Skip to content
This repository was archived by the owner on Aug 5, 2026. It is now read-only.

Data: Check for empty tracks map during data gathering or rebroadcasts fail on Windows #39

Open
danrossi wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Data: Check for empty tracks map during data gathering or rebroadcasts fail on Windows #39

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 8 additions & 0 deletions src/moq-output.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -132,6 +132,9 @@ void MoQOutput::Stop(bool signal)
}
audio_tracks.clear();

// Clear bytes sent
total_bytes_sent = 0;

if (signal) {
obs_output_signal_stop(output, OBS_OUTPUT_SUCCESS);
}
Expand All @@ -147,6 +150,11 @@ void MoQOutput::Data(struct encoder_packet *packet)
return;
}

// OBS tries to gather data on signalling a stop success. Check once if the video tracks map is filled or new broadcasts fail on Windows.
if (video_tracks.empty()) {
return;
}
Comment on lines +153 to +156

@coderabbitai coderabbitai Bot May 18, 2026 •
edited
Loading

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🔴 Critical | ⚡ Quick win

Guard drops the first video packet and prevents track initialization forever.

This return runs before VideoData(), so video_tracks never gets initialized and all packets are skipped indefinitely.

Suggested fix 
-	// OBS tries to gather data on signalling a stop success. Check once if the video tracks map is filled or new broadcasts fail on Windows.
-	if (video_tracks.empty()) {
+	// Ignore non-video packets until at least one video track is initialized.
+	// This still allows the first video packet through to create the track.
+	if (video_tracks.empty() && packet->type != OBS_ENCODER_VIDEO) {
 		return;
 	}
🤖 Prompt for AI Agents 
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/moq-output.cpp` around lines 153 - 156, The early return when
video_tracks.empty() causes the first video packet (and thus track
initialization in VideoData()) to be dropped and prevents any future
initialization; instead of returning from that block, change the logic to avoid
discarding the first packets—either remove the return and skip only
shutdown/failure paths, or replace the return with a no-op/continue and only
perform the guard after VideoData() runs (or after a dedicated initialized flag
is set). Locate the guard referencing video_tracks (and the VideoData()
consumer) and adjust it so the check does not run before track initialization
completes, ensuring VideoData() can populate video_tracks on the first packet.

@kixelated kixelated May 22, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't understand what this is doing? VideoData() should return if no track is found.

@coderabbitai coderabbitai Bot May 22, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@kixelated — Great question, let me clarify the conflict.

VideoData() does guard against a missing track — but it also creates the track when it's missing:

auto it = video_tracks.find(encoder);
if (it == video_tracks.end()) {
    VideoInit(encoder);   // ← populates video_tracks[encoder]
    it = video_tracks.find(encoder);
}
if (it == video_tracks.end() || it->second < 0)
    return;  // ← only returns if init actually failed

This means video_tracks starts empty and is only populated on the first video packet flowing through VideoData() → VideoInit().

The guard added in this PR fires before that can happen:

Data() called with 1st video packet
  → video_tracks.empty() == true  (nothing initialized yet!)
  → early return ← VideoData() / VideoInit() never called
  → video_tracks stays empty forever
  → all subsequent packets also dropped

The video_tracks.empty() check can't distinguish between two very different states:

State video_tracks.empty()
Stream just started, tracks not yet initialized true
Stop() was called and cleared the map true

The PR wants to catch the second case (post-Stop() stray packets on Windows), but the guard also blocks the first case and breaks normal startup. A dedicated boolean flag (e.g. bool stopped) set in Stop() and cleared in Start() would be a cleaner way to distinguish the two situations.


if (packet->type == OBS_ENCODER_AUDIO) {
AudioData(packet);
} else if (packet->type == OBS_ENCODER_VIDEO) {
Expand Down
Loading