# Issue 29 probes

These snapshots reproduce the EQ retention and video emission-rate differences; `before/VideoEqualizerProcessor.swift` adds only `retainedConfigurationCount` for measurement. `Models.swift` stubs unrelated localization/settings keys and declares the same response rates as the app. `EqualizerSettings.swift` is the unchanged production model.

From this directory, with Xcode 27.1 selected:

```sh
xcrun swiftc -swift-version 6 -O -parse-as-library -D BASELINE Models.swift EqualizerSettings.swift Probe.swift before/*.swift -o /tmp/issue29-before
xcrun swiftc -swift-version 6 -O -parse-as-library Models.swift EqualizerSettings.swift Probe.swift after/*.swift -o /tmp/issue29-after
/tmp/issue29-before
/tmp/issue29-after
xcrun swiftc -swift-version 6 -sanitize=thread -g -parse-as-library Models.swift EqualizerSettings.swift RaceProbe.swift after/*.swift -o /tmp/issue29-race
/tmp/issue29-race
```

The before snapshot is based on main `6ef0b74`. The fixed source snapshots accompany the code PR linked from the Wiki page. Source-audio frame counts and retained configuration counts are the measured values; physical-device dropouts and Instruments render-thread traces are unmeasured.

Review follow-up: the archived `after/` snapshots correspond to the initial fix `969f878`. The follow-up fixes and regression tests are committed at `886f1eb` in the main application repository. `review-baseline-results.log` preserves one failure per newly added regression test, and `review-final-results.log` records the 30-test pass after the fixes. `review-race-results.log` records the stress-probe rerun using production service sources from `886f1eb`; no Thread Sanitizer race was reported.