Review — PR #2178

Thanks for working on the DSP filter visibility gating for 6000-series radios. The core feature (hiding NRL/NRS/RNN/NRF on radios that lack them) looks correct and well-placed. However, this PR has significant scope issues that need to be addressed before merge.

Scope concern — SliceColorManager removal reverts #2155

The stated scope is "Hide 8000-series DSP filters on 6000-series radios," but the PR also:

• Deletes SliceColorManager entirely (.h + .cpp, ~193 lines)
• Removes the "Themes" tab from RadioSetupDialog (~161 lines of UI code)
• Reverts all consumers (CatControlApplet, RxApplet, SpectrumWidget, VfoWidget, main.cpp) from SliceColorManager::instance() back to raw kSliceColors[] access
• Removes the colorsChanged signal connections from VfoWidget and SpectrumWidget

This effectively reverts the entire Customizable Slice Colors feature from PR #2155 (merged 2 commits ago). That's a user-facing feature regression that isn't mentioned in the PR title, body, or issue #2177. These changes should either be split into a separate PR with its own justification, or removed from this PR entirely.

Potential crash — hardcoded DSP row indices without bounds check

In SpectrumOverlayMenu::setHasExtendedDsp():

hideRow(m_dspRows[4], has);
hideRow(m_dspRows[5], has);
hideRow(m_dspRows[6], has);
hideRow(m_dspRows[8], has);

The early return (if (m_dspRows.isEmpty()) return;) only guards against a completely empty vector. If m_dspRows is non-empty but has fewer than 9 elements (e.g., a build configuration that excludes some DSP backends), this is an out-of-bounds access and will crash. Suggest:

auto safeHide = [&](int idx, bool visible) {
    if (idx >= m_dspRows.size()) return;
    // ... hide logic
};

Minor: inconsistent bounds checking on kSliceColors[]

RxApplet.cpp correctly guards access:

const char* color = (i < kSliceColorCount) ? kSliceColors[i].hexActive : "#888888";

But CatControlApplet.cpp uses unguarded direct indexing:

.arg(kSliceColors[i].hexActive)
.arg(kSliceColors[ch].hexActive)

The loop variable i runs to kMaxRigctlChannels (currently 4, safely under kSliceColorCount = 8), so this isn't a bug today, but it's inconsistent and fragile. Worth adding the same bounds check for consistency.

DSP filter gating logic looks good

The actual feature implementation is clean:

• RadioModel::hasExtendedDspFilters() correctly identifies BigBend/DragonFire platforms via model string matching — follows the existing pattern used by maxPanadapters() and other model-detection helpers
• setHasExtendedDsp() is called at the right time in onSliceAdded() (before wireVfoWidget) and wirePanadapter()
• Both VfoWidget and SpectrumOverlayMenu correctly gate the 4 extended filters

Recommendation

Please split this into two PRs:

1. The DSP filter hiding (the actual fix for #2177) — this is ready
2. The SliceColorManager removal — needs its own discussion and justification, as it reverts a recently merged feature