Commit Graph
7 Commits
Author SHA1 Message Date
Chris LeeandClaude Opus 5 8cdee83ff4 Windows: decide a surface's visibility when the call is queued, not before
connect() registers window surfaces in a loop that reads the foreground
window once, before the loop, then awaits each surface's configure call
before queuing its visibility. On a fresh process restoreOpenApps runs at
the same time and focuses windows as it relaunches them. The bridge runs
Java calls in order, so a visibility decided before the await lands after
the focus change's own calls and undoes them: the old foreground can stay
visible and cover the new one, and a later window switch doesn't clear
it, because a switch only hides the window being left.

configureWindowSurface now also accepts visibility as a function, asked
after the configure call, when the visibility call is queued. The display-
attach loops (glasses and preview), the display-mode, per-app layout and
vertical-position changes, and an in-process window's height-mode change
pass one. Launch paths that pass a literal false are unchanged.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-25 22:24:22 -07:00
Chris LeeandClaude Opus 5 54eae0c6b9 Stop leaking every submitted frame's pixel buffer in native memory
Passing a JS ArrayBuffer to a Java ByteBuffer parameter leaks it in
@nativescript/android 9.0.4 through 9.1.1: JsArgConverter wraps the backing store with
NewDirectByteBuffer and takes a JNI global ref that it never deletes. The
ref keeps the ByteBuffer reachable, so makeInstanceWeakAndCheckIfAlive
always answers "alive", the JS finalizer re-arms, and V8 never frees the
ArrayBuffer. submitSurfaceFrame passed a fresh copy of each frame this way,
so every frame leaked a full surface (576x260, 640x452 or 640x480 bytes)
plus one global ref, including frames that Java then discards as unchanged.
The draw-list buffer and buzzer payloads leaked the same way.

Each payload now goes through a reused Java direct ByteBuffer
(app/native/java-direct-buffer.ts): JS writes into it through an
ArrayBuffer.from view, and the Java object itself is passed, which takes
NativeScript's ordinary object path. Every consumer (SurfaceCompositor
applyAndComposite, playBuzzerSequence) copies before returning, so one
buffer per JS context is safe.

Covered: the communicator bridge (frames, draw lists, buzzer, which also
carries EvenHub tones), the preview-only display, and all eight worker
apps. Workers submit through the Android getActiveDisplay() (frames) and
worker-buzzer (tones), so the reused buffers live there rather than in
each worker; the worker sources are shared with iOS, whose active display
takes the ArrayBuffer itself. The
smaller one-off crossings (glyph and image atlas registration, the gray
preview helper, firmware and file writes) are left as they are.

tests/java-direct-buffer.test.cjs covers the helper against a fake
ByteBuffer, and the Android worker-buzzer test stubs it.

(Rebased onto main by James Babcock: worker changes moved into
active-display.ts and worker-buzzer.ts.)

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-25 22:24:22 -07:00
Chris LeeandClaude Opus 5 22c851a6b9 Health: on-device store, derivations and a glasses glance for ring data
The companion half of the ring protocol, offered less as finished UI than as
an executable answer to "what does this ring data actually mean" -- the part
that took two sessions to work out and that a future contributor would
otherwise have to redo.

- app/health/: wire->domain conversion, an append-only on-device store with
  deduping ingest, rollups, a per-day step-bucket ledger, sleep stage mapping
  (0=wake 1=REM 2=light 3=deep, verified against 34 real nights), and chart
  derivations.
- app/apps/health/: a glasses glance page over that store.
- tools/health-preview.cjs: renders the real chart code headlessly, so layout
  can be checked without an emulator.
- 26 new tests, all passing.

Three clock frames meet in this data and mixing them up causes real bugs, so
they are documented at each site. The ring's clock is deliberately set to
now + 4h (matching what a real Even write carries), so every timestamp it
reports is in that shifted frame and must be corrected on read. Sleep is the
exception: it needs -8h, twice the correction, confirmed against a known
night. Why the ring wants these quantities is unexplained -- the code says so
rather than inventing a tidy mechanism, because the tidy story was tried and
measurement falsified it.

Collection runs on a wall-clock-aligned tick at :01 and :31. Aligned because
the ring's step buckets close on 10-minute boundaries, and because an elapsed
timer drifts off the boundary it is meant to track; offset by a minute because
firing exactly on the boundary risks catching a bucket before it is finalised.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-17 20:14:09 -07:00
Chris LeeandClaude Opus 5 0db295d802 R1 ring: health-data BLE protocol (codec, five record decoders, page ACK)
Implements the Even R1 ring's health-sync protocol, reverse-engineered from
Bluetooth HCI snoop captures of Even's own app and confirmed field-by-field
against that app's CSV export. Answers the open question in issue #22, and in
notes/apps.txt, about where interpretation happens: it happens ON THE RING.
Sleep arrives already classified into stages; the phone does no inference.

RingProtocol.java: 15-byte header, 2-byte command id, CRC-32C (non-reflected)
over the frame body, fragment reassembly, and decoders for all five record
types -- heart rate, HRV, SpO2, sleep and steps. Hourly types share the shape
[hour_index][avg][max][min]. Steps are 10-minute buckets counted from the
record's anchor. RingProtocolSelfTest.java covers the codec against captured
bytes.

Three traps are documented in the code because each one costs days and a
reviewer "tidying" any of them would silently break the sync:

1. The 00:08 handshake frame is the ONE request whose payload carries no
   nonce -- it is exactly 3f 01 01. Measured across 46 real writes with zero
   exceptions: the 3-byte form is always answered, the 5-byte form never is.
   Prepending a nonce makes the ring discard the frame and then ignore every
   subsequent request on that connection.

2. The ring holds only one health type's response in flight. Requesting the
   next type before the previous type's DATA arrives silently drops it, and a
   lost race appears to discard that backlog PERMANENTLY rather than delaying
   it. requestRingHealth() is therefore response-paced, not sleep-paced, and
   ACKs each page as it lands.

3. HEALTH_COMMANDS mirrors the order Even's app uses. Whether the ring cares
   is untested, and matching is the cheap way not to find out the hard way.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-17 20:14:09 -07:00
Chris LeeandClaude Sonnet 5 2d717a9cdb Add on-device Whisper as a second offline transcription option
sherpa-onnx's native library already reads an offline Whisper config
(verified against sherpa-onnx/jni/offline-recognizer.cc at v1.13.0);
only the Java wrapper was a package-private stub. This adds a public
OfflineWhisperModelConfig builder and wires base.en (int8) in as a
second on-device model next to Moonshine.

- FaceclawVoiceController: setOnboardModelKind() selects the model
  before start("onboard"); defaults to Moonshine. Whisper skips the
  700ms live-partial redecode (each call re-encodes the whole utterance
  plus ~10s of tail padding) and gates near-silent segments before
  decode.
- asr-model.ts: ASR_MODELS registry with per-model download state.
- Settings > Voice: "On-device (Moonshine)", "On-device (Whisper)",
  and "OpenAI (Whisper)" for the existing cloud provider (stored
  value unchanged); one model row per on-device model.
- Microphones live captions pass "moonshine" explicitly (unchanged).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-14 22:46:22 -04:00
Chris Lee 6f304fc076 Wake up notifyAll() waiters when a BLE ack lands (fixes up-to-100ms dead time on every EvenHub resume)
awaitEvenHubSessionReady() polls fixedLayoutCreated/displayedFingerprint at up to 100ms
granularity because resolveAckLocked() never notified waiters when the relevant ack
actually arrived. Confirmed live on real hardware (2026-09-08): full resume (prelude +
create-layout + content frame) now completes in 449ms with a clean, tight ack sequence
and no regressions. Safe by construction: the one caller already holds the lock, and
every lock.wait() consumer in this file rechecks its own condition in a loop.
2026-09-08 22:54:27 -04:00
Chris Lee ec12a4845b Fix onboarding firmware-check: kind mismatch broke every real result
classifyOnboardingFirmware() (firmware-compat.ts) returns kind values of
"custom" | "flashable-stock" | "newer-stock" | "unknown", but
applyClassification()'s switch (onboarding-firmware-check-view-model.ts)
checked for "flashable" / "newer" -- names borrowed from the unrelated
CheckPhase type used a few lines later via setPhase(). Since kind was
typed as plain string, nothing caught the mismatch at compile time.

Effect: any device that successfully reported a firmware version (stock,
flashable-stock, or newer-stock alike) fell through to the switch's
default case and showed "Connected, but couldn't read a firmware
version" -- the hard-block message meant for when the probe gets no
data at all. Confirmed on real hardware (glasses on 2.2.9.22): the BLE
probe read the version correctly every time, logged it correctly, and
the UI still reported it as unreadable.

Fix: match the real OnboardingFirmwareKind strings, and tighten the
parameter's type from `string` to `OnboardingFirmwareKind` so this
category of typo fails to compile instead of failing silently.
2026-08-30 10:19:35 -04:00