feat(gui): add RADE status applet to sidebar panel

[NF0T]

Summary

Adds `RadeApplet`, a new sidebar applet that mirrors the RADE/FreeDV
status block from the VFO widget as a standalone, dockable panel in the
right-side AppletPanel.

The applet shows sync state (LED: green synced, grey unsynced), mode label
("RADE" or "FreeDV"), SNR value with color coding (yellow < 5 dB, green
≥ 5 dB), EOO-decoded far-end callsign, and carrier frequency offset. All
fields are driven directly by existing RADEEngine signals — no separate
timer or polling. Content is hidden and state is reset when RADE or FreeDV
deactivates.

The FreeDV waveform path (FDVU/FDVL via the radio's EXT_WVF engine) shows
sync and SNR from the `FreeDV_SNR` meter; it does not populate callsign or
frequency offset as those are RADE-specific.

Constitution principle honored

Principle IX — Surface Only What Survives: the applet exposes only data
already computed by the RADE engine and emitted via existing signals.
No new polling, no new timers, no new state derivation.

Test plan

• Local build passes (`cmake --build build`)
• Activate RADE on a DIGU slice: applet shows "RADE", LED goes green
  on sync, SNR populates, EOO callsign appears when decoded
• Deactivate RADE: applet resets to "RADE inactive"
• Activate classic FreeDV (FDVU/FDVL): applet shows "FreeDV" label,
  LED and SNR update from meter; no callsign or offset rows shown
• Build without `HAVE_RADE` (`-DENABLE_RADE=OFF`): applet absent,
  no compilation errors in AppletPanel or MainWindow FreeDV paths —
  verified clean on Windows (MinGW GCC 13.1.0, Qt 6.11.0)
• Existing tests pass (CI)

Checklist

• Commits are signed (`docs/COMMIT-SIGNING.md`)
• No new flat-key `AppSettings` calls — applet is display-only,
  no persistence needed
• No new meter UI — SNR is a plain `QLabel`, not a `MeterSlider`
• Documentation updated if user-visible behavior changed
• Security-sensitive changes reference a GHSA if applicable — N/A

[aethersdr-agent]

Thanks for the clean, self-contained applet @NF0T — this is a nice surface-only addition. The signal-driven design matches Principle IX, and the HAVE_RADE guards are correctly placed (FreeDV path is independent of the RADE engine, and the new code respects that).

A few small observations, none blocking:

1. Inline stylesheets duplicate the same QLabel rule many times

src/gui/RadeApplet.cpp repeats this string at least four times:

"QLabel { color: #8090a0; font-size: 10px;"
" background: transparent; border: none; padding: 0; margin: 0; }"

…and setRadeSnr / setRadeSynced both re-apply the full sheet just to change the color. Could you pull a single kLabelBaseStyle constant (or set the base style once in the ctor and only toggle the dynamic property / color via a smaller sheet)? It would also make the contrast with the two ThemeManager::applyStyleSheet callsites less jarring — right now the file uses both styles side-by-side and it's not obvious which labels are themable and which aren't.

2. setRadeActive(true) doesn't reset m_offsetLabel/m_callsignLabel contents

setRadeActive(false) clears them, so a normal activate-after-deactivate is fine. But the first activation (with both labels constructed but never cleared) and a hypothetical activate→activate-with-different-label sequence both leave whatever was last shown. Probably benign in practice — just worth a one-line m_offsetLabel->hide(); m_callsignLabel->clear(); m_callsignLabel->hide(); in the on==true branch to keep state machine purely a function of inputs.

3. static_cast<int>(snrDb) truncates toward zero

-2.7 becomes -2, 2.7 becomes 2. Not wrong, but qRound() would match the more conventional "nearest integer dB" display you see elsewhere in the meter UI.

4. Build without HAVE_RADE test plan item is unchecked

The guards in MainWindow.cpp, AppletPanel.{h,cpp}, and the file-level #ifdef HAVE_RADE in RadeApplet.{h,cpp} look correct to me, but since CMake registers src/gui/RadeApplet.cpp unconditionally (it's added outside any if(HAVE_RADE) block), the source file will compile to an empty TU when the flag is off — which works, but you'll want to actually verify it before merge given that's the box you haven't checked yet.

Nice clean PR overall.

---

_{🤖 aethersdr-agent · cost: $16.8321 · model: claude-opus-4-7}

[aethersdr-agent]

Hi @NF0T — thanks for this contribution! 👋

The Qt Accessibility Static Analysis check (run 27069409424) is the only thing red right now. It's the lint added recently in a8bafc73 (tools/check_a11y.py, wired up by .github/workflows/a11y-check.yml). The other CI jobs (build, check-macos, check-windows, analyze (cpp)) are still in-progress and unrelated.

What's tripping the check

The new RadeApplet is a QWidget subclass whose constructor creates several QLabel children (m_inactiveLabel, m_statusLabel, m_snrLabel, m_callsignLabel, m_offsetLabel, plus the "SNR" caption) but never calls setAccessibleName() on any of them. The lint's check #3 (constructor-missing-accessible-names) is designed to catch exactly this:

creates child widgets but calls setAccessibleName() zero times — AT users cannot identify controls.

Since the widget exposes live information (sync state, SNR, callsign, frequency offset), screen-reader users would have no way to identify what each label is reading out.

Two ways to fix

Option 1 — add accessible names (recommended): Give each informational label a name in RadeApplet::RadeApplet(). Something like:

m_inactiveLabel = new QLabel(tr("RADE inactive"));
m_inactiveLabel->setAccessibleName(tr("RADE status"));

m_statusLabel = new QLabel;
m_statusLabel->setAccessibleName(tr("RADE sync indicator"));

m_snrLabel = new QLabel(QStringLiteral("---"));
m_snrLabel->setAccessibleName(tr("RADE SNR"));

m_callsignLabel->setAccessibleName(tr("Far-end callsign"));
m_offsetLabel->setAccessibleName(tr("Frequency offset"));

You may also want to follow up with QAccessibleValueChangeEvent notifications in setRadeSnr / setRadeFreqOffset / setRadeCallsign so screen readers announce the values when they change — see docs/a11y.md for the canonical patterns.

Option 2 — opt out if you'd rather defer the a11y work to a follow-up: add // a11y-check: skip-file anywhere in RadeApplet.cpp. The lint is documented as informational, but since the readouts here are real data (not decoration), Option 1 is the better fit.

Heads up on a separate compile issue I noticed while looking

RadeApplet is declared inside namespace AetherSDR { ... } in the header, but in MainWindow.cpp it's referenced bare (RadeApplet*) and forward-declared bare (class RadeApplet; in AppletPanel.h). Once the build job catches up it will likely error on this — you'll probably need either a using AetherSDR::RadeApplet; in those translation units, or to drop the namespace (the rest of src/gui/ mostly doesn't namespace its widget classes). Worth checking in your local cmake --build build with -DHAVE_RADE=ON.

Appreciate the careful PR — the applet itself is a nice mirror of the VfoWidget status. Once the names are in, this should sail through. 🛰️

---

_{🤖 aethersdr-agent · cost: $10.0815 · model: claude-opus-4-7}