `import`s,
so parsing it as a script would be a syntax error. VERIFIED BY BITE TEST: making
plugin-loader import app.js back fails with "Dependency cycle detected".
HARNESSES (the R3a note said budget one conversion per carve — it was five):
retargeted capability_inspector_nav, plugin_hydration_wipe,
plugin_loader_script_type, plugin_style_injection, legacy_shim_hits (SPLIT — one
test needs the loader, one still needs app.js) + test_plugin_runtime_idempotence.
legacy_shim_hits was missed by a symbol-name grep because it greps for a code
STRING; only the failing run found it. test_capability_events' NEGATIVE asserts
now span app.js + the loader — carving code out of app.js would otherwise make
them vacuous instead of failing.
VERIFIED: A/B against origin/main in two browsers — mounted plugin screens, 14
loaded plugin scripts, the 3 module plugins injected as <script type="module">,
37 capability participants, 14 shims, window.loadPlugins: IDENTICAL, zero
console/page errors on both. /static/js/plugin-loader.js serves 200; R0 rails
intact (src/main.js 200, conditional GET 304, script_type passthrough).
pytest 2396, node 1032/1032, ESLint 0, Codex 0.
Codex preflight caught a REAL [P1] first pass: static/js/plugin-loader.js was
untracked, so a checkout would have served an app.js importing a nonexistent
module — a failed static import kills the whole module and every window handler
with it. Now tracked.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Byron Gamatos
|
92c86f5393
|
refactor(ui): load app.js as an ES module (R3a) (#876)
One attribute. #871/#872/#874/#875 exist to make this line safe. app.js's 385 top-level `function` declarations stop being implicit `window` properties: 87 stay reachable via the explicit contract (#874's Object.assign block + the 47 pre-existing `window.X = X` assignments), and 298 become module-private. Verified NO unexposed name is read from outside app.js. Strict mode (modules are always strict) checked ahead of the flip: app.js parses clean as `sourceType: module` (no octal, dup params, `with`), and has no implicit globals, no `eval`/`new Function`, no top-level `this`. `registerShortcut` is called bare at 15 top-level sites but is assigned at `window.registerShortcut` (app.js:10387) before its first call (10648), and a bare identifier in a module still resolves through the global object — verified `typeof window.registerShortcut === 'function'` in the browser. HARD GATE — app.js IS the plugin loader: - /api/plugins script_type passthrough: editor/stems/studio = "module" - /api/plugins/stems/src/main.js -> 200; conditional GET -> 304 (live-edit ETag) - deep graph: stems/src/transport.js, editor/src/state.js -> 200 - window.loadPlugins present; 5 plugin screens mount; the 3 migrated plugins injected as <script type="module"> - 37 capability participants, 14 compatibility shims, bus + capabilities v1 Every one of the shell's 336 inline handlers resolves on window under module scope, and the A-Z rail / pagination execute 6/6 with no ReferenceError. A/B against origin/main: the ONLY unresolvable handler is `editorToggleStemMixer`, which is equally broken on main (a dead handler in the editor plugin — not defined anywhere in its source; pre-existing, flagged separately). Codex preflight raised a [P1] claiming restartCurrentSong / requestExitSong / editRegionInEditor / returnToEditorFromHighway would ReferenceError — FALSE POSITIVE. It scanned only #874's new Object.assign block and missed app.js's 47 scattered `window.X = X` assignments; all four are at app.js:7086/7204/8492/8511 and all four resolve as `function` in the browser with app.js loaded as a module. pytest 2396, node 1032/1032, ESLint 0 errors, tailwind-fresh clean. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |