feat(cli): check_report counts the browser-playable copies a check waited for by miguel-heygen · Pull Request #7316 · heygen-com/hyperframes

miguel-heygen · 2026-11-06T12:46:49Z

What changed

check_report gains copies_total, copies_waited_ms and copies_unfinished, sent only when the project had videos Chrome cannot decode: how many there were, how long the check waited for their browser-playable copies, and how many were still not done when the wait budget stopped it (copies ffmpeg failed on in a normal run are not counted). Counts and milliseconds only, no names or paths. preResolveHostileMediaProxies reports the numbers through a callback just before it may throw CheckCopiesUnfinishedError, runBrowserCheck keeps them as a partial result, so a check that stopped on the copy wait still sends them.

Why

A check that stops with media_copies_unfinished cannot be counted today: the event has no copies facts, and stop_code (#7305) says only that it stopped. Whether the wait is stuck or just slow on many large videos is not answerable from the data.

What I measured

fails without the fix: packages/cli/src/commands/check.test.ts "a check stopped by the copy wait still sends the copy facts and stays browserSkipped" (copy facts then CheckCopiesUnfinishedError; with the earlier head's pipeline it read browserSkipped false), and packages/cli/src/telemetry/events.test.ts "sends what making browser-playable copies cost as counts and milliseconds, and omits it without videos" (with events.ts at main it fails; 1 failed, 82 passed). With the change that file and commands/check.test.ts pass (246).

What I did NOT exercise

No end-to-end check run that waits for copies and reads the sent event; the plumbing from the copy step to the event is typed but not run.
tsc on the box printed errors for @hyperframes/inspector/live (not built in that tree), none in the lines this changes.

Size

About 30 changed lines in 5 files: a lone telemetry addition; nothing open on this event to carry it.

…ited for

heygengenesis

Verdict: REQUEST CHANGES · 2 blocker · 1 important · 0 nits · head 5cbc9a13

blocker: required Comments check fails at this head: checkTypes.ts comment share rises 8.2% (38/465 → 39/473); the PR body doesn't mention it. ↓ inline
blocker: the new partial flips browserSkipped to false on the exact stop this PR targets. ↓ inline
important: copies_unfinished also counts copies that failed. ↓ inline

Scope

Read all 5 files at head, plus checkDeadline.ts (keepPartial/partialResult), checkPipeline.ts catch path (L1748–1751) and commands/check.ts browserSkipped consumers. The PR names a test that fails without the fix (events.test.ts), which covers the event mapping only. Nothing exercises the stop path end to end.

heygengenesis · 2026-11-06T12:49:43Z

  const html = await loadCheckHtmlWithDeferredMedia(project.dir);
  deadline?.enter("copies");
+  let mediaCopies: MediaCopyFacts | undefined;
+  deadline?.keepPartial(() => (mediaCopies ? { mediaCopies } : {}));


blocker: when the copy wait throws CheckCopiesUnfinishedError, checkPipeline.ts:1750 builds skipped: Object.keys(found).length === 0 from the partial. found is now { mediaCopies }, so browserSkipped becomes false even though Chrome never launched. The human report then drops the "Browser session never ran — … empty placeholders, not a clean pass" warning (commands/check.ts:688), and the JSON report and artifact metadata show browserSkipped: false over empty layout/motion/contrast sections.

Fix: carry mediaCopies outside keepPartial, or have line 1750 ignore mediaCopies when deciding skipped. Then add a check.test.ts case where runBrowserCheck reports copy facts and then throws CheckCopiesUnfinishedError, asserting browserSkipped === true and that copies_* are sent.

heygengenesis · 2026-11-06T12:49:43Z

  const unfinished = sources.flatMap(({ file }, index) =>
    results[index]!.status === "rejected" ? [`${file} ${heartbeat.percent(index)}`] : [],
  );
+  record?.({ total, waitedMs: Date.now() - startedAt, unfinished: unfinished.length });


important: unfinished counts every rejected result, including copies that ffmpeg failed on (lines 341–344 handle those as warnings when the budget isn't hit). So a check that finishes normally with 2 failed copies sends copies_unfinished: 2. That mixes slow or stuck copies with broken ones, which is the distinction this field exists for. Count only the rejections caused by the abort, or send failures as their own field.

heygengenesis · 2026-11-06T12:49:43Z

  overlapMs: number;
 }

+/** What a check's browser-playable copies cost: videos, ms waited, videos not done. */


blocker (CI): this line makes the required Comments check fail (RAISED comment share: checkTypes.ts 8.2% (39/473), base 8.2% (38/465)). The field names already say what the doc comment says. Drop the comment, or cut one elsewhere in the file.

…t only abort-caused unfinished

miguel-heygen · 2026-11-06T12:55:07Z

Head a523981: the doc comment on checkTypes.ts is dropped (Comments check); the catch path ignores mediaCopies when deciding browserSkipped, with a check.test.ts case for the copy-wait stop (copy facts sent, browserSkipped true); copies_unfinished is 0 unless the budget stopped the wait, so ffmpeg-failed copies are not counted. The red/green run for the new test is running on a loaded box; I will post it.

heygengenesis

Verdict: APPROVE — 3 resolved, 0 open, 0 new · 5cbc9a13 → a523981f

Comments now passes. With mediaCopies left out of the skipped count, the copy-wait stop reads browserSkipped: true, and the new check.test.ts case fails without that change. unfinished is zeroed unless the budget fired. The rest of CI was still running when I approved, so required checks still gate the merge.

miguel-heygen · 2026-11-06T12:58:03Z

Run on a Linux box (bun install, build, vitest): the new check.test.ts case "a check stopped by the copy wait still sends the copy facts and stays browserSkipped" fails with the previous head's checkPipeline.ts (1 failed), and events.test.ts plus check.test.ts pass with this head (247).

github-actions · 2026-11-06T13:04:37Z

Edit accuracy: accurate 267 (base branch 267), smooth 106 of those

The gate passes.
Smoothness is reported in the artifact, not gated. A case fails only if it fails 2 of 3 runs.

Quarantined, measured but not gated (8)

seqplainkeys-none-center-r0-nested-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): pass, passes
seqplainkeys-none-center-r0-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): pass, passes
seqplainkeys-none-center-r30-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run
seqplainkeys-none-pct-r0-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): pass, passes
seqplainkeys-none-pct-r30-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run
seqplainkeys-none-px-r0-nested-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): fail / pass / pass, passes
seqplainkeys-none-px-r30-nested-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run
seqplainkeys-none-px-r30-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run

heygengenesis

Verdict: APPROVE — 0 resolved, 0 open, 0 new · a523981f → 6acbc419

Only formatting changed, in the two test files. CI, preview-regression, Comments and CodeQL pass on this head. The earlier preview-regression red came from the format preflight, not from main. Windows render verification was still running; it also fails on main at 410a516c (engine-cli-1).

…view to heygen-com#7250 Adopts heygen-com#7310 (MPEG-4 Part 2 / MJPEG get browser-playable copies), heygen-com#7316 (check_report counts the copies a check waited for) and heygen-com#7317 (render reads a composition src like the preview and refuses paths outside the project). Defers heygen-com#7309 (needs heygen-com#6912); heygen-com#7314 does not apply. PR axis advances to heygen-com#7250 (heygen-com#7243 closed unmerged, heygen-com#7249 open and pinned as monitored; monitored 96->97). Co-authored-by: Miguel Ángel <miguel.sierra@heygen.com> Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

miguel-heygen added 2 commits November 6, 2026 07:35

feat(cli): check_report counts the browser-playable copies a check wa…

fadd474

…ited for

style(cli): shorten the copy facts comment under the line limit

5cbc9a1

heygengenesis Bot requested changes Nov 6, 2026

View reviewed changes

fix(cli): a check stopped by the copy wait stays browserSkipped; coun…

a523981

…t only abort-caused unfinished

heygengenesis Bot approved these changes Nov 6, 2026

View reviewed changes

style(cli): format the copy facts tests

6acbc41

heygengenesis Bot approved these changes Nov 6, 2026

View reviewed changes

miguel-heygen added this pull request to the merge queue Nov 6, 2026

Merged via the queue into main with commit de2d21a Nov 6, 2026
84 checks passed

miguel-heygen deleted the drenderq/check-copies-fields branch November 6, 2026 14:03

Navigation Menu

Repository navigation

feat(cli): check_report counts the browser-playable copies a check waited for - #7316

feat(cli): check_report counts the browser-playable copies a check waited for#7316
miguel-heygen merged 4 commits into
mainfrom
drenderq/check-copies-fields

miguel-heygen commented Nov 6, 2026 •

edited

Loading

Uh oh!

heygengenesis Bot left a comment

Uh oh!

heygengenesis Bot Nov 6, 2026

Uh oh!

heygengenesis Bot Nov 6, 2026

Uh oh!

heygengenesis Bot Nov 6, 2026

Uh oh!

miguel-heygen commented Nov 6, 2026

Uh oh!

heygengenesis Bot left a comment

Uh oh!

miguel-heygen commented Nov 6, 2026

Uh oh!

github-actions Bot commented Nov 6, 2026 •

edited

Loading

Uh oh!

heygengenesis Bot left a comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

miguel-heygen commented Nov 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What changed

Why

What I measured

What I did NOT exercise

Size

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot Nov 6, 2026

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot Nov 6, 2026

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot Nov 6, 2026

Choose a reason for hiding this comment

Uh oh!

miguel-heygen commented Nov 6, 2026

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

miguel-heygen commented Nov 6, 2026

Uh oh!

github-actions Bot commented Nov 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Edit accuracy: accurate 267 (base branch 267), smooth 106 of those

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

miguel-heygen commented Nov 6, 2026 •

edited

Loading

github-actions Bot commented Nov 6, 2026 •

edited

Loading