Steer TX to A2DP when Default output shares a BT link with SCO

[patrickrb]

Closes #759

What changed

Follow-up to PR #772. That change made the RX side of Bluetooth work reliably on Android 8.1 (the SCO link is now brought up and the mic re-routed to it), and a tester confirmed decoding worked -- but reported a mirror bug on the TX side:

If I set audio input and output to "Default", audio received, but not transmitted to the transceiver. However, if I leave the audio input as "Default" and select a device with the A2DP profile for the audio output, audio is transmitted and received correctly.

Root cause. On Android 8.1, once the SCO (hands-free) link is up for the mic, the OS keeps the USAGE_MEDIA stream on the SCO speaker path too. The paired transceiver only listens for the FT8 tone on its A2DP music channel, so TX audio never leaves the phone -- exactly the "audio received, not transmitted" report. Manually picking the paired device's A2DP profile as the output moves TX back to A2DP and cures it.

Fix.

• AudioOutputRoutingPolicy (pure ints/Strings, unit-tested): when the user chose "Default" output, our own SCO link for the rig was up when the over was keyed, and the output list has an A2DP endpoint on the same Bluetooth device as that link, steer TX to it. It leaves routing to the OS for a rig on USB/network with a paired car kit nearby, for an A2DP endpoint on a different device, and whenever two hands-free devices are enumerated and it cannot tell which carries our link.
• TxScoLatch: the keying path (MainViewModel.beginKeying) stops SCO before PTT, and the TX worker only asks after the PTT settle delay -- by then the tracker already says "down". So keying snapshots the link state (and the address of the device the mic is captured from over SCO) into the latch before stopSco(), and the TX path reads the snapshot. The latch engages for any Bluetooth-rig TX (needControlSco(), so VOX included) and stops SCO only on the control-path keying that always did.
• DefaultOutputRouting: the Android adapter from the enumerated outputs to the policy to AudioTrack.setPreferredDevice, with the API < 28 / denied-BLUETOOTH_CONNECT address gating. Both TX playback paths in FT8TransmitSignal (playViaAudioTrack() and playTuneTone()) call it on the Default-output branch.

debug.log says what happened on each transmit: playFT8Signal: default output steered to A2DP (app SCO session up) when the route was applied, or playFT8Signal: A2DP steering REJECTED by setPreferredDevice; TX audio stays on the OS route when the framework refused it. No line at all on that branch means the policy left routing to the OS.

Also fixed on the way: ScoLinkCoordinator (merged in #772) took a java.util.function.LongSupplier clock, which does not exist before API 24 and would have failed to load on Android 6 (minSdk 23) while constructing MainViewModel; it now uses an app-local Clock interface, as the latch does.

Files changed:

• ft8af/app/src/main/java/com/k1af/ft8af/bluetooth/AudioOutputRoutingPolicy.java (new)
• ft8af/app/src/main/java/com/k1af/ft8af/bluetooth/TxScoLatch.java (new)
• ft8af/app/src/main/java/com/k1af/ft8af/bluetooth/DefaultOutputRouting.java (new)
• ft8af/app/src/main/java/com/k1af/ft8af/bluetooth/ScoLinkCoordinator.java (API 23 clock)
• ft8af/app/src/main/java/com/k1af/ft8af/MainViewModel.java (keying snapshot, TX accessors)
• ft8af/app/src/main/java/com/k1af/ft8af/wave/MicRecorder.java (routed SCO address)
• ft8af/app/src/main/java/com/k1af/ft8af/ft8transmit/FT8TransmitSignal.java (both TX playback paths)
• tests: AudioOutputRoutingPolicyTest (27 cases), TxScoLatchTest (6, drives the real ScoLinkTracker through the snapshot-then-stop order), DefaultOutputRoutingTest (7, Robolectric, real AudioDeviceInfo objects and a capturing sink)

Caveat

Like #772 this is verified with unit tests, not on an Android 8.1 device. If the tester still sees TX not reaching the rig, a fresh debug.log will now say either default output steered to A2DP (app SCO session up) (the override fired -- something further downstream is broken), A2DP steering REJECTED by setPreferredDevice (the framework refused the route), or nothing at all on that branch (the policy left routing to the OS -- most likely only one of A2DP/SCO enumerated, or our SCO link was not up at keying).

How to test

• Unit tests: from ft8af/ run ./gradlew testDebugUnitTest --tests 'com.k1af.ft8af.bluetooth.*' (JDK 17). Windows: cmd.exe /c "gradlew.bat testDebugUnitTest --tests com.k1af.ft8af.bluetooth.*".
• On-device (Android 8.1 with a BT-audio transceiver, Default I/O selected): confirm TX now reaches the rig and debug.log contains playFT8Signal: default output steered to A2DP (app SCO session up) on each transmit; RX behaviour from Bluetooth SCO: track link state, retry, and rebuild the mic on connect (Android 8.x RX dead, #759) #772 is unchanged.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 42.40%. Comparing base (6ab072f) to head (343d137).
⚠️ Report is 18 commits behind head on dev.

Additional details and impacted files

@@             Coverage Diff              @@
##                dev     #790      +/-   ##
============================================
+ Coverage     42.34%   42.40%   +0.06%     
- Complexity      226      227       +1     
============================================
  Files           267      267              
  Lines         31834    31862      +28     
  Branches       3650     3649       -1     
============================================
+ Hits          13479    13510      +31     
+ Misses        18097    18087      -10     
- Partials        258      265       +7     

Flag	Coverage Δ
android	17.05% <ø> (+0.15%)	⬆️
native	9.93% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.
see 3 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

🟡 Changes recommended

Device types alone cannot establish an active SCO link or matching A2DP device, risking transmission to an unrelated Bluetooth sink.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Steers default TX audio toward Bluetooth A2DP when SCO is detected, addressing Android 8.1 transmission routing.

Changes:

• Adds a pure-Java output-routing policy.
• Applies routing to FT8 and tune playback.
• Adds eight policy tests.

File summaries

File	Description
AudioOutputRoutingPolicy.java	Selects A2DP when A2DP and SCO types coexist.
FT8TransmitSignal.java	Applies the override to both TX paths.
AudioOutputRoutingPolicyTest.java	Tests policy decisions and constants.

Review details

• Files reviewed: 3/3 changed files
• Comments generated: 2
• Review effort level: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

🟡 Changes recommended

The SCO state is normally cleared before playback checks it, preventing the routing override from activating.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

• Files reviewed: 4/4 changed files
• Comments generated: 1
• Review effort level: Balanced

[Copilot]

🟡 Changes recommended

Routing can still select an unrelated Bluetooth endpoint and incorrectly affects USB rigs using a Bluetooth microphone.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

• Files reviewed: 6/6 changed files
• Comments generated: 3
• Review effort level: Balanced

[Copilot]

🟡 Changes recommended

Critical API 27 compatibility failures and unresolved Bluetooth routing cases block approval.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

Suppressed comments (1)

ft8af/app/src/main/java/com/k1af/ft8af/wave/MicRecorder.java:314

• The pure policy tests never execute this new Android-facing address lookup, so they cannot catch platform API-boundary failures such as the API-27 crash above. Add a Robolectric test for the new extraction path on SDK 27 (unknown address/fallback) and SDK 28+ (SCO address returned), including non-SCO and blank-address cases.

    public synchronized String routedScoInputAddress() {

• Files reviewed: 7/7 changed files
• Comments generated: 4
• Review effort level: Balanced

[Copilot]

🟡 Changes recommended

The API 23 compatibility failure and unresolved routing, permission-handling, and test-coverage issues must be addressed.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

Suppressed comments (3)

Previously missed (1) — in code that hasn't changed since the last review.

ft8af/app/src/main/java/com/k1af/ft8af/ft8transmit/FT8TransmitSignal.java:3346

• The pure policy and latch are tested, but no test executes this new Android routing adapter, so the wiring from enumerated outputs and the TX latch into setPreferredDevice() (including the rejection diagnostic) can regress unnoticed. The project requires every new helper/branch to have unit coverage; please extract injectable device/route collaborators or add a focused Robolectric test for this helper.

ft8af/app/src/main/java/com/k1af/ft8af/bluetooth/AudioOutputRoutingPolicy.java:170

• When activeScoAddress is known but its A2DP endpoint is not found, this fallback can discard contradictory known output identity. For [A2DP(blank), SCO(OTHER)] with the active SCO address RIG, it selects the blank A2DP endpoint even though the only enumerated SCO endpoint is another device, potentially sending TX to that device. Preserve the all-addresses-withheld fallback, but do not use it when a known active address conflicts with any known output SCO address; add this partial-redaction case to the policy tests.

        boolean addressesUninformative =
                !isKnownAddress(addressAt(deviceAddresses, firstA2dpIdx)) || !anyScoAddressKnown;
        if (a2dpCount == 1 && scoCount == 1 && addressesUninformative) {
            return firstA2dpIdx;

ft8af/app/src/main/java/com/k1af/ft8af/ft8transmit/FT8TransmitSignal.java:3356

• The PR's caveat and on-device test instruct the tester to search for default output steered to A2DP (SCO link present), but this success message is now ...(app SCO session up) and rejection has its own message. Following the documented check will therefore falsely indicate that the override did not fire; update the PR description and its test expectations to the emitted strings (and refresh the stale file/test counts).

        GeneralVariables.fileLog(applied
                ? "playFT8Signal: default output steered to A2DP (app SCO session up)"
                : "playFT8Signal: A2DP steering REJECTED by setPreferredDevice;"
                        + " TX audio stays on the OS route");

• Files reviewed: 7/7 changed files
• Comments generated: 2
• Review effort level: Balanced

[patrickrb]

Also addressed the three suppressed comments from the latest review in 343d137:

• Untested Android adapter — the enumeration → policy → setPreferredDevice wiring is now DefaultOutputRouting.apply(outputs, scoHeld, scoAddress, Sink), and DefaultOutputRoutingTest (Robolectric, real AudioDeviceInfo objects via AudioDeviceInfoBuilder, a capturing sink) covers the steer + LOG_STEERED, the LOG_REJECTED path, the no-SCO / no-Bluetooth / null no-ops, and the API 27 address gate. FT8TransmitSignal only supplies the real track and fileLog.
• Withheld A2DP address contradicting a named SCO — [A2DP(blank), SCO(OTHER)] with our link on RIG now leaves routing to the OS; [A2DP(blank), SCO(RIG)] still uses the single-pair fallback. Two new policy tests (27 total).
• Stale PR description — rewritten: the emitted log strings, the latch, the adapter, the coordinator clock fix, and the current file/test lists.

[Copilot]

🔵 Needs a closer look

Bluetooth routing behavior remains unverified on the target Android 8.1 device.

Review details

• Files reviewed: 10/10 changed files
• Comments generated: 0 new
• Review effort level: Balanced