Complete managed macOS platform bring-up

docs/porting-to-dotnet.md

@@ -791,16 +791,17 @@ criterion from `roadmap.md`, re-proven against the new server.
 6. **Add the allocation-regression test.**
 
 **Exit criterion:** `test_jitter_depth`'s bounded-depth invariant holds; zero allocations per
-mix cycle; a manual listen test on Windows and macOS with no audible glitching over 10
-minutes.
+mix cycle; real multi-human calls on Windows and macOS have no reproducible audio-quality
+defects once each client is feature-complete.
 
 **Checkpoint (2026-09-16):** `VoiceCat.Audio` now owns local Opus streams, reframing at every
 protocol frame size, VAD/PTT/DTX, DRED → FEC → PLC receive recovery, bounded jitter, per-stream
 controls, RNNoise and stereo mixing. Its normal encode/decode/NR/mix cycle allocates zero
 managed bytes. Capture inputs use bounded non-waiting PCM rings. The Windows implementation
 uses direct C# WASAPI capture, loopback and playback instead of the proposed miniaudio device
-shim; the codec/DSP shim remains the only native component. A real device smoke passed, but
-the required ten-minute Windows/macOS listen test is still a manual release gate.
+shim; the codec/DSP shim remains the only native component. A real device smoke passed;
+final listen validation is performed with real multi-human calls after feature completion,
+not a synthetic ten-minute sine-wave gate.
 
 ---
 
@@ -850,7 +851,7 @@ Per §8.2. AppKit port, ScreenCaptureKit per-app audio selection, VoiceOver pari
 **Exit criterion:** feature parity with `VoiceCatMac`, VoiceOver smoke-tested, notarized
 build produced.
 
-**Checkpoint (2026-09-16):** `clients/apple/dotnet/VoiceCat.Mac` is a separate .NET 10
+**Checkpoint (2026-09-18):** `clients/apple/dotnet/VoiceCat.Mac` is a separate .NET 10
 AppKit application that consumes `VoiceCat.Core` directly. It implements guest and account
 connection, validated saved server profiles whose JSON never contains passwords,
 interactive TOFU approval, protected-channel password prompts, channel browsing, roster,
@@ -862,11 +863,17 @@ Core Audio capture is resampled
 and converted through `AVAudioConverter` before entering the managed audio engine. Stereo
 playback uses `AVAudioSourceNode` and the shared bounded PCM ring; its render callback does
 not allocate, lock or block. Voice stream lifetime follows subscription, disconnect and
-channel changes. Current Apple API calls compile warning-free against Microsoft's
-26.4.10259 macOS reference assembly, and a macOS CI job builds the native codec/DSP shim and
-Apple solution. A real macOS device run and listen test remain required. The existing Swift
-app remains the release client until selectable devices and the moderation/settings
-surface, ScreenCaptureKit, VoiceOver
+channel changes. The app now targets Microsoft's Xcode 27 preview bindings from workload
+set 10.0.401. On Apple Silicon it selects a native Homebrew `protoc`, deduplicates the
+transitive media dylib before AppKit bundling, uses the macOS 27 error-returning Core Audio
+overloads, and keeps hardened runtime Release-only so local ad-hoc Debug builds can load
+their runtime libraries. It enumerates and selects Core Audio input/output devices, requests
+microphone permission explicitly, and renders through Core Audio's native planar Float32
+layout. The arm64 app builds, signs, launches and displays on macOS 27. Live testing proved
+physical-microphone transmission and clean peer playback; Lobby's expected DTX comfort noise
+was distinguished from corruption by repeating the tone in the DTX-disabled Music Room. A
+macOS CI job builds the native codec/DSP shim and Apple solution. The existing Swift app
+remains the release client until the moderation/settings surface, ScreenCaptureKit, VoiceOver
 validation, signing and notarization are complete.
 
 ---