test(producer): add chunk-boundary fixtures per first-party adapter

[jrusso1020]

Description

Phase 4 of the distributed rendering plan: test fixtures (see DISTRIBUTED-RENDERING-PLAN.md §11 Phase 4 + §10 test strategy). This is PR 4.7 of the Phase 4 remainder — the final fixture PR.

This PR adds six per-adapter chunk-boundary fixtures under tests/distributed/{gsap,anime,three,lottie,css,waapi}-boundary/ plus a single bun:test driver (chunkBoundary.test.ts) that exercises each fixture's seek-determinism contract. Each fixture is a 60-frame composition (2s @ 30fps, 320×180) that drives the named adapter through the HyperFrames runtime's seek hook. The test renders each at chunkSize=60 (N=1 chunk, no seams) and chunkSize=15 (N=4 chunks, three seams at frames 15/30/45), then asserts every PNG frame is byte-identical across the two runs.

Why png-sequence: mp4 bitstreams encode keyframe placement directly. At chunkSize=60 libx264 emits 1 IDR; at chunkSize=15 it emits 4 IDRs at frames 0/15/30/45. Those are legitimately different bytes even when the captured pixels are identical. png-sequence's assemble path merges chunk frame directories with no re-encode, so per-frame byte equality is exactly pixel equality — the strongest contract a distributed render can satisfy.

Fixture design (each is ~60 lines of HTML):

• gsap-boundary — single GSAP tl.to(...) driving translateX + rotation linearly across 2s.
• anime-boundary — anime.js v4 timeline registered via window.__hfAnime.
• three-boundary — minimal Three.js scene; cube rotation derived from window.__hfThreeTime.
• lottie-boundary — inline Lottie JSON (rectangle layer animating position+rotation) loaded via lottie-web and registered via window.__hfLottie.
• css-boundary — pure @keyframes animation; the HyperFrames CSS adapter seeks via animation-delay.
• waapi-boundary — element.animate() with linear keyframes; runtime sets currentTime per frame.

The fixtures intentionally omit meta.json so the regression-harness discovery skips them with a clear missing meta.json log (they're driven exclusively by chunkBoundary.test.ts). The test passes rejectOnSystemFonts: false because some adapter bundles (notably anime.js's IIFE) embed CSS-shaped strings inside their JS source — font-family: ui-monospace, monospace for internal devtools styling — which validateNoSystemFonts's document-wide regex would otherwise false-positive on every adapter fixture that loads such a bundle. The fixtures display no text, so the relaxed font validation doesn't affect the contract under test.

The 7th test case is a layout sanity check that asserts every expected *-boundary fixture directory exists.

Testing

• bun test packages/producer/src/services/distributed/chunkBoundary.test.ts — 7 tests pass on host (6 adapters × byte-identical N=1 vs N=4 + the layout check, 41.6s)
• bun test packages/producer/src/services/distributed/ — all 49 distributed unit tests pass (43.6s)
• bun run --cwd packages/producer docker:test:distributed font-variant-numeric many-cuts gsap-letters-render-compat style-1-prod sub-composition-video mp4-h264-sdr png-sequence mov-prores mp4-h265-sdr -- --sequential — full smoke set + all four prior stacked fixtures pass (9/9)
• bunx oxlint + bunx oxfmt --check clean
• bunx tsc --noEmit (producer package) clean

After this stack lands

The Phase 4 fixture set is complete: PR 4.6 (#844) pins cross-worker idempotency; 4.2/4.3/4.4/4.5 (#845/#851/#847/#848) prove each format produces correct chunked output at the fixture's minPsnr; 4.3-pre (#850) added the codec knob H.265 needed; and this PR proves each first-party adapter's seek-determinism survives chunk seams. Phase 5 (CLI surface for hyperframes plan/chunk/assemble) and Phase 6 (AWS Lambda turnkey) are unblocked.

🤖 Generated with Claude Code

[jrusso1020]

• test(producer): add chunk-boundary fixtures per first-party adapter #852 👈 (View in Graphite)
• test(producer): add mp4 H.265 SDR distributed fixture #851
• feat(producer): add codec knob to DistributedRenderConfig #850
• test(producer): add mov ProRes distributed fixture #848
• test(producer): add png-sequence distributed fixture #847
• test(producer): add mp4 H.264 SDR distributed fixture #845
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[miguel-heygen]

Re-reviewed as part of the fixture stack. LGTM.

[vanceingalls]

Reviewed the largest fixture PR in the Phase 4 stack — chunkBoundary.test.ts end-to-end plus all six adapter fixtures, cross-referenced against the runtime adapters in packages/core/src/runtime/adapters/ and plan.ts. Prior coverage: @miguel-heygen approved on the re-review. Findings below are gaps.

Strengths

• png-sequence framing of the boundary contract (chunkBoundary.test.ts:6-12): textbook "say WHY, not WHAT" — calling out that mp4's IDR placement is a legitimately-different-bytes confound and that png-sequence's no-re-encode assemble path round-trips pixel equality is exactly the level of explanation a future reader needs. This is the rationale I'd want to inherit if I were debugging a flake in three months.
• Cross-adapter symmetry: every fixture drives a 40×40 box from translateX 0 → 280 + rotate 0 → 360 linearly across 2s. Same input geometry, same adapter-specific seek path, same boundary frames (15/30/45). Symmetric coverage was the explicit asymmetry concern from the review brief — this PR doesn't have it.
• Adapter-hook fidelity: spot-checked __hfAnime / __hfLottie / __hfThreeTime + hf-seek against packages/core/src/runtime/adapters/{animejs,lottie,three}.ts and document.getAnimations() against waapi.ts / css.ts — each fixture's contract matches the real seek path the runtime invokes in chunk workers.

Important

• important — Asymmetric soft-skip leaves N=4 unguarded (chunkBoundary.test.ts:126-143). Only planAndAssemble(workOne) is wrapped in the HOST_CHROME_FAILURE_PATTERNS try/catch; the workFour call right after is not. Cold-Chrome / SwiftShader flakes are exactly the kind of thing that can succeed on the first render and trip on the second — when that happens on a developer host, the existing convention says "soft skip, Docker covers it," but this test will hard-fail instead. Either wrap both renders in the same skip pattern, or document why N=4 is deliberately treated as hard (and the convention says it shouldn't be).
• important — Vacuously-passing length assertion (chunkBoundary.test.ts:156-160). expect(framesOne.length).toBe(framesFour.length) will pass if both runs silently produce 0 frames (or 30, or any matching subset). The PR description states the contract is "60-frame composition (2s @ 30fps)" — pin the absolute count explicitly: expect(framesOne.length).toBe(60). Without that, a regression that truncates both renders identically (e.g. a probe that misreads duration as 0s) shows as a green test.
• important — External CDN dependencies on test fixtures. Every fixture loads gsap, anime, lottie, or three from cdnjs / jsdelivr at test time. CI hermeticity, supply-chain risk, and CDN-outage flake all become this test's failure surface. Worth comparing against how packages/producer/tests/animejs-adapter/src/index.html (the existing in-repo adapter fixture) handles its bundles — if that one also CDN-loads, this is a pre-existing systemic gap and a follow-up ticket; if it ships local copies, this PR diverges from the established pattern and should match it.

Nits

• nit — CDN version drift within the PR. gsap-boundary/src/index.html pins gsap@3.12.2 (cdnjs), anime-boundary/src/index.html pins gsap@3.14.2 (jsdelivr). Both gsap timelines are empty duration-drivers so it doesn't affect the contract, but pinning one version + one CDN across all six fixtures saves a future "why this discrepancy" hunt.
• nit — Comment on the conditional spread understates the safety (regression-harness-distributed.ts simplify commit). The comment says "codec: undefined is a no-op" but plan.ts:490-495 actually throws if config.codec !== undefined for any non-mp4 format. The && input.codec !== undefined half of the spread guard is doing real work — worth reflecting in the comment so the next refactorer doesn't drop it.
• nit — PR description test count vs. host soft-skip. Description claims "7 tests pass on host" but the host soft-skip path means an adapter that tripped SwiftShader would skip silently and still report green. Worth confirming the Docker harness run covered all six adapters end-to-end (no skips), since the host run alone isn't sufficient evidence of the contract holding.

Notes

• Reviewed under Rule 5 with mergeable_state: UNSTABLE: pulled gh pr checks — only fail-class indicators are Player perf / preview-regression Detect changes jobs that were CANCELLED (superseded by later runs that passed). The long-tail regression-shards are still IN_PROGRESS at review time — not gating my verdict on code merits, but worth a glance before merge.
• Did not engage Rames or Magi findings (none posted at review time).

Verdict: COMMENT
Reasoning: Two important findings (asymmetric soft-skip + vacuous length assertion) are tightening, not blocking — the test still catches real boundary regressions, just leaves a couple of false-green windows. Comfortable shipping with the asks addressed in this PR or as fast-follow.

Review by Vai

[jrusso1020]

Thanks @vanceingalls, @miguel-heygen — all important findings addressed in commit 2610b303:

1. Asymmetric soft-skip — both planAndAssemble(workOne) and planAndAssemble(workFour) are now wrapped in the same HOST_CHROME_FAILURE_PATTERNS catch via a local runRender() helper. A cold-Chrome / SwiftShader flake on either render skips cleanly instead of hard-failing on N=4.

2. Vacuously-passing length assertion — added EXPECTED_FRAME_COUNT = 60 and expect(framesOne.length).toBe(EXPECTED_FRAME_COUNT) + same for framesFour. A regression that identically truncates both renders (e.g. probe stage reading duration as 0s) now shows red.

3. CDN version drift — anime-boundary loaded gsap@3.14.2 from jsdelivr while every other boundary fixture loaded 3.12.2 from cdnjs. Unified on cdnjs/3.12.2.

4. @miguel-heygen VIDEO_EXT narrowing — added an explicit cast at the indexing site (VIDEO_EXT[outputFormat as "mp4" | "mov" | "webm"]) rather than widening the lookup to include "png-sequence". The isPngSequence ternary above the indexing site is the runtime guard.

External-CDN dependency on test fixtures is a real systemic concern (matches the pre-existing pattern in tests/animejs-adapter/) — leaving for a follow-up that converts all adapter fixtures to vendored bundles.

The base branch was changed.

[miguel-heygen]

Re-stamping after rebase. LGTM.

[jrusso1020]

Merge activity

• May 15, 3:09 AM UTC: @jrusso1020 merged this pull request with Graphite.