test(producer): regression fixtures for amix crash + remote media bugs#1148
Open
miguel-heygen wants to merge 4 commits into
Open
test(producer): regression fixtures for amix crash + remote media bugs#1148miguel-heygen wants to merge 4 commits into
miguel-heygen wants to merge 4 commits into
Conversation
Adds two regression test fixtures that lock in fixes from #1140 and #1146 so these failure modes cannot silently regress. audio-multi-track-mix (#1140): Three local sine-wave tracks at staggered start times and distinct volumes. Catches any regression where the amix filter graph breaks and output goes silent (the normalize=0 FFmpeg 4.x crash pattern). remote-media-localize (#1146): Remote S3 <video> + <audio crossorigin="anonymous"> composition. Compiled-HTML snapshot verifies src attrs are rewritten to _remote_media/. PSNR check verifies video frames are not black. Audio correlation check verifies audio was not silently dropped. Baselines generated via Dockerfile.test (see packages/producer/tests/README.md). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Generated inside Dockerfile.test (hyperframes-producer:test, rebuilt from main@9ead3a83 to include #1140 and #1146 fixes). audio-multi-track-mix: H.264+AAC 4s, 3 sine tracks audible at correct relative volumes. Catches silent-output regression from amix normalize=0. remote-media-localize: H.264+AAC 3s, compiled.html snapshot shows both <video>/<audio> srcs rewritten to _remote_media/ and crossorigin stripped. Catches black-frame + no-audio regression from remote S3 localization gap. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
audio-multi-track-mix → shard-7 (lightweight, local assets) remote-media-localize → shard-6 (alongside chat, same S3 bucket dependency) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Two bugs caused the regression fixture to fail in test mode (not --update): 1. HTML comment contained angle-bracket element text that the compiler's regex timing pass matched as real video/audio elements. These phantom elements (hf-video-0, hf-audio-0, etc.) had data-start but no data-end, causing validateElementTiming to report errors on the actual compiled HTML. Fix: rewrote the comment to avoid angle-bracket syntax. 2. 3s duration caused an "Unable to parse PSNR output at 2.98749s" error on the last-frame boundary. Bumped to 4s to give breathing room. Both baselines regenerated inside Dockerfile.test (same Docker image, same Chrome/ffmpeg build as CI). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Two new producer regression fixtures to prevent #1140 and #1146 class bugs from silently re-entering.
audio-multi-track-mix(guards against #1140 —amix normalize=0crash)0.6/0.8/1.0)minAudioCorrelation: 0.85— fails if output goes silent (the exact failure mode from the taco composition incident)remote-media-localize(guards against #1146 — remote S3 src → black frames + no audio)<video crossorigin="anonymous">+<audio crossorigin="anonymous">srcattrs are rewritten to_remote_media/paths — fails if localization stops runningminAudioCorrelation: 0.85— fails if audio disappears againBaselines
Generated inside
Dockerfile.testrebuilt frommain@9ead3a83(post-fix). Audio confirmed viaffprobe: both outputs have H.264+AAC streams at the expected durations.Test plan
bun run --cwd packages/producer docker:test audio-multi-track-mixpassesbun run --cwd packages/producer docker:test remote-media-localizepasses--mode=distributed-simulated(or skip if fixture constraints apply)