Repository navigation
Invented or hard-coded values shown as measurements #765
Description
Activity
Timer slice implemented in PR #792 (0246509). Confirmed current main still advanced bridge elapsed time by 16 ms per poll plus 500 ms per capture spine tick, and TransportViewModel used that diagnostic timestamp for LIVE. The repair uses monotonic bridge-session time and an independent output-session clock starting at first observed local recording/stream production. Connecting alone does not start it; interruptions/retries/finalization remain in the session, and all outputs ending resets it. Explicit test timestamps no longer contaminate later requests. Eight clock regressions passed, including fake-core IPC, cadence independence, retries and combined outputs; Release WinUI build passed with existing warnings. Installed acceptance remains open. Remaining network/audio/caption claims are still present; this is a focused timer slice, not closure of #765.
- added a commit that references this issue
on Oct 6, 2026
Symptom / cause (read, not run).
talking ? 78 : 12(ZoomEngineRuntime.cpp:1142-1143,ZoomEngineState.cpp:365-366); the engine'sonUserNetworkStatusChangedis an empty override._elapsedMs, advanced 16 ms per 250 ms poll and 500 ms per spine tick (TransportViewModel.cs:226, 252;MediaCoreBridgeService.cs:393, 779): about 6% of real time with capture off, 106% with it on.max(82, 97 - len/28)(MediaCore.cpp:3743, 5240-5258); the merger hard-codes 95.rmsDbfs/peakDbfsare derived from a 0-100 level plus a fixed crest factor, not measured (MediaCore.cpp:4557-4567).Done when. Each value is measured, or removed from the wire and the UI.
Found by the 2026-10-02 half-wired-feature audit, by reading code on main (8b84a8f). Not yet reproduced in the running app unless stated.