Route MQTT updates to local antenna display names (#2880). Principle I.

[aethersdr-agent]

Summary

Fixes #2880

What was changed

Route MQTT updates to local antenna display names (#2880). Principle I.

Files modified

• CMakeLists.txt
• src/core/MqttAntennaAliasParser.cpp
• src/core/MqttAntennaAliasParser.h
• src/gui/MainWindow.cpp
• src/gui/MqttApplet.cpp
• src/gui/MqttApplet.h
• tests/mqtt_antenna_alias_test.cpp

 src/gui/MainWindow.cpp              |  22 ++++
 src/gui/MqttApplet.cpp              |  10 ++
 src/gui/MqttApplet.h                |   2 +
 tests/mqtt_antenna_alias_test.cpp   | 252 ++++++++++++++++++++++++++++++++++++
 7 files changed, 381 insertions(+)

---

Generated by AetherClaude (automated agent for AetherSDR)

[aethersdr-agent]

Reviewed the diff against the surrounding code. Looks good — focused, well-bounded, and the test coverage is genuinely strong for this kind of integration.

What I like

• Pure parser in its own TU. MqttAntennaAliasParser takes only QString/QByteArray, links no Qt UI types, and the test drives it directly. That's the right factoring for something whose "correctness" is really a topic-contract spec.
• Defense in depth at the MainWindow seam. Gate on isConnected() and knownAntennaTokens().contains(token). Even during the brief window where connectionStateChanged(true) has fired but m_chassisSerial/antenna list haven't populated, known will be empty so the alias gets dropped rather than persisted under the "unconnected" fallback key from antennaAliasRadioKey() (RadioModel.cpp:541). Nice.
• Fixed contract topics held outside MqttTopics. Keeping aethersdr/antenna/names and aethersdr/antenna/name/+ subscribed independently of the user-editable topic list (MainWindow.cpp:1710-1714) means an operator can't accidentally unsubscribe the automation contract by editing the applet's topic settings.
• Bulk-JSON semantics for clear are explicit. Both JSON null and "" collapse to "empty alias → clear", and the comment in MqttAntennaAliasParser.h documents it. The tests pin both forms.

One small thing worth confirming (not a blocker)

MqttApplet::onMessageReceived still writes every incoming message — including the contract topics — to the visible m_messageLog (lines 266-275). For an operator who's set up bulk JSON re-publishes from Antenna Genius this could be a steady trickle into the message log. Probably fine, but if you wanted to suppress the contract topics from the human-readable log specifically, the early-return point would be right before the m_messageLog->append(line) for any topic that matches kNamesTopic or kNamePrefix. Easy follow-up if anyone complains.

Thanks for the careful work on this one — the parser/transport split and the test that exercises the disconnect-gate via a StubRadio are both above the typical bar.

[ten9876]

Closing in favor of #2881 by @s53zo, which was opened ~1.5 hours
before this PR and addresses the same issue (#2880).

Both PRs implement the MQTT → antenna-display-name routing, but
@s53zo's:

• Adds operator-facing documentation (resources/help/aethersdr-help.md)
  explaining the topic contract
• Was tested against a real MQTT broker (10.0.10.9:4883) with a
  live retained-message round-trip on the author's station
• Already passed full review (conflict-resolved against current main,
  25/25 parser tests green, build clean)

This PR is a process miss on the orchestrator's part — the
`aetherclaude-eligible` gate should have included a "check for
existing open PR addressing this issue" pre-flight, and didn't.
Filing a follow-up to add that check.

Thanks to the bot for the parallel implementation — the architectural
shape was similar, which is a good signal that #2880's design space
converges naturally. But community-contributor PRs take precedence
when both exist.

73,
Jeremy KK7GWY & Claude (AI dev partner)