* fix(core): discover sub-composition hosts to a fixed point during inlining
inlineSubCompositions collected [data-composition-src] hosts from the
document once, before inlining began, then processed that fixed list in
a single flat loop. A host's inlined content could introduce new hosts
of its own (a sub-composition nesting another sub-composition), and
those were never discovered: two-level nesting worked because the
second level's host already existed in the root document, but any
third level's host only appeared inside content the single pass had
already finished walking, so its content silently never rendered.
Host discovery now runs as a work queue: after a host is inlined, the
newly inserted subtree is re-scanned for further hosts, which are
enqueued with their ancestry chain. A host whose source already appears
in its own ancestry is a circular reference and is reported through the
existing onMissingComposition channel instead of being inlined; a depth
ceiling backstops any gap in that check. Existing single- and two-level
fixtures are unaffected.
* fix(producer): assign runtime composition ids to hosts discovered mid-inline
The producer's per-instance runtime id assignment (assignBundledRuntimeCompositionIds)
ran once over the root document's initial hosts, before inlining. With
host discovery now iterating to a fixed point (previous commit), hosts
revealed inside an already-inlined sub-composition were never in that
pre-pass, so the identity map returned undefined for them: two sibling
instances of the same sub-composition, discovered mid-traversal, both
fell back to their shared authored id and clobbered each other's
variablesByComp entry.
hostIdentityMap is now a lazy map: a pre-pass cache hit returns
unchanged, a miss reads the host's authored data-composition-id,
allocates a collision-checked runtime id by scanning the live document,
writes it back, and caches it. No change to the shared inliner's call
signature. Existing single-instance and root-level reusable-template
cases (#2064) are unaffected.
* test(producer): add depth-3 nested sub-composition regression fixture
Adds a minimal three-level chain (root -> level-2 -> level-3) where the
deepest level renders a distinguishing full-frame marker, following the
existing sub-comp-* fixture pattern. Confirmed against the pre-fix
commit (735128a61) in a separate scratch checkout that the fixture fails
without the previous two commits (level-3's marker absent, ~0.4B fewer
matching pixels than the golden expects) and passes with them.
The golden reference video was captured on the team's render host
(devbox) rather than locally, to avoid PSNR drift from font/GPU
differences against whatever renders the regression suite in CI. Local
render against that golden also passes (PSNR ~24.5dB throughout, 0
failed frames of 100 checkpoints), confirming cross-machine consistency
as well.
Producer regression test fixtures
Each subdirectory under this folder is a regression fixture for the
HTML-to-video pipeline. The harness at
packages/producer/src/regression-harness.ts walks every subdirectory,
runs the composition, and PSNR-compares the rendered output against a
checked-in golden baseline.
Fixture layout
<fixture-name>/
├── meta.json # name, tags, PSNR threshold, renderConfig
├── src/
│ ├── index.html # composition entry point
│ └── assets/... # any locally-referenced media
└── output/
├── compiled.html # golden compiled HTML (validated as a snapshot)
└── output.mp4 # golden rendered video
meta.json is validated by validateMetadata in
src/regression-harness.ts. The required fields are:
name(string),description(string),tags(string[])minPsnr(number, dB)maxFrameFailures(integer)minAudioCorrelation(0..1),maxAudioLagWindows(integer ≥1)renderConfig.fps(integer like30or a rational string like"30000/1001")
Optional renderConfig fields:
format—"mp4"(default) or"webm"workers— integer ≥ 1hdr— boolean (defaultfalse)variables— JSON object of render-time variable overrideschunkSize— integer ≥ 1 (used by--mode=distributed-simulated)maxParallelChunks— integer ≥ 1 (used by--mode=distributed-simulated)
Generating / updating a baseline
Always inside Docker. Host Chrome / FFmpeg versions drift across distros, so a baseline captured on the host won't match the bytes CI renders.
# From the repo root.
docker build -t hyperframes-producer:test -f Dockerfile.test .
# Generate a baseline (single fixture):
bun run --cwd packages/producer docker:test:update <fixture-name>
# Generate all baselines (rarely needed):
bun run --cwd packages/producer docker:test:update
The --update flag writes output/compiled.html and output/output.mp4
from the current render. Without --update, the harness compares against
those baselines.
Running the harness locally
# Run every fixture (parallel, in-process mode — the default).
bun run --cwd packages/producer docker:test
# Run a single fixture:
bun run --cwd packages/producer docker:test font-variant-numeric
# Run sequentially (lower memory):
bun run --cwd packages/producer docker:test -- --sequential
Harness modes
--mode=<value> chooses which render path the harness exercises:
| Mode | What it calls | Use for |
|---|---|---|
in-process (default) |
executeRenderJob |
Day-to-day baselines. This is the same path the hyperframes render CLI takes, and it is what produced every existing output/output.mp4. |
distributed-simulated |
plan() → renderChunk() × N → assemble() from @hyperframes/producer/distributed |
Validates the distributed pipeline against the in-process baseline. No Temporal or Lambda involvement — the controller and chunk worker are both this process. |
--mode=distributed-simulated
bun run --cwd packages/producer docker:test -- --mode=distributed-simulated
bun run --cwd packages/producer docker:test font-variant-numeric -- --mode=distributed-simulated
The distributed pipeline cannot run every fixture. Fixtures that fail any of these gates are skipped with a clear log line (and counted as passing in the summary):
fps.den !== 1— distributed mode is integer-fps only (no NTSC).fps.num ∉ {24, 30, 60}— closed set perDistributedRenderConfig.format === "webm"—plan()refuses webm.hdr === true— distributed mode is SDR-only at v1.
Both modes use the fixture's authored minPsnr as the per-test
threshold — distributed must clear the same quality bar in-process
clears against the same frozen baseline. (Internal contract: distributed
vs in-process renders of the same fixture should clear 50 dB PSNR
against each other within the same Docker image. Against the frozen
committed baseline, neither mode reaches that consistently due to
shared encoder/JPEG-capture jitter — that's why the fixture's authored
threshold gates here, not the 50 dB contract value.) An absolute 10 dB
pathology floor catches fully-black-output regressions when a fixture
authors a permissive threshold. A distributed failure at the fixture's
own threshold means the distributed pipeline has drifted — file an
issue rather than relaxing the fixture.
--update is incompatible with --mode=distributed-simulated: the
in-process renderer is the source of truth for baselines, and the
distributed mode's job is to verify the contract against the same
baseline.
Validating PR 4.1 (the harness mode itself)
The smallest fixtures (font-variant-numeric, many-cuts) are sufficient
to verify the mode plumbing end to end:
docker build -t hyperframes-producer:test -f Dockerfile.test .
# In-process: existing behavior, unchanged.
bun run --cwd packages/producer docker:test font-variant-numeric
bun run --cwd packages/producer docker:test many-cuts
# Distributed-simulated: same baselines, distributed pipeline.
bun run --cwd packages/producer docker:test font-variant-numeric -- --mode=distributed-simulated
bun run --cwd packages/producer docker:test many-cuts -- --mode=distributed-simulated
Both modes must pass at each fixture's authored minPsnr against the
existing baseline. If --mode=distributed-simulated fails where
--mode=in-process passes, the distributed primitive has a regression —
file an issue rather than relaxing the fixture's threshold.
Distributed-only fixtures
Fixtures under tests/distributed/<name>/ are authored specifically for
the distributed pipeline. They follow the same meta.json schema as the
top-level fixtures, but they always set chunkSize / maxParallelChunks
so a plan() over the fixture produces N>1 chunks. Each fixture
exercises one of:
- per-format chunk-boundary correctness (mp4 H.264, mp4 H.265, ProRes, png-sequence)
- per-adapter chunk-seam state preservation (GSAP, Anime.js, Three.js, Lottie, CSS, WAAPI)
Each distributed fixture covers one or more equivalence axes — see the
meta.json description field for what a given fixture is locking in.
Fixture pattern (4.2 onward)
Each tests/distributed/<name>/ fixture has the same structure as a
top-level fixture (meta.json + src/index.html + output/output.mp4).
Differences worth knowing:
renderConfig.chunkSizeis required — pick a value that yields N≥2 chunks for your fixture's frame count (e.g. 60 frames atchunkSize: 15produces N=4). Without this the fixture renders in a single chunk and never exercises the seam.- The fixture's ID on the CLI is just
<name>(nodistributed/prefix).bun run --cwd packages/producer docker:test mp4-h264-sdrworks the same as for a top-level fixture. - The
distributedtag is informational — it doesn't gate any tag-based filter today. Add it so the fixture is easy to find by tag. - The composition should stress state continuity across the chunk seams: an animation crossing a seam, a counter, a rotation. A fully-static composition would pass even if chunk-boundary state was broken.
- Baselines must be generated inside Docker — see the section above.
The baseline is rendered by the in-process renderer (the source of
truth for golden output);
--mode=distributed-simulatedis validated against the same baseline.
Tags
Common tags values control which fixtures the default bun test
invocation runs. --exclude-tags transparency (the default for
bun test) skips webm/png-sequence alpha fixtures that need a working
chrome-headless-shell alpha pipeline.