You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Call video retains only one frame with ScreenCaptureKit queue depth 1 #84
Installed notarized builds 37 and 38 finalized valid HEVC files but retained only one frame per video span, even while a Finder window moved visibly. Both authoritative audio tracks continued normally, so the failure was video-only and easy to miss from Call readiness alone.
Physical evidence
Build 37: live audio → video → audio → video stayed in one audio epoch, but both video spans contained one frame.
Build 38: BGRA was restored, but the one-frame queue still produced one frame (ffprobe: 0.0667 s).
Build 39: restoring ScreenCaptureKit queue depth 2 produced 274 HEVC frames over 18.675 s at 1920×1080, with both PCM legs continuous.
Resolution
Keep queue depth 2 and the separate immediate screenshot-yield controller. Add physical multi-frame evidence to candidate qualification; a finalized MP4 alone is insufficient. Fixed on PR #69 at 2f925e3e7dae2779b090b53ed583ad336c37c62b; close after merge.
Installed notarized build 39 passed the physical regression check: Finder window motion produced 274 HEVC frames over 18.675 seconds at 1920×1080 (15 fps target), rather than the single 0.0667-second frame seen in builds 37–38. A later 175-second Call produced six sequential crash-forward video segments. Both microphone and system PCM remained continuous. Fix is physically proven but the issue stays open until PR #69 merges.
Activity
nikshilov commentedon Aug 16, 2026
Installed notarized build 39 passed the physical regression check: Finder window motion produced 274 HEVC frames over 18.675 seconds at 1920×1080 (15 fps target), rather than the single 0.0667-second frame seen in builds 37–38. A later 175-second Call produced six sequential crash-forward video segments. Both microphone and system PCM remained continuous. Fix is physically proven but the issue stays open until PR #69 merges.