The CDI softkey was in the PFD root row but had no handler — it never cycled the
HSI/CDI source (GPS↔VLOC1↔VLOC2). Wire it to write cdiSrc
(HSI_source_select_pilot, now writable), cycling GPS→NAV1→NAV2.
Also fix a latent demo bug: the 'needs a live sim socket' guard sat above the
setDataref/command handlers, so in DEMO (no sim socket) every dataref write
(transponder code, baro, AP bugs, CDI) was silently dropped. Reflect writes
locally before that guard so cockpit controls respond without a sim.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The steam-panel KAP140 LCD decoded the same unreliable autopilot_state bitfield
as the G1000 panel, so its lateral/vertical annunciation could be wrong. Read the
*_status datarefs instead, consistent with AutopilotPanel and the PFD.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Key feedback came from decoding the autopilot_state bitfield, whose bit
positions don't reliably match X-Plane (e.g. bit 0 is auto-throttle, not the
flight director), so several keys (FD/NAV/APR/BC/VS/VNV/FLC) never lit even when
the mode was engaged. Read the dedicated *_status datarefs (off/armed/active) —
the same source the PFD AFCS bar uses — so every key and the lateral/vertical
annunciator reflect the real mode. NAV also lights on GPS-coupled (gpss) mode.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The 3D terrain showed almost no sky — the horizon sat far above the attitude
horizon line. Base camera pitch was 72°, but with MapLibre's 36.87° vertical FOV
the flat horizon only reaches the attitude line (28% of the SVT box) at ~82°.
Invert the perspective to derive the camera pitch from aircraft pitch so the
synthetic horizon lands exactly on the attitude horizon and tracks it 1:1
(accounting for the 1.5× canvas scale). Raise maxPitch to 85.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
So a per-machine release (macOS here, Linux on Arch) of the same version keeps
both platform entries instead of clobbering the one not built this run.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
linuxdeploy injects a RUNPATH ($ORIGIN/../lib) via patchelf into every
usr/bin executable when building the AppDir. The Bun-compiled xpbridge
sidecar (self-contained, JS/assets appended past the ELF) does not
survive ELF rewriting — the patched copy core-dumps on start, so the app
launches but the bridge never listens.
Add scripts/fix-linux-appimage.sh: extract the built AppImage, restore the
pristine repo sidecar, repack with linuxdeploy-plugin-appimage (which does
not patchelf), verify the sidecar is byte-identical, and regenerate the
updater .sig. Wired into scripts/build-linux.sh after `tauri build`.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
FlyWithLua auto-install: bridge drops fms-sync/ui-sync/terrain-probe into
X-Plane's FlyWithLua Scripts dir on startup and self-updates (content-compare).
Graceful when no X-Plane / no FlyWithLua. /api/lua/install + status in health.
Desktop app bundles the scripts and passes LUA_SRC_DIR to the sidecar.
Smoothing: shared useEased/useEasedAngle hook (api/ease.js) with render-bail on
settle. VFR steam gauges now interpolate to 60fps instead of stepping at the
~10Hz value stream. MFD ownship no longer vibrates — position/heading eased in a
single rAF loop, follow-pan without animated-panTo pile-up (pauses on range zoom).
Airspace overlay: server/airspace.js loads per-region GeoJSON, classifies
(B/C/D/TMA/CTR/MOA/Restricted/Prohibited/Danger), bbox query, and downloads
regions on demand — FAA (US, key-free) and OpenAIP (Europe, user key). New
AIRSPACE softkey draws chart-coloured boundaries (B blue, C magenta, D dashed),
non-interactive so map-clicks still drop waypoints. Launcher gains a "Lufträume"
section to pick/download regions via the running bridge.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- electrical readout now shows M (main) and E (essential) bus volts from
bus_volts[0]/[1], M (alternator generator_amps) and S (battery_amps) separately,
and ENG hours from flight time — replacing the hardcoded duplicate volts / +0.0
amps / 0.0 HRS placeholders (manual S.54 C172 layout)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- PFD VNAV: magenta flight-plan target altitude on the alt scale (S.110), V DEV
deviation scale + chevron (left, shown in VNAV when not on an ILS), VS TGT
chevron on the VSI (S.113)
- GPS phase annunciation is now dynamic: APR (approach leg) / TERM (<30 nm to
destination) / ENR, instead of a fixed label
- flight-plan ALT can be toggled designated(blue) <-> reference(white) by clicking
the cell (S.106); only designated altitudes drive the VNAV profile
- setPlan now preserves the dsgn/appr waypoint flags across the shared plan
- AFCS vertical mode labelled VPTH (manual) instead of VNV
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- CURRENT VNV PROFILE panel on the MFD flight-plan page: active VNV waypoint +
target altitude, VS TGT (−3° path), VS REQ, V DEV, FPA, TIME TO TOD (manual
S.64 / S.107)
- enriched the VNAV computation (vsTgt / vDev / FPA / time-to-TOD) shared by the
PFD VnavBox
- flight-plan ALT column now shows designated (VNAV) altitudes in blue (S.105)
- new Audio Panel + earlier manual-alignment batch already in this branch
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Aligned to the official X-Plane 1000 manual:
- NAV radio: active RIGHT / standby LEFT (boxed) per S.12 (COM already correct)
- ALT UNIT softkey (IN / HPA) in the PFD submenu, baro readout converts (S.20)
- DCLTR cycles 3 levels (land / +NDB / flight-plan only) with DCLTR-n label (S.56)
- TOPO and TERRAIN are now independent toggles (relief vs awareness overlay) (S.57)
- Barometric MINIMUMS: BARO MIN bug + readout on the altimeter, amber "MINIMUMS"
annunciation at/below the decision altitude; set via TMR/REF (lifted to App)
- OBS mode: HSI course follows the CRS knob (magenta "OBS"), sequencing suspended
- New Audio Panel tab (COM mic/receive, MKR/DME/ADF, intercom, Display Backup) (S.91)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
karim