)]}'
{
  "commit": "8bca63f27840903ce5b62f7edb18ed2f491194ce",
  "tree": "1158017fe631d46e6ffe46af39d6d1514c9633b4",
  "parents": [
    "98708d25e1d3deb81e3fbf6d9f3ed67b2dea948c"
  ],
  "author": {
    "name": "Tommi",
    "email": "tommi@webrtc.org",
    "time": "Fri May 22 06:07:00 2026"
  },
  "committer": {
    "name": "webrtc-scoped@luci-project-accounts.iam.gserviceaccount.com",
    "email": "webrtc-scoped@luci-project-accounts.iam.gserviceaccount.com",
    "time": "Fri May 22 08:10:11 2026"
  },
  "message": "Unify worker and network threads in full stack tests\n\nUnify worker and network threads by default in `webrtc_perf_tests` to eliminate the startup packet-loss race condition introduced in CL 472240.\n\nBy making worker and network threads refer to the same thread object, the receiver sink registration (`AddSink`) runs synchronously on the same thread. This avoids the race window where incoming packets fail demuxing on the network thread before the asynchronous `AddSink` task is processed.\n\nPerformance verification results (comparing separate vs unified threads):\n\n- Pc_Simulcast_HD_High:\n  * dropped_frames: 63 -\u003e 34 (-46% improvement)\n  * max_skipped: 0.048 -\u003e 0.025 (-47% improvement)\n  * actual_encode_bitrate: 2011.686 kbps -\u003e 2071.463 kbps (+3% recovery)\n  * retransmission_bitrate: 0.134 kbps -\u003e 0.000 kbps (-100% fully eliminated)\n\n- Pc_Screenshare_Slides_Simulcast_No_Conference_Mode:\n  * dropped_frames: 2 -\u003e 1 (-50% improvement)\n  * max_skipped: 0.045 -\u003e 0.022 (-50% improvement)\n  * actual_encode_bitrate: 240.117 kbps -\u003e 352.958 kbps (+47% recovery)\n  * transmission_bitrate: 371.318 kbps -\u003e 520.972 kbps (+40% recovery)\n\n- Pc_Simulcast_Vp8_3sl_High:\n  * dropped_frames: 39 -\u003e 27 (-30% improvement)\n  * max_skipped: 0.029 -\u003e 0.020 (-31% improvement)\n\nBug: webrtc:514760674\nChange-Id: Ie6ad4542ac30a365af464854df383f0f14a03ed2\nReviewed-on: https://webrtc-review.googlesource.com/c/src/+/474120\nCommit-Queue: Tomas Gunnarsson \u003ctommi@webrtc.org\u003e\nReviewed-by: Per Kjellander \u003cperkj@webrtc.org\u003e\nReviewed-by: Markus Handell \u003chandellm@webrtc.org\u003e\nReviewed-by: Rasmus Brandt \u003cbrandtr@webrtc.org\u003e\nCr-Commit-Position: refs/heads/main@{#47772}\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "1d039d702c483c56911f0d2e3e328c06ff1966e8",
      "old_mode": 33188,
      "old_path": "video/pc_full_stack_tests.cc",
      "new_id": "87959b8d215bc48e40dabc030b48266bc37b14b6",
      "new_mode": 33188,
      "new_path": "video/pc_full_stack_tests.cc"
    }
  ]
}