Anything landing inside a running value curve is refused unless the parameter is
cancelled first, and two paths were not cancelling: the chain observer wrote each
knob straight onto its AudioParam before rescheduling, and a bent segment read as
straight because only the curve exponent was checked, never the via point the
timeline actually writes. The first threw NotSupportedError into the console and
abandoned the rest of the envelope; the second played a dragged bend as a ramp.
Measured against Chrome, in a live context and in an offline one suspended
mid-curve: any cancel frees the span, and only a missing cancel is refused.
clearParamLane takes the strongest form on purpose, because curve-over-curve
refusals were reported with a cancel at the new schedule time already in place
and have never reproduced; emit keeps a ramp fallback as the backstop for
whatever that mechanism turns out to be.
Dynamic carve is what exercises all of it, so it lands here too:
- a `gain` primitive, so a carve can match levels as well as carve bands
- carve settings collapse to one `strength`, with carveProfile deriving the six
numbers that always moved together anyway
- analyseCarveDynamics / analyseCarveDuck turn the analysis into envelopes, with
a slow release so the bed does not snap back the instant a word ends
- worklets are awaited inside attach, so adding a compressor to a carved bed no
longer kills its envelopes and freezes every later edit
- per-track failure detail in the render's audio stage, which was being discarded
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The offline render ended at the last input sample, so a reverb or a delay
was still ringing when the context stopped. Measured on a 1.5 s tone
through a default reverb, the render cut at 1.524 s while the tail was
still at -29.7 dB — an audible chop, and the one place the render did not
match preview.
The length does not have to be guessed. Every tail here follows from its
own settings: a convolution is exactly as long as its impulse, and
`synthesizeReverbImpulse` derives that from room size; a delay's repeats
fall by `feedback` every `time`, so the count down to -60 dB is a log.
Everything else settles with its input — an all-pass chain has group
delay, not a tail, and a 9-second compressor release has no signal to
release once the clip stops.
`chainTailSeconds` sums them (the chain is serial, so a delay in front of
a reverb hands each repeat to the room), reads a lane's maximum rather
than the static knob where one is automated, and caps at 5 s — 5 s
between repeats at 0.95 feedback is eleven minutes of decay, and the
panel can dial exactly that.
The mixer's per-track atrim now allows the clip plus its tail; the atrim
after apad still holds every track to the composition's length, so a tail
can run over what follows but never extends the video.
Same fixture after: a smooth decay to -72 dB, last non-zero sample at
3.306 s against the 3.4 s the settings predict.
A structural edit — an effect added, removed, bypassed, or a filter's pole
count switched — was dropped. `buildFxChain`'s update reports false when the
change is not merely new values, and the attribute observer ignored that, so
the edit only took hold when the persisting write reloaded the composition.
That reload restarted every playing track, which is what was heard as the
audio chopping.
The graph is now swapped in place: the old effects are detached, the new ones
built and connected between the same source and gain, and any lanes
re-scheduled onto the new nodes. The source node is never touched, so playback
does not restart.
A track with no chain is watched too, rather than wired through and forgotten,
so adding its first effect is heard the same way. That means the function
always returns a disposer instead of null for the empty case.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The envelope was scheduled onto the transport's gain AudioParam, but the
runtime rewrites that gain every tick from `data-volume` and the GSAP-seeked
value — so it was erased within a frame. Volume automation was correct in the
render and inaudible while previewing.
The lane now feeds the per-tick path where the probed volume keyframes already
sit, checked ahead of them so the two cannot fight, and the transport no
longer schedules volume at all: one mechanism instead of two racing.
The cost is honest — in preview the level steps per tick rather than per
sample, exactly as the existing keyframe path does. The render still bakes it
into the PCM sample-accurately, and FX parameters are still scheduled on their
own AudioParams, since nothing rewrites those.
Parsed lanes are cached by attribute text: the runtime asks once per tick per
track, and parsing there would run the JSON parser 60 times a second for a
value that only changes on an edit.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The offline render schedules FX lanes with the same scheduler preview uses,
inside the OfflineAudioContext that already runs the same graph builders. The
input WAV is the clip's own audio from its first sample, so clip-local time
is offline time and the envelope needs no offset.
Volume lanes take the existing PCM bake rather than a second mechanism: the
lane is converted to keyframes, so a straight fade stays two of them and only
a bent segment is sampled — the baker interpolates linearly and would
otherwise quietly straighten the curve. A volume lane supersedes keyframes
probed from the timeline, which `lint` already warns about.
A browser test sweeps a lowpass from below a 2 kHz tone to well above it and
measures both ends. Parsing the envelope is not the same as scheduling it,
and only running the real thing tells the two apart.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Schedules each lane onto the AudioParams behind its knob using native ramps
and value curves. Nothing evaluates the envelope per frame: it is handed to
the audio thread once, so it stays sample-accurate however busy the main
thread is, and the offline render will schedule it the same way.
Timing comes from the transport, so an envelope survives seeking into the
middle of a clip, a clip that has not started yet, and a playback rate that
compresses clip seconds into context seconds.
A straight line is only scheduled as a ramp when nothing bends it — no
curvature, a linear parameter scale, and no unit mapping. Log-scaled
parameters and mapped ones are sampled instead, since a delay knob in
milliseconds and a wet/dry pair moving in opposition are not linear in the
parameter they drive.
Lanes with nowhere to write are skipped rather than reported: a one-pole
filter exposes no frequency param, and the worklet effects expose none at
all. Editing an envelope mid-playback re-aims it at the live playhead rather
than restarting the track.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Marks the knobs an automation lane can drive and has each graph builder hand
back the AudioParam behind them, so a scheduler can write to a running effect
without knowing what the effect is.
A knob is not always one AudioParam. A wet/dry mix is two gains moving in
opposition, and a knob in milliseconds drives a delay time in seconds, so
each target carries the mapping out of the knob's own declared unit.
What stays unautomatable is stated where it is decided: a WaveShaper curve, a
convolution impulse and a one-pole filter's coefficients are all rebuilt
wholesale rather than scheduled, and the four worklet effects take values by
postMessage rather than through AudioParams.
The registry flag is written by hand, so a test builds every effect and
checks the exposure both ways — nothing flagged is missing, nothing exposed
is unflagged. A flag that lied would offer a lane that silently did nothing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds the data model behind Ableton-style automation lanes: breakpoint
envelopes over track volume or one knob of one effect in the track's FX
chain, stored on the element as `data-automation`.
Times are clip-local, so an envelope travels with the clip when it moves —
the clip-envelope model rather than arrangement automation.
`sampleAutomationLane` is the single interpolator. The lane drawing, the
preview scheduler and the render bake all call it, so the picture and the
sound cannot disagree about the curve. Log-scaled parameters interpolate in
log space, matching what their own knob already promises.
FX nodes gain a stable `id`, minted by count rather than randomly so the
document is the same on every machine. Lanes address nodes by id, so
reordering a chain never re-points a lane at a different effect, and a lane
whose effect was deleted is dropped rather than left to reattach.
Also warns when a track carries both a volume lane and a GSAP volume tween,
since only the lane is heard and the tween silently does nothing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Clears the health findings the FX stack left behind: the chain-node render
callback was a 70-line closure over half of FxSection's state, and the two
reorder arrows were the same button written twice.
Also drops two exports with no consumers, and registers the audio FX runtime
stub as an entry point — it is bundled by file path, so nothing imports it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
An AudioWorkletNode cannot be constructed before its processor is registered —
it throws, and the surrounding chain is lost with it. `attachElementFxChain`
built the chain first and only then called `ensureAudioFxWorklets`, so every
worklet-backed effect (compressor, limiter, gate, bitcrush) threw on
construction and the track fell back to dry. Instrumenting the preview showed
`hf-compressor: InvalidStateError` with addModule never called at all.
When the module has not landed yet the track now plays dry and the graph is
swapped in once registration resolves, so the effect arrives a moment late
instead of never.
Registration is also tracked per context rather than in one module-level
promise. A processor registered on one AudioContext does not exist on another,
so the shared promise made every context after the first believe it was ready
when it was not — the studio's transport owns its own context, which is exactly
that case.
With the worklets actually running, the compressor's per-sample log10 and pow
became real audio-thread work. Samples below the knee have a gain of exactly
unity and need neither, so the envelope is now compared in the linear domain
and the transcendentals only run for samples that are actually being
compressed.
Dragging a knob wrote the chain through the persisting attribute path on every
input event. That path refreshes the preview, which reloads the composition and
reschedules audio — so a single drag reloaded dozens of times and playback
stuttered the whole way.
Drags now go through `onSetAttributeLive`, the same path colour grading uses for
scrubs: it coalesces undo entries and sets `skipRefresh`, so no reload happens.
The persisting write fires once, when the gesture ends — pointer-up or blur for
a slider, Enter or blur for a typed value. A select commits immediately since
there is no drag to wait for.
While dragging, the control is driven from local state. Waiting for the value to
round-trip through the element attribute made the knob lag behind the pointer.
For the change to be audible without a reload, the graph now follows the
attribute: the chain installed by the transport observes the element and
re-parameterises itself in place, so a value change lands on the next
128-sample quantum. A shape change (effect added, bypassed, pole count) cannot
be patched into a running graph, so it still waits for the next schedule rather
than cutting the audio mid-play.
The regression test drags a slider through several values and asserts the
persisting handler is untouched until release.
Splices an element's FX chain into the playback graph so preview stops being
silent about effects, and wires the carve button that was previously inert.
The chain goes between the decoded source and its gain stage: effects see the
raw signal and volume automation rides on their output, matching the order the
offline render uses. Since preview and render call the same graph builders,
what is heard while scrubbing is what gets written.
The splice lives in the transport rather than on the `<audio>` element. The
transport plays each track from a decoded AudioBuffer and mutes the element to
avoid doubling, so capturing the element with createMediaElementSource would
have processed a stream nothing is listening to — it looked like it worked
because the call succeeded, and the audio was unchanged.
A chain that cannot be built plays dry rather than silencing the track, which
is the right failure in preview: the author keeps working and hears the source.
The render still refuses, because shipping the dry signal there would be wrong.
Carve now analyses for real: it decodes the chosen voice track, ranks its bands
and writes the resulting peaking filters onto this track. Generated nodes are
tagged `fromCarve`, so re-running replaces the previous carve instead of
stacking another set on top of hand-added effects.
Known limitation: the graph is built when a source is scheduled, so a knob
turned mid-playback takes effect on the next play or seek rather than
immediately. Live re-parameterisation needs the transport to hold the handle
and forward updates.
Adds `audioFx` to the editing-affordances contract and renders the FX panel in
the inspector when an `<audio>` element is selected.
The section is audio-only. A `<video>` carries its sound on a separate
`<audio>` element, so an FX chain on the video would have nothing to process.
Chain and carve settings are written straight back onto the element as
serialised attributes, the way colour grading carries its config, so
persistence is an ordinary attribute write and needs no new server route. A
chain that cannot be parsed renders as empty rather than breaking the panel,
and the attribute is left untouched until the user changes something.
The collapsed group summarises what is on the track ("2 effects + carve") so
the state is visible without expanding it.
Wired into PropertyPanelFlat rather than PropertyPanel: STUDIO_FLAT_INSPECTOR_ENABLED
defaults to true, so the flat inspector is what actually renders.
Three defects in the offline FX path, none of which any test could see.
**Float output silently disabled sample-accurate volume automation.** The writer
emitted 32-bit IEEE float; the very next mixer step bakes the volume envelope
into the samples and accepts only 16-bit PCM, returning null otherwise. So
enabling any effect downgraded that track to the ffmpeg expression path — capped
at 32 straight segments, quantising a curved envelope, and on a dense one falling
back to base volume. It now writes 16-bit PCM, clamped rather than wrapped so a
limiter at 0 dB or a resonant filter cannot turn overshoot into a click. A test
asserts the baker accepts the writer's own output and actually fades it.
**Everything was folded to mono.** `prepareAudioTrack` goes out of its way to
emit stereo — its pan filter exists to dodge ffmpeg's 3 dB mono-to-stereo
rematrix — and this folded it, then wrote one channel. So adding a single peaking
EQ collapsed a bed's width and cost ~3 dB in the render, while preview stayed
stereo. Channels now travel as one plane each, through an OfflineAudioContext of
the same width, and come back interleaved.
**Small results decoded the wrong length.** `new Float32Array(buf.buffer)`
discards byteOffset and byteLength, and Node pools small allocations: a 400-byte
payload sits at offset 8 inside an 8 KiB pool, so a clip under ~1024 samples
decoded as 2048 samples of unrelated memory — and the empty-result guard could
not see it. The reader has the mirror-image fix: a float data chunk on an odd
boundary (ffmpeg's pcm_f32le writes fmt(18) + fact, landing `data` at 58) now
copies instead of throwing RangeError on an unaligned view.
The tail limitation is now stated rather than mis-stated: the context is exactly
as long as the input, so a reverb or delay still ringing is cut there. The old
comment claimed the opposite. How far a tail may run past a clip's end changes
the clip's length in the mix, so it is a product decision, not one to make here.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Finds the bands a voice occupies so a music bed can be dipped there, letting
the voice sit in front without ducking the whole track.
Carve is a relationship between two tracks rather than an effect on one, so it
stays out of the FX chain. What it emits is an ordinary chain of peaking
filters, so a carve composes with whatever else is on the track and needs no
separate rendering path.
Selection is weighted toward intelligibility rather than raw voice energy.
Ranking purely by power lands on the fundamental almost every time, because
that is where a voice is loudest — but the masking that actually hurts a
voiceover happens higher up, and dipping 160 Hz mostly just thins the bed. The
bias is a control, not a constant: at 0 it follows raw energy, at 1 it weights
toward 1-3 kHz.
Ranking happens in dB, which matters more than it looks. Speech spreads 20-30 dB
across these bands — it falls off roughly 6 dB per octave above the fundamental
— so a weighting has to be on that scale to move anything at all. A
multiplicative weight of `1 - bias + bias * shaped` is bounded below by
`1 - bias`, capping its influence at 10*log10(1/(1 - bias)): 5.2 dB at the 0.7
default, 3 dB at 0.5. That is no influence against a real voice — every bias
short of ~0.95 would rank exactly like bias 0 and carve the fundamental, the
outcome the bias exists to prevent, while looking decisive against a fixture
whose bands sit 2 dB apart. So the bias is a dB penalty, zero at 2 kHz and worth
up to 30 dB at full strength, and relative cut depths come from a dB difference
rather than a ratio of weighted linear powers.
The bias reweights ranking without overriding the spectrum — a band the voice
has no energy in is not worth carving, and scores -Infinity rather than
competing — so a strongly low-pitched voice can still select low at full bias.
What the tests hold is that biasing never selects lower than the unbiased
ranking, that the DEFAULT bias reaches the presence region on a voice with a
realistic tilt, and that bias 0 still follows raw power exactly.
Includes a radix-2 FFT rather than a dependency; one Welch-style averaged
spectrum over third-octave bands does not justify pulling in a DSP library.
Reads `data-fx-chain` off an audio element and runs the chain over the trimmed
WAV before volume automation is baked in — effects should see the raw signal,
and the envelope belongs on their output.
The processing happens in an OfflineAudioContext inside the headless browser
the engine already drives, running the same graph builders the studio previews
with. That is the point of the approach: one implementation per effect, so the
render agreeing with the preview is a property of the architecture rather than
a tolerance to police. Reimplementing each effect as an FFmpeg filter would
mean two implementations to keep in step, and for the dynamics processors and
modulated delays there is no filter that behaves the same way.
`build:audio-fx-runtime` bundles the graph builders into an injectable IIFE,
following the same pattern as the existing runtime artifacts, so the browser
runs exactly the code the studio does.
The page loads from a file:// URL rather than about:blank because AudioWorklet
is only exposed in a secure context — the compressor, limiter, gate and
bitcrush processors would otherwise fail to register with an opaque error.
file:// qualifies and needs no listening socket.
The chain is serialised into the attribute the way colour grading carries its
config, so there is no side-car file to resolve or lose.
An FX failure is fatal for the whole mix rather than a per-track soft failure.
Every other audio failure mode degrades gracefully — the track drops, siblings
continue — but substituting the dry signal for a processed one ships a render
that sounds plausible and is not what the author set up. Since the per-element
work races under Promise.all, an internal AbortController chained off the
caller's signal aborts in-flight siblings before workDir is removed.
Four defects in the graph builders, all found by review rather than by ear.
**Reverb was unusable at its own defaults.** A ConvolverNode applies the
impulse's gain whole — the graph sets `normalize = false` so a room is
deterministic rather than browser-defined — but the impulse was raw decaying
noise. Measured L2 at the registry default (size 0.7 / damping 0.5): 46.4, or
+33.3 dB, putting the wet path ~24 dB over dry at the default `wet: 0.35`. It is
now normalised to unit energy, so the wet knob means what it says. Preview and
render both convolve this buffer, so they stayed identical throughout — equally
deafening before, equally correct now.
**Phaser in_gain/out_gain trim the signal entering and leaving the effect**, not
a wet/dry pair. Wired to the wet and dry legs, "Input" muted the dry path and
the two defaults summed to 1.14, so inserting a phaser raised the track level.
They are now input and output trims with the legs summed at unity. Its declared
waveform is also honoured: `lfo.type` was never assigned, so the default
"Triangular" was silently a sine.
**The dynamics worklets held one envelope across a channel-major loop.** The
followers advance per sample, so on stereo a 20 ms attack behaved as 10 ms, and
the right channel's gain came from an envelope that had already traversed the
left — the two ducked differently from the same input and the image pumped.
State is now per channel, as is the gate's smoothed gain and bitcrush's
sample-hold counter, which previously advanced only on the last channel and left
every earlier one frozen for a whole quantum. The gate also honours the knee it
declares instead of chattering on material sitting at the threshold.
**A one-pole filter's cutoff was swallowed in preview.** Its coefficients are
fixed at construction, so `update` cannot push a new frequency — but the shape
signature carried only type and pole count, so a cutoff change looked like a
values-only edit and went into a no-op updater. Preview kept filtering at the
old frequency while the render used the new one: a preview/render divergence in
exactly the two effects that do not use a BiquadFilterNode.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
One graph builder per `web` id, turning the registry's declarations into
running audio.
Every node exposes `update`, so turning a dial re-parameterises the live graph
rather than rebuilding it: an AudioParam change lands on the next 128-sample
quantum, about 2.7 ms at 48 kHz. `buildFxChain` reports whether an update could
be applied in place — adding or bypassing an effect, or switching a filter
between one and two poles (which changes the node type from BiquadFilterNode to
IIRFilterNode), changes the graph's shape and returns false so the caller
rebuilds.
Four effects have no native node and run as AudioWorklet processors:
compressor, limiter, gate and bitcrush. The module is registered from a data:
URL rather than a blob:, because a blob inherits the page origin and is opaque
on a file:// page, where it fails with an unhelpful AbortError.
Reverb has no single node either. `synthesizeReverbImpulse` generates a tail
from the room parameters, seeded so the same room sounds the same on every
machine, and the ConvolverNode uses it.
Tests cover the wiring — which nodes get built, how they connect, parameter
application and clamping, in-place update versus rebuild, disposal — against a
fake AudioContext, since happy-dom has no Web Audio.
Three parameters were declared with ranges, defaults and hints, and read by no
builder — dials an author could turn with no audible result.
- `chorus.decay` and `bitcrush.aa`: removed. FFmpeg's chorus feeds a decay back
into its delay line and a bitcrusher's anti-alias needs a real filter; adding
either is new DSP, not a fix, so the honest move is to stop advertising them.
- `lowshelf.q` / `highshelf.q`: removed. The Web Audio spec leaves Q unused for
shelving filters, so the control moved nothing — and because the shared Q
helper marks it automatable, an author could draw an envelope on it and hear
nothing at all.
`phaser.decay` and `gate.knee` stay: the first drives the sweep depth, and the
second is now read by the gate's processor.
A test asserts each of these directly, since the existing exposure invariant only
checks that a flagged parameter reaches an AudioParam — a parameter the node then
ignores passes it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
One declarative description of every effect that can be applied to an audio
track: fourteen across filters, dynamics, non-linear and time, each exposing
its full parameter surface rather than a curated subset.
Parameters carry the range, step, unit and scale a control needs, so a panel
can generate its UI from this rather than hard-coding a form per effect, and a
value that survives `normalizeAudioFxParams` is always safe to realise.
Everything is declared in the units a person thinks in — dB, ms, Hz.
Parsing rejects an unknown effect id rather than skipping the node. A chain
that quietly loses an effect renders something other than what was authored,
which is worse than refusing to load it.
Data only: no audio is produced here. The graph that realises each effect is
referenced by the `web` id and lands in the next change, which keeps this
module free of browser globals so the engine and the linter can import it.
Extends the studio-preview fix to the render path and the asset-discovery
utilities, which share the same resolver and had the same defect.
`rewriteAssetPath` takes an optional `assetExists` probe. A plain relative ref
authored in a sub-composition (`_shared.css`, `clip.mp4`) is re-pointed at the
composition's own directory when that sibling exists on disk; project-root refs
with no sibling (the registry's `assets/logo.png` convention) stay as authored.
Callers that can see the filesystem supply the probe, so the module stays free
of node:fs.
Also fixes a second defect in the inliner: `<head>` <link> hrefs and external
script srcs are hoisted into the root document but never went through the
rewrite at all, so even the documented `../` form escaped the project and 404'd
at render time.
Wired into the preview bundler, the producer compiler, the studio preview
builder, the HEVC preview lint, the project lint's asset scans, publish proxy
baking, and media-treatment source resolution.
R6/R7 blockers.
An already-open Studio kept emitting server-side render telemetry after
another process disabled CLI telemetry. refreshTelemetryPosture() only ran
while serving a fresh SPA document and on /api/telemetry-identity, which
Studio has no consumer for, so the render POST and its async outcome used
the posture cached when the preview server booted. It now refreshes at the
render boundary and again immediately before the completion/error event,
so an opt-out during a long render is honoured.
The identity tests were passing vacuously: their mocks omitted
readConfigFresh and resetTelemetryPostureCache, and the resulting
missing-export error was swallowed by the refresh's own catch. Mocked
properly, plus the enabled -> external disable -> next response transition
and the suppression path at the layer that drops the event.
A full reset also did not persist its new lineage in a long-lived process:
syncInstallState returned early on a process-lifetime memo even after
~/.hyperframes was deleted, so install-state was never recreated and the
next config-only re-mint rolled a third seed instead of inheriting the
second. The memo is now revalidated against the file.
Also drops a stale reference to assertNoOverdueCanaries and stops the
workflow and docs claiming the sunset job routes anything to the owner —
it names them in the run log and notifies nobody.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- A long-lived preview cached its telemetry posture in two places
(readConfig and shouldTrack). Running `telemetry disable` in another
terminal left it resolving canaries and injecting the CLI id for hours.
Both caches are now dropped together at a request boundary.
- Studio minted and shipped a telemetry id for every render regardless of
the browser profile's opt-out, and the server emitted the outcome under
CLI policy, which cannot see localStorage or DNT. The browser now sends
an explicit telemetryOptOut, distinct from an old client's omission.
- Any non-empty HYPERFRAMES_PREVIEW_HOST disabled the DNS-rebinding guard,
so even a loopback bind accepted a hostile Host. The guard now holds for
loopback binds and, on a LAN bind, admits only names this machine
answers on.
- sunsetAfter had no reader of the current date. A scheduled workflow runs
scripts/check-canary-sunset.ts weekly, so a failure lands on the
rollout's owner rather than on an unrelated PR author.
- The install-state seed memo outlived `rm -rf ~/.hyperframes`,
resurrecting a cleared cohort. Removed; it only saved a read on a
readConfig cache miss.
Docs updated for the Host rule and the 100% exclusion carve-out.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
R5 blockers
- Negative install-state latch was cached for the process lifetime, but
only `true` is monotonic across processes. A long-lived preview server
held a stale `false` and could re-enrol after another process tripped
the breaker. Only the positive is cached now; `false` re-reads.
- The real breaker writer used writeConfig(), which collapses
{ok:true, mirrored:false} to success, so a run that mirrored nothing
reported done with the latch only on the erasable store. It consumes
writeConfigWithResult and retries until both stores carry it.
Bucketing integrity
- Storage-restricted Studio profiles all bucketed on the literal
"anonymous": computed against the shipped hash, 100% of them were
enrolled in calibration-50 rather than 50%, and they merged into one
PostHog person. Per-session random id instead — persists nothing.
- bucketSeed had read/write authority backwards: install-state is
write-once authoritative, but readConfig took config.json's blindly, so
the stores could hold different seeds until a re-mint flipped every
cohort. Merged on read, like the latch.
- An unwritable ~/.hyperframes with no config.json re-minted per call,
re-rolling the seed on every command, and the "cohorts will not be
stable" warning was unreachable on that path.
- A corrupt PRE-MOVE state file was never deleted, so a machine reset
with `rm -rf ~/.hyperframes` reported predecessorFound/stateFileCorrupt
forever — poisoning the exact metric this work exists to produce.
Opt-out honoring
- CLI canary decisions memoized per process, so `hyperframes telemetry
disable` during a running preview server was ignored for hours while
the server kept serving pre-opt-out decisions. The memo is keyed on the
telemetry posture.
- shouldTrack() memoized, contradicting policy.ts's documented "not
memoized" contract that policy.test.ts asserts.
- The Studio override path resolved the bucket unit eagerly as an
argument, minting and PERSISTING a tracking id for an opted-out profile
— a value evaluateCanary discards unread.
- Storage reads could throw out of telemetry into a post-commit catch
block, reporting an already-committed edit as failed.
- readConfig printed an unsilenceable stderr warning on every invocation
for installs that opted out of telemetry entirely.
Host split
- isLoopbackHost rejected 0.0.0.0, so the documented
HYPERFRAMES_PREVIEW_HOST LAN mode silently lost CLI→Studio identity
stitching and split one user across two PostHog persons. Identity is
now allowed when the operator explicitly opted into LAN binding.
- Corrected the comment claiming the guard refuses spoofed Hosts: a
non-browser client sets Host freely. It is a browser DNS-rebinding
mitigation, not access control, and now says so.
Semantics and test hygiene
- percentage:100 did not mean everyone — exclude and no_unit_id sat above
the fast path, so the registry's "delete the entry at 100" step was an
unstaged flip for CI and seedless installs.
- CLI cohort adoption returned before evaluateCanary, dropping Studio's
own webdriver exclusion.
- overdueCanaries() was asserted against wall-clock time, so the whole
core suite would go red on 2026-09-15 for every unrelated PR; and `>`
against midnight made a canary overdue ON its sunset date.
- Statistical assertions ran on unseeded randomUUID() populations tight
enough to fail ~1 run in 200. Seeded.
Also: broke a config -> policy -> transport -> config import cycle by
moving POSTHOG_API_KEY to a leaf module.
Tests: 2347 CLI (bundle absent), 3153 Studio, 1450 core. Fault injection
covers the latch, seed authority, LAN identity, webdriver exclusion and
the anonymous-bucket fix. Two pre-existing tests asserted behaviour these
findings identify as wrong (shouldTrack memoization, 100%-excludes-CI)
and were rewritten with the reasoning stated.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Six findings from review, none behaviour-critical on their own but three
of them quietly corrupt the data the rollout is judged by.
Endpoint no longer serves bucketSeed (studioServer.ts). Studio gets its
canary answers from the injected decisions map now, so nothing needed the
seed over HTTP — and an unauthenticated local endpoint is a strictly
worse place for it than a script scoped to Studio's own document. The
endpoint itself predates this PR and still serves distinctId, so it also
gains a Host guard: a remote page can rebind its hostname to 127.0.0.1
and read the response as same-origin, but the request still carries THAT
hostname, which is what makes it refusable.
predecessorFound no longer reports corruption as a fresh install. It
returned null for both "file absent" and "file unreadable", so a partial
disk write looked like a new machine — understating recoverable churn,
the one thing the field measures. Now distinguishes absent from corrupt
and emits install_state_file_corrupt alongside.
A mangled markerAt no longer discards a salvageable bucketSeed. markerAt
is only a timestamp and can be restamped; the seed cannot be recovered,
and losing it silently re-rolls the install's cohort.
The seed backfill no longer ignores its write result. An unwritable
~/.hyperframes meant a different seed every invocation with no
diagnostic, and made the field's own "backfilled once" docstring false.
Warns once per process with the underlying error.
FNV-1a's ASCII constraint is now explicit rather than incidental. It
hashes UTF-16 code units while reference FNV-1a is byte-oriented, so the
two agree only on ASCII; the registry's kebab-case assertion is what
makes non-ASCII unreachable, and both ends now say so. Not a live bug —
names are kebab-case and units are UUIDs.
de-parallel-router is pinned at 0%. The registry is data, so a ramp is a
one-line edit with no review surface, and its own description says to
ramp only alongside the circuit breaker.
Tests: 8 new (corruption vs absence, seed salvage, backfill write
failure, 17 host-guard cases, registry pin). One existing test asserted
predecessorFound: false on corruption — that was the bug, updated with a
note. Fault injection: restoring the old corrupt handling fails 4.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Both reviewers flagged the same gap: seed injection was gated on
telemetryShouldTrack(), but canary EVALUATION was not. An install with
DO_NOT_TRACK=1 was still bucketed and still had real code paths flipped
(e.g. HF_DE_PARALLEL_ROUTER), silently and unmeasurably.
A canary is a measured rollout — we enrol a slice precisely so it can be
compared against everyone else. An install that sends nothing can't be
compared, so enrolling it buys no signal and only changes that user's
code path, on an experimental feature, without their knowledge. That is
the wrong side of an opt-out.
Resolves to a new `telemetry_opt_out` reason BEFORE bucketing, so no
cohort is assigned at all. Distinct from `excluded` because "why is my
canary off" has a very different answer for CI than for opted-out, and
the reason never reaches telemetry by construction.
Covers every opt-out route: persisted preference, the runtime env vars
and dev/telemetry-disabled builds via policy.ts, and Studio's
hyperframes-studio:telemetryDisabled.
An explicit HF_CANARY_* / ?hf_canary_*= override still wins — a
deliberate local choice, not silent enrolment, and the documented way to
exercise a canary with telemetry off.
The CLI check mirrors shouldTrack() rather than importing it: client.ts
already imports canary.ts for canaryEventProperties, so depending on it
would be a cycle. Both read the same two inputs, so they cannot disagree.
Tests: 9 new across CLI and Studio (preference off, each runtime
override, no bucket assigned, override still honoured, flag properties
all-false). Fault injection: removing the CLI gate fails 6, removing the
Studio gate fails 3.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Verified against packages/*/src env reads: 57 distinct HF_*/PRODUCER_*
toggles pre-existing this branch (the raw grep said 59, but two of those
are HF_CANARY_TEST_* fixtures introduced by this branch's own tests).
The number is cited externally now, so it should match what the repo
actually has.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Registers two canaries that gate nothing — `calibration-10` (10%) and
`calibration-50` (50%) — so the rollout mechanism can be proven against real
traffic before any real feature depends on it. Zero behavioural risk: they are
read by nothing.
They answer what the unit tests structurally cannot. The tests bucket
generated UUIDs and weight every install equally; real render volume is
heavily skewed toward a few heavy installs, and real install ids churn (~25x
more distinct ids over 30 days than in any single day on the desktop render
population).
Four checks, pre-registered in the docs so the read is not post-hoc:
1. ACCURACY — does 10% land at 10%, install-weighted AND event-weighted?
2. DRIFT — how fast does CUMULATIVE exposure climb above target as ids churn?
The instantaneous share is flat by construction; the set of installs
enrolled at some point is not.
3. STABILITY — does any install ever change cohort? Must be zero. Percentages
are held FIXED for the window precisely so a flip is unambiguously a bug;
during a real ramp a false->true flip would be correct instead.
4. CROSS-SURFACE — do the CLI and Studio bindings agree for the same install?
A CLI-launched Studio adopts the CLI id, and 16,961 installs currently
share an id across both surfaces, so this is measurable.
Plus an independence check: overlap between the two calibration canaries
should be ~p1*p2 (~5%), not ~min(p1,p2) (~10%, which would mean every canary
lands on the same unlucky cohort).
The docs also record what calibration CANNOT fix: per-install cohorts never
flip, but a person who wipes their config gets a new id and a fresh roll.
Preventing that needs stable identity across resets, and both candidates were
rejected — hardware fingerprinting correlates the cohort with hardware (fatal
for a rendering experiment, and it survives uninstall) and account identity
covers only ~3.6% of local rendering installs. The drift is therefore a
measured, accepted limit, and the point of calibrating is to size it and pick
canary window lengths accordingly.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Replaces the single `canaries: "a,b"` telemetry property with PostHog's own
flag shape, one property per registered canary:
$feature/canary-de-parallel-router: "true" | "false"
PostHog treats `$feature/<key>` as a first-class flag property, so breakdowns,
funnels split by cohort and the experiment surfaces work on a canary with
nothing configured server-side. The decision still happens locally: the render
path forbids render-time network calls, behaviour must not depend on analytics
being reachable, and neither the CLI nor Studio ships posthog-js (both
hand-roll a batch POST, so there is no SDK to evaluate a real flag with).
Decide locally, analyse natively.
Two decisions worth recording:
- BOTH ARMS ARE EMITTED. A non-enrolled install reports "false" rather than
omitting the property. Absent means "this build predates the canary", which
is a different fact from "this install is control" — collapsing them makes a
ramp unreadable, because you cannot separate a control group from an old
version.
- KEYS ARE NAMESPACED with a `canary-` infix. A real PostHog flag namespace
already exists in this project, owned by the web app (`enable-chat-tab`, set
by posthog-js from `$lib=web` events). Namespacing guarantees a canary key
can never alias a real flag key and have the two fight over one property.
Values are the strings "true"/"false" to match how PostHog records boolean
flag values, so the property is directly comparable to a real flag.
98 core / 1437, 166 cli / 2194, 269 studio / 2982 green; tsc clean across all
three packages.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds the Studio (browser) binding so a canary can span the CLI and the editor,
and fixes a bundling mistake the studio test suite caught.
## The binding
Same public API as the CLI — `isCanaryEnabled("name")` — so a call site reads
identically whether it runs in Node or the browser. Three inputs differ:
- UNIT ID: `resolveStudioDistinctId()`, which already adopts
`window.__HF_CLI_DISTINCT_ID` when the CLI launched Studio. A CLI-launched
Studio therefore lands in the SAME cohort as the CLI: a rollout spanning
render and editor is coherent for that user instead of enrolling their
terminal but not their editor. A test pins that the id is passed through
unmodified — prefixing or re-hashing it would silently break that parity.
- OVERRIDE: no `process.env` in a page, so `?hf_canary_<name>=on` mirrored
into sessionStorage. Session scope is deliberate. A URL is the right carrier
(shareable — "support: open this link"), but persisting a URL-borne override
to localStorage would let one click silently pin a browser into a cohort
forever, long after anyone remembers why. Closing the tab is the reset;
`=reset` clears it explicitly.
- EXCLUSION: `navigator.webdriver` stands in for the CLI's `is_ci`. Automated
browsers mint a fresh localStorage id per run, so they would hop cohorts
between runs — noise in the signal, nothing learned about real users. An
override still reaches them, which is how you test a canary under Playwright.
Studio's `trackEvent` now attaches `canaries` to every event, mirroring the CLI.
## The bundling fix
Importing the `@hyperframes/core` barrel into studio browser code broke two
unrelated hook test files with an esbuild TextEncoder invariant violation. The
barrel re-exports the whole core surface (parsers, lint, studio-server), so it
drags a Node-oriented dependency graph into a browser bundle — the test
failure was the symptom, the bundle bloat was the bug.
`@hyperframes/core` now exposes `./canary` and `./canary-registry`, declared in
packages/core/package-subpaths.json (the generated source of truth for exports —
hand-editing package.json is reverted by the sync script) and marked
`environments: [browser, bun, node]`. Both the studio AND cli bindings import
the leaf modules; the CLI gets the same benefit for a different reason, since
this resolves on the startup path — the reason the producer is lazily loaded.
Verified: the two hook files pass again; 269 studio files / 2982 tests, 98
core / 1433, 166 cli / 2194 green, `bun run lint` clean including the subpath
check. Fault-injection confirms both design decisions are pinned — swapping
session for local storage fails the scope test, prefixing the unit id fails the
CLI/Studio cohort-parity test.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Follow-up to the canary primitive.
Telemetry: every event now carries a `canaries` property listing the cohorts
the install is enrolled in, attached in `trackEvent` so it lands on ALL
events rather than renders only — a staged rollout is only as useful as the
ability to split any metric by cohort. Resolved after the shouldTrack guard,
so opted-out installs never pay for it, and omitted entirely (not null or "")
when the install is in no canary, since PostHog treats those as real values.
Test hardening, after validating the shipped code against 60k synthetic and
101 real fleet install ids:
- Pin FNV-1a against canonical vectors, AND assert the shipped canaryBucket
actually uses that hash. Without the second assertion the first is
tautological — it would only prove the test's own copy is correct while
canary.ts drifted to a different hash, silently reshuffling every live
cohort. Fault-injection confirms only this assertion catches a hash change;
the distribution tests stay green because a perturbed hash is still
well-distributed.
- Tighten the share test from a 0.6x-1.4x band to +/-1 percentage point.
Measured error was 0.16pp at n=60k, so the old band would have passed a
badly skewed hash.
- Add chi-square uniformity across all 100 buckets (chi2 89.0 vs 148.2
critical at p=0.001). A lumpy hash yields roughly the right total share
while overloading some buckets, so the share test alone cannot catch it.
- Assert N concurrent canaries enrol binomially rather than in lockstep:
8 canaries at 10% put ~43% of installs in none and zero in all eight,
matching binomial(8, 0.1). Correlated slices would put ~10% in all eight.
Also verified 88,443 of 88,448 fleet install ids are well-formed UUIDs; the
5 that are not fail closed, which is the intended direction.
Docs: docs/contributing/canary-rollouts.mdx, registered in docs.json (an
unregistered page is invisible in the nav).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds a reusable staged-rollout primitive so a change can ship to a stable
slice of installs instead of all-or-nothing.
The gap it fills: the repo carries ~49 HF_*/PRODUCER_* booleans and every one
is binary — a feature is either off (and therefore unexercised on real
traffic) or on for everyone (and therefore a fleet-wide bet). The
parallel-drawElement router sat in that gap for weeks: default-off produced
almost no signal, and flipping it default-on would have exposed 100% of
eligible installs at once.
Shape:
- packages/core/src/canary.ts — pure evaluator. No fs, no network, no
`process`; the caller supplies the unit id and overrides, so it imports
cleanly into the CLI, producer, engine, studio-server, the browser-side
studio bundle and the embeddable player. FNV-1a rather than node:crypto for
the same reason.
- packages/core/src/canaryRegistry.ts — every rollout in one table (name,
percentage, owner, description, sunsetAfter), so "what is rolling out, to
whom, owned by whom" is answerable without grepping 49 env vars.
- packages/cli/src/telemetry/canary.ts — supplies the three things only the
CLI knows: anonymousId, the HF_CANARY_<FEATURE> override, and is_ci.
Day-to-day API is `isCanaryEnabled("name")`.
Three properties the tests pin, because getting them wrong is subtle:
- Slices are INDEPENDENT per feature: the bucket hashes `feature:unitId`, not
the id alone. Bucketing on the id would hand every concurrent experiment to
the same unlucky cohort and make two rollouts unreadable apart.
- Ramping is INCLUSIVE: `bucket < percentage`, so widening 10 -> 25 keeps the
original cohort and before/after comparisons survive the ramp.
- It fails CLOSED: no unit id, unknown name, or CI install means not enrolled.
A canary exists to bound blast radius, so "we don't know who this is" must
never mean "enrol everyone".
Registry entries also carry a sunset date, and a test fails once one is past
due — a canary that outlives its rollout is a permanent fork of the product
with none of the review a permanent fork would get.
Ships with de-parallel-router registered at 0%: inert, and ready to ramp in a
patch release once #2840 lands.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
