MinGW / no-HIDAPI build broken: unguarded TMate2Overlay calls in applyFlexControlWheelAction

[NF0T]

What happened?

PR #3401 (ELAD/WoodBoxRadio TMate 2 USB HID support) added calls to
triggerTMate2Overlay(TMate2Overlay::...) inside applyFlexControlWheelAction.
applyFlexControlWheelAction is declared outside any #ifdef HAVE_HIDAPI block —
it handles wheel inputs from FlexControl, MIDI, and other non-HID sources — but
both TMate2Overlay (an enum class) and triggerTMate2Overlay are declared
inside the #ifdef HAVE_HIDAPI block in MainWindow.h (lines 573–623). The
five unguarded call sites cause a build failure on any configuration where
HIDAPI is not present.

Compiler output (MinGW GCC 13.1.0, Qt 6.11.0, no hidapi):

src/gui/MainWindow.cpp:7823:34: error: 'TMate2Overlay' has not been declared
src/gui/MainWindow.cpp:7823:13: error: 'triggerTMate2Overlay' was not declared in this scope
src/gui/MainWindow.cpp:7840:30: error: 'TMate2Overlay' has not been declared
... (5 sites total)

This does not reproduce on CI because the Docker image and MSVC Windows path
both include HIDAPI. It only surfaces locally on Windows without
scripts/setup/setup-hidapi.ps1 having been run, or on Linux/macOS builds
where the HID optional dependency was not installed.

Steps to reproduce

1. Clone the repo on Windows without running setup-hidapi.ps1
2. Configure and build normally
3. Observe compile failure in MainWindow.cpp

Expected behavior

Build succeeds on all target platforms regardless of HIDAPI availability,
matching the pattern established in PR #3344.

Affected lines

src/gui/MainWindow.cpp — five call sites inside applyFlexControlWheelAction:
WheelRit (×1), WheelVolume (×2), WheelPower (×1), WheelCwSpeed (×1).

Fix

Wrap each of the five sites with #ifdef HAVE_HIDAPI / #endif.
Branch fix/tmate2-wheel-mingw-no-hidapi on NF0T/AetherSDR has the patch
ready for PR.

[aethersdr-agent]

Triage

Confirmed. The report is precise and the fix is mechanical.

Root cause

TMate2Overlay (enum class) and triggerTMate2Overlay() are declared inside an #ifdef HAVE_HIDAPI block in src/gui/MainWindow.h:

• src/gui/MainWindow.h:590 — enum class TMate2Overlay { ... };
• src/gui/MainWindow.h:608 — void triggerTMate2Overlay(TMate2Overlay overlay, int value);
• Block opens at line 573 and closes with #endif at line 618.

MainWindow::applyFlexControlWheelAction() lives at src/gui/MainWindow.cpp:7785 outside any HIDAPI guard — correctly, because it also serves MIDI and other non-HID wheel sources (see the MIDI/FlexControl/dial dispatch sites around lines 6956–6980, 7117, etc.). The five call sites inside it reference the HIDAPI-guarded symbols unconditionally:

Line	Branch	Call
7824	WheelRit	triggerTMate2Overlay(TMate2Overlay::Rit, hz);
7841	WheelVolume / WheelMasterAf	triggerTMate2Overlay(TMate2Overlay::Volume, next);
7847	WheelHeadphoneVolume	triggerTMate2Overlay(TMate2Overlay::Volume, next);
7870	WheelPower	triggerTMate2Overlay(TMate2Overlay::Power, next);
7875	WheelCwSpeed	triggerTMate2Overlay(TMate2Overlay::Wpm, next);

(The reporter's "WheelVolume ×2" count is the WheelVolume + WheelHeadphoneVolume pair — five sites total, matching the compiler output.)

CI is green because both the Docker/Linux image and the MSVC Windows path pull HIDAPI in. Local MinGW or any Linux/macOS dev box without the optional HID dep hits this immediately.

Proposed fix

Wrap each of the five call sites with #ifdef HAVE_HIDAPI / #endif. The surrounding wheel logic (RIT/volume/power/CW speed) must stay unguarded so MIDI and other transports keep working. This matches the pattern from #3344.

Single file touched: src/gui/MainWindow.cpp.

Handoff summary

• What's known: clear build break on every non-HIDAPI configuration; reporter has the patch ready on fix/tmate2-wheel-mingw-no-hidapi.
• What to fix: five #ifdef HAVE_HIDAPI wraps around the triggerTMate2Overlay(...) calls at lines 7824, 7841, 7847, 7870, 7875.
• Why it's safe: all five sites are pure overlay-UI hints with no functional side effects on the wheel action itself; guarding them preserves the wheel behavior for every transport.
• Scope: small, mechanical, no protocol or behavior change — flagged good first issue if you'd rather hand it off.

Ready for aetherclaude-eligible if you'd like me to take it.

---

_{🤖 aethersdr-agent · cost: $2.2256 · model: claude-opus-4-7}