Playing a move used to show two progress indicators at once: the thin
bar on the card and a sticky bottom toast (Loading / Uploading NN% /
Playing NN Hz) that also collided with the playback dock on phones.
Move every piece of live feedback into the card via a shared
createPlaybackReadout helper:
- spinner on the play button + "Uploading NN%" in the meta line while
the move streams to the daemon (the bare spinner never told you how
far along a big move was)
- top bar + "M:SS / M:SS" ticking in the meta line once the daemon
reports playback started
- scrollIntoView so a card far down a long folder is still visible
- progress bar can resume from a fraction so the folder's async
render() swapping the card mid-playback no longer restarts the fill
The wizard's review preview uses the same helper (it previously had
the bar but no upload feedback). The Hz readout is gone: Playback.liveHz
was a hardcoded 100, and the debug HUD already has a live rate.
Share: drop the sticky "Uploadi…
fix: delay motion for robot-audio playback so it no longer leads the sound
The daemon's play_move kicks off play_sound then enters the motion loop
immediately, but the robot's GStreamer playbin has a warmup before sound
reaches the speaker — so with a zero audio lead the head starts moving
before the music ("animation starts too early"). Add a positive baseline
audio_lead_ms (ROBOT_AUDIO_PLAYBIN_LEAD_MS, in lib/motor.js) for daemon-side
audio, layered on top of the user's Settings/folder offset, applied in both
the folder playback and the wizard review-preview paths. Device-output audio
(browser playbin-free) keeps the raw offset. Bump version to r2.
Co-authored-by: Cursor <cursoragent@cursor.com>
fix: repair topbar title, move-card, and toast layout on narrow phones
- topbar: compact "EXP" chip on phones (full "EXPERIMENTAL" from 480px up)
and let the title claim leftover width, so short titles no longer
truncate to a couple of characters
- move-card: single-line ellipsised meta and wrap the action row (share /
rename / delete) onto a second line when it would starve the title,
fixing the overlapping/multi-line breakage on owned moves
- toast: anchor above the bottom dock (via --toast-dock-offset set in
main.js) and clear the safe-area inset so it can't be clipped or land
on top of the playback sliders
- cta-dock: a bit more bottom padding so the Record button isn't flush
to the screen edge
- bump version label to exp 2026-07-30 r1 ux-layout-fixes
Co-authored-by: Cursor <cursoragent@cursor.com>
docs: record viewer-fix implementation status in the spec copy
RemiFabre
7faac64
sweep: cover datasets seeded from other devices via ?author= enumeration
RemiFabre
56b26fe
repairOwnedViewerConfigs now unions the caller's local-moves repo map
with HF's ?author=<namespace>&filter=<community tag> listing, so 'all of
your datasets' really means all: repos shared from this browser (even
vintage untagged ones) plus tagged repos created via the Python app or
another device. A failed enumeration counts as an error so the kv
done-flag isn't persisted and the sweep retries next session.
Bump to exp 2026-07-04 r1 hf-viewer-fix
RemiFabre
d0fa207
HF dataset-viewer fix: configs + metadata.jsonl on share/compress, auto-repair sweep at startup
RemiFabre
2a4933f
Implements docs/hf-viewer-fix-spec.md (root cause: extension-inferred
JSON loader chokes on motion files and disables the whole viewer).
Share & compress paths (deliverable A):
- viewerConfigPlan/renderViewerConfigBlock/ensureViewerConfig/
buildViewerMetadataRows/buildMetadataJsonl/parseMetadataJsonl helpers
(exported for tests); ensureViewerConfig only touches the frontmatter's
configs: key — byte-preserving, idempotent
- uploadMove: README is no longer regenerated on every share (the old
clobber-customized-READMEs failure mode). Fresh repos get the community
README (+configs when audio); existing repos get only the configs key
ensured; metadata.jsonl rebuilt to cover ALL audio in the repo
(titles = stems, descriptions from local moves via opts.localMoves,
existing rows kept). Rename-cleanup-doomed files excluded (rule 4).
fetchRepoText distinguishes 404 from network failure: on failure the
share proceeds WITHOUT touching README/metadata (never clobber).
- compres…
Bump to exp 2026-07-03 r2 tfrere-ux-port; doc: record port status
New tests/stubs/hf-hub.js aliases the @huggingface/hub CDN import so
lib/hf.js loads under vitest; 7 new tests pin the F1-F3 behavior:
delete ordering, .ogg-aware stale paths, protectedStems skip, failed
upload leaves sharedStem untouched, failed delete degrades to a warn,
container-correct sidecar naming. 68/68 green.
components: leak/robustness sweep
RemiFabre
da98a65
- modal: dismissTransientOverlays closes menu + select through their own
close handlers (detaching document keydown listeners) and sweeps
.select-popover too — it removed the backdrop but orphaned the popover
- menu/select: module-level active-close registry so external dismissal
never strands a document-level Escape listener
- trim-field: tear down the previous instance's ResizeObserver on
re-render instead of piling registry entries holding detached DOM
- mic-meter: start() restarts cleanly instead of orphaning the previous
rAF loop + AudioContext on a double start
- ui: drop unused sectionHeader export
folder: don't block playback start on the folder re-render
RemiFabre
19b7fc0
playMove awaited render() so the spinner/progress bound to the fresh
card — but on big folders that rebuild decodes every audio blob and
delayed the robot's first motion by a user-visible amount. The progress
binder now re-resolves its card lazily on every event (update/start/
stop), so render() can be fire-and-forget and playback starts
immediately.
toast: keep the bar node attached across refreshes (F8)
RemiFabre
be5dfc7
remove()+prepend() cancels running CSS animations even when done
synchronously, which defeated exactly what the comments promised: the
indeterminate loop restarted (looked frozen) on every 150 ms playback
repaint and the determinate fill stepped instead of gliding. The bar is
position:absolute, so child order is irrelevant — rebuild only the
non-bar content nodes and leave the bar in place; replace it solely on
a real mode change.
topbar: restore the EXPERIMENTAL chip (F6)
RemiFabre
5b036e9
The redesign dropped the APP_VERSION-gated chip; on this Space users
had no in-app signal they're not on stable. Re-added in topbar-left,
after the contextual title — same 'exp ' prefix gate, so it still
auto-disables on the production naming patch.
wizard: block Back + abort continuation during audio prep (F4 ghost-take fix)
RemiFabre
985fbb8
During the 'Preparing/Uploading audio…' awaits none of the in-progress
flags (Capture.active/rafId/countdownIv) are set, so topbar Back could
tear the view down and the pending goBtn continuation then started a
ghost countdown on the detached view: beeps, motors limp, invisible
take. Two guards: onBack no-ops while preparing (baseline behavior),
and the continuation re-checks `cancelled` after the awaits so any
other teardown path (stopAllActivity -> runCleanups) also aborts it.
hf: harden rename cleanup (F1-F3) — delete after upload, only existing paths, skip sibling-slug collisions
RemiFabre
80e8277
- Cleanup now runs AFTER hfUploadFiles: a failed upload can no longer
leave the dataset with neither old nor new files
- Stale paths come from listDatasetFiles + a SOUND_EXTENSIONS match, so
Opus-era .ogg sidecars are cleaned too (was hardcoded .json/.wav) and
the atomic delete commit never trips on a missing path
- opts.protectedStems (wired from the folder share handler): when the
old slug now belongs to a sibling move, skip the delete instead of
destroying that move's live remote copy
Port tfrere UX redesign onto compression base (mechanical replay)
RemiFabre
477cd2b
- Thibaut's one-sided files wholesale: views/components/* (13), welcome,
topbar, settings, main.js, index.html, style.css, db.js, mic-meter,
waveform, hands illustration, dev tooling (live-server, dev-local.js)
- 3-way merged (base 9e420d7): hf.js, state.js, util.js, folder.js,
folders.js, wizard.js — compression era (canonicalizeMotion, Opus
record/save, compress-&-republish banner, Ogg playback, direct share)
grafted into his redesigned structure
- Dropped wizardFromFolder in favor of his wizardOrigin (subsumes the
back-from-folder fix)
- Ported the Opus upload diff into views/components/upload-field.js
(transcodeToReachyAudio, real ext/mime instead of hardcoded .wav)
- SDK pin 1.8.0-rc1-main.dbe26af -> 1.8.4 everywhere (daemon runs 1.8.4;
his 1.8.1 pin skipped)
61/61 unit tests pass. Known follow-ups land next: rename-cleanup
hardening, wizard Back guard, EXPERIMENTAL chip, toast bar fix.
docs: Rémi's inline decisions on the port review (SDK 1.8.4 direct, drop sign-out, fix the rest)
RemiFabre
e8e14f4
docs: review of tfrere UX redesign + staged import plan
RemiFabre
87552c3
Sync from marionette-js @53b07de: compression era (Opus audio, canonical 50Hz/6dp motion, compress & re-publish, folder-share fixes)
RemiFabre
ff2b76b
r3 (Jun): "Share this tuning with viewers" button
RemiFabre
9e420d7
Author can tune the per-folder Audio sync offset slider and propagate
the value to viewers without re-sharing a move. Until now the marionette.json
sidecar was only committed as a side effect of uploadMove, so a slider
edit after the last share never reached HF; viewers were stuck on the
offset baked into the most recent move commit.
New piece: a "📤 Share this tuning with viewers" button rendered below
the slider, visible only when the user is the author of the folder's
HF repo AND the local audioOffsetMs differs from what was last
published. Click → HF.publishFolderTuning commits a marionette.json-only
change → folder.publishedAudioOffsetMs updated to match → button
hides.
Plumbing:
- lib/hf.js: publishFolderTuning(repoId, audioOffsetMs, token).
Always writes the file (even at 0) so an author reverting a
non-zero published value back to 0 propagates correctly.
- views/folder.js share path: persists folder.repoId (parsed from
the uploadMove URL) and folder.publishedA…
r2 (Jun): trim sync banner copy + move it from home view to folder view
RemiFabre
1d7ed2e
r1 (Jun): surgical community-folder update — keep local recordings
RemiFabre
d28dca6
The Update banner inside a community-imported folder used to call
deleteDatasetAndMoves(folder.id) before re-importing, which nuked
every move in the folder regardless of provenance. A colleague who
filed local recordings into a community folder lost all of them
after clicking Update.
The schema already knows what's local vs imported: every move carries
`fromCommunity: true|false` (set by lib/hf.js at import time, defaulted
to false by the wizard). The fix is to partition before importing:
1. List local moves; collect ids where fromCommunity === true
(prevCommunityIds).
2. importDataset returns the set of ids it imported (one tweak to
lib/hf.js: track + return importedIds alongside the existing
{ datasetId, imported }).
3. Orphans = prevCommunityIds minus importedIds. These are community
moves the upstream owner deleted; we drop them via deleteMove.
4. Local recordings (fromCommunity:false) are never touched.
UX: the confirm-dialog now says "Your N local recordings in th…
r8: replace softReturnToBase / prepForPlayback with safelyReturnToPose
RemiFabre
99b04e6
Delete marionette's local wrappers (softReturnToBase, prepForPlayback,
_waitFreshState, sampleAtTime, planDurationToPose, poseSnapshot) plus
the now-unused HEAD_LEAD_S / ANTENNA_LEAD_S / SLEEP_* tunables. All
call sites switch to safelyReturnToPose from
@pollen-robotics/reachy-mini-sdk/animation, passing MARIONETTE_DURATION_PRESET
(min 0.2 s, max 1.5 s; SDK default minDurationSec of 0.01 reads as a
snap for auto-returns).
The pin-then-enable scaffolding existed because setMotorMode("enabled")
used to be unsafe (daemon kept a stale commanded target and would snap
the motors toward it on torque-up). Daemon PR #1138 fixed that by
pinning to the present pose internally on enable, which lets
safelyReturnToPose collapse the whole sequence to enable -> read state
-> scaledDuration -> gotoTarget. Six lines of marionette glue replaced
with three SDK-helper calls.
Motor.wakeUpRobot (failsafe recovery) splits into Motor.stepThroughDisabled
(gotoSleep + sleep + setMotorMode("disabled") + sleep) …
r7: freshen state before post-playback softReturnToBase
RemiFabre
658a3eb
User's log captured the smoking gun: at "playback ended, entering
return-to-base" the cached robot.robotState showed is_move_running:true
and was up to 500ms stale (state events arrive at ~2Hz). Stale state
meant setTarget(stalePose) snapped the motors to where they used to
be, then gotoTarget(INIT, 0.22s) smoothly moved them to INIT. The
combined effect is the antenna oscillation the user observed at the
end of every recorded move.
motor.js's softReturnToBase already exposes {freshen: true} which
requests a state and waits one round-trip before reading; the
post-recording path uses it. Plumb the same flag through the
post-playback path.
r5: diagnostic logs around end-of-playback return-to-base
RemiFabre
d11a35d
User reports a brutal short motion at the end of a recorded move even
after r3's three-channel scaledDuration. Adds two log lines:
- "playback ended, entering return-to-base" in views/folder.js,
captures the wire-shape pose snapshot at the boundary between the
daemon's last play frame and softReturnToBase.
- "softReturnToBase plan" in lib/motor.js, captures the current pose
snapshot, the SDK's distanceBetweenPoses result, the full
scaledDuration plan (duration, limiter, perChannel), and the goto
target with its computed duration.
User can copy/paste these from Settings → Diagnostics → View log when
the brutal motion reproduces, and we can see whether the duration was
computed from a stale pose, missing a channel, or correctly computed
but still snapping for a different reason.
r4: trim sync banner copy + reorder Settings rows
RemiFabre
03bfcaa
Per user direction: drop the "drift was invisible to whoever recorded
them" sentence from the HomeView sync banner (banner stays concise but
keeps the actionable "tune Audio sync offset" guidance). Reorder the
Settings sheet to: Play audio through, Robot motors, Audio sync offset,
Robot microphone, Debug HUD, Streaming benchmark, Diagnostics.
r3: three-channel scaledDuration + defensive mic gain
RemiFabre
a63b402
motor.js was computing the auto-return duration from head distance
only. That produced a snap on the antennas at end of playback when the
head was already near base but the antennas were at full extension.
Swap the local lib/pose.js (head-only, retired) for the SDK's
@pollen-robotics/reachy-mini-sdk/animation helpers: distanceBetweenPoses
+ scaledDuration consider head (magic-mm), antennas (|Delta deg| each)
and body_yaw (|Delta deg|) and take the slowest channel.
lib/animation-helpers.js is a thin re-export so vitest can stub the
jsDelivr URL (Node ESM can't resolve https: imports). Pose unit tests
go away; the SDK's own tests cover the math.
We also push state.settings.robotMicVolume (defaults to 100) to the
daemon on every connect because robot mics have been observed booting
at 0% (OS-level PulseAudio/ALSA setting that persists). Settings sheet
gets a "Robot microphone" slider for manual override.
Bump SDK pin to current @main head (fd4354c → dbe26af)
RemiFabre
1c3dd04
`fd4354c` (what the reference apps ship) is six commits behind upstream
main. Marionette experimental's reason to exist is validating bleeding-
edge changes before they reach the stable Marionette, so the pin now
points at the current `@main` head — `1.8.0-rc1-main.dbe26af`.
Bump is manual rather than via the floating `@main` dist-tag so the
version pin in the repo stays the source of truth (jsDelivr caches
dist-tag URLs for up to ~12 h).
Migrate to host-shell pattern (r1 host-shell-migration)
RemiFabre
f10576b
mountHost() in standalone mode + connectToHost() in iframe replaces the
custom gate (deleted) and the legacy CDN `@main/+esm` import. SDK pinned
to the exact tarball the Pollen reference apps ship against
(1.8.0-rc1-main.fd4354c). Patterned after webrtc_example/index.html.
Icon moved to public/icon.svg so the HF mobile catalog API can probe it
via the Space's siblings listing.
Pin SDK to @main dist-tag (bleeding-edge tracking)
RemiFabre
f5f4031
Pin SDK to npm @rc dist-tag; drop importmap; refresh HANDOFF
RemiFabre
f15a015
main.js now imports from cdn.jsdelivr.net/npm/@pollen-robotics/reachy-mini-sdk@rc/+esm
(currently resolves to 1.8.0-rc1). The /npm/+esm endpoint rewrites bare
specifiers so the @huggingface/hub importmap in index.html is no longer
needed. HANDOFF.md fully refreshed: SDK pin policy section explains
when to flip to @latest; logo placement flagged as provisional; ancient
rollback checkpoints trimmed.
Per-folder offset travels through HF; banner phrasing; no em-dashes
RemiFabre
eb05859
Banner text updated per user request, "visible if played on a different
setup". uploadMove now writes a marionette.json sidecar at the repo
root carrying audio_offset_ms when it is non-zero; importDataset
reads it on import so downloaders get the corrected offset by
default. Opportunistic em-dash cleanup in the files touched this turn.
Banner on HomeView (dismissible) explaining the 2026-05-25 architecture
change. Per-folder audioOffsetMs slider in FolderView header — adds to
the global offset at play time so authors can salvage legacy datasets
recorded before the upload-then-play rollout.
HANDOFF: record r8 state + r7 rollback checkpoint
RemiFabre
1c5642e
Bump SDK pin to 3e9f6a60 + new path js/reachy-mini-sdk.js (post-#1141)
RemiFabre
125125e
Restore importmap (load-bearing, confirmed by r6 test)
RemiFabre
278f25a
TEST: revert importmap to verify it was load-bearing (will revert)
RemiFabre
70dd199
Add importmap for @huggingface/hub bare specifier
RemiFabre
9f86b05
The SDK at 628870a9 imports @huggingface/hub as a bare specifier
(post-PR #1135 npm packaging). Browsers can't resolve that without an
importmap — without this, r4 failed to load on a fresh page. r5
importmap-for-hf-hub.
Bump SDK pin to 628870a9 + new path js/sdk/reachy-mini-sdk.js
RemiFabre
312d93d
Tracks the daemon PR branch after merging upstream main (PR #1135
renamed the SDK file). r4 sdk-path-post-merge.
PR notes: document the JS SDK consumer API + 1c0f0cc0 commit
RemiFabre
1e33815
HANDOFF: record r3 SDK typed helpers + r2 rollback checkpoint
RemiFabre
f1dd6a5
docs: SDK + docs gap analysis report (rationale for the refactor)
RemiFabre
d0281d3
Use the new SDK typed helpers; drop lib/move-upload.js
RemiFabre
793741a
Marionette now calls robot.playMove / robot.cancelMove /
robot.uploadAudio / robot.playUploadedAudio / robot.cancelAudio
directly. SDK pin bumped to commit 1c0f0cc0 on
feature/daemon-side-move-upload. Net delete ~725 lines; new contract
test guards the wrapper layer.
Docs cleanup + new HANDOFF.md entry point
RemiFabre
bda2007
Prune stale pre-2026-05-20 docs that no longer reflect current
architecture; their content was either obsolete or rolled into the
daemon-side rewrite:
deleted: HANDOFF.md (pointed at obsolete DESIGN/SDK/etc.)
deleted: DESIGN.md (April-23 v2 design draft, pre-daemon-side)
deleted: FEEDBACK.md (pre-our-work feedback log)
deleted: REFACTOR_PLAN.md (pre-our-work refactor notes)
deleted: SDK_NOTES.md (SDK gap closed in v1.7.2)
deleted: daemon-goto-target-fix.patch (landed upstream)
deleted: docs/upstream-sdk-prs.md (situation closed)
deleted: docs/upstream-sdk-status-2026-05-19.md (confirmed closed)
New HANDOFF.md (~120 lines) is the single entry point a fresh agent
reads first. Covers repo map, deploy loop, push-permission rules,
architecture in one paragraph, audio invariants, slider polarity,
removed live-recording mode, daemon CI requirements, current state,
and pointers to the four still-relevant docs + the memory files.
RE…
Review screen: mount VolumeRow at the top (only if move has audio)
RemiFabre
134ba4e
The Review step plays the move on the robot via Playback.play, so
the user should be able to adjust the robot speaker volume from
here without bouncing back to Settings or the folder view.
Reuses the same VolumeRow component the folder view embeds; passed
through createWizard so the wizard doesn't have to import settings
directly. Skipped for silent moves -- no audio means the slider
would just be noise.
Version exp 2026-05-21 r2 review-volume.
Wizard review-step fixes: pass Playback dep + defensive enterMarionette
RemiFabre
c32a39c
Two regressions in r13 review-step:
1. Preview crashed with 'Playback is not defined'. createWizard
never listed Playback as a dependency; the new Review UI's
Preview button reaches for it from closure. Add Playback to
the createWizard destructuring + the main.js call site.
2. Torque reportedly stayed ON during recording. enterMarionette
IS still called at the GO! beep; the audio-upload-before-
countdown path shouldn't have affected it. Two defensive
changes to narrow down the cause if it persists:
- Log line at the GO!-beep enterMarionette call (with the data
channel readyState) so we can confirm dispatch.
- A second enterMarionette in startRecording right before
Capture.start. set_motor_mode is idempotent so a duplicate
is harmless when the first one worked; it rescues the case
where the first one was silently lost (data channel hiccup,
etc).
Version exp 2026-05-21 r1 review-fixes.
Wizard: post-record Review step (Preview / Record again / Cancel / Save)
RemiFabre
387daf6
Capture no longer auto-saves on natural finish or Stop. Instead it
enters a Review state where the user can preview the just-recorded
move on the robot and decide what to do with it.
UI flow:
Recording -> [Stop early] -> Review
-> [natural end] -> Review
Review:
[▶ Preview on robot] plays motion + audio via the daemon
(same Playback.play folder.js uses, so
preview == final playback)
[✕ Cancel] discard and go back to the folder list
[🎙️ Record again] discard and return to IDLE in the same
step with state.audio still loaded
[✓ Save] putMove() + return to the folder list
Preview is a play/pause toggle: a second click while playing
cancels the daemon-side playback. Save / Cancel / Record again
are disabled while preview is running so the user can't
accidentally Save mid-preview and trip the busy lock.
The pendingMove dict lives in the Ste…