Repository navigation
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Live MPEG-TS input shutdown discarded packets buffered in the demuxer/parser, and FFmpeg closed live SRT publishers before the peer could deliver their final buffered messages. A 90-second reframer run received all 2,250 source frames but previously processed only 2,248.
Add an explicit InputRec
request-eofoperation that interrupts reads while draining valid packets and emitting the common A/V EOF. IntentionalAVERROR_EXITrecovery is scoped to this operation; other errors remain failures. ExposePacket.eof()for native drain control. With positive SRT linger, the FFmpeg 9 patch drains pending sends and waits for negotiated peer delivery within the bounded close deadline; steady-state buffering is unchanged.Validation: native connected-TCP regression exactly matches file-EOF video/audio packet hashes and timestamps. Production-image GCP transport tests preserved five byte-identical 90-second SRT captures at 200 and 100 ms without a shim. The full GCP GPU reframer at 100 ms preserved all 2,250 frames in each of three outputs, decoded cleanly, and aligned output boundaries. FFmpeg patch-series checks and close-helper regression tests passed. Evidence: gs://aiml-recorder-ai-tests/reframer-srt-e2e/reframer-nba-e2e-4d76fe0b7a/ .