preston-peterson/aerodrome v3.4.31

on GitHub

The aircraft details map now draws each aircraft's track as a multi-color line instead of a cloud of per-point dots. Same data underneath — every position sample from the ADS-B history feed — but rendered as connected polyline segments rather than discrete circle markers. The line color encodes altitude on the same six-stop ramp the page's altitude legend uses, so a glance tells you both the path the aircraft took and what altitude band it was in at every point. Climb, cruise, and descent phases read directly from the color transitions; the path shape reads from the line itself instead of having to be assembled mentally from a dotted cloud. The altitude legend strip below the map is unchanged — it was already the legend for these colors, just now it's a legend for the line instead of for the dots.

• The line breaks honestly across coverage gaps. When the time between two consecutive position samples exceeds 120 seconds (the default ADS-B coverage-loss threshold for this rendering), the line stops at the last in-coverage point and resumes at the next one — it does not draw a connecting segment across the gap. Drawing across would imply a path the aircraft didn't actually fly. Two minutes between samples means real coverage loss in normal operation (most aircraft sample every 5-30 seconds when in range); the threshold is conservative enough to not break the line for momentary dropouts, strict enough to catch real terrain or over-horizon losses. If you have a single isolated position with no neighbors within the gap threshold on either side, it falls back to a small dot at its bin color so it remains visible.

• Altitude bin transitions use a 500 ft hysteresis to prevent flicker. ADS-B altitude readings can wobble ±100 ft even during steady cruise. Without hysteresis, an aircraft cruising at FL299 (29,900 ft) would flicker between the orange and red bins on every other sample, producing a visually noisy striped line that doesn't reflect actual altitude behavior. The fix: transitioning to a higher bin requires the altitude to cross the boundary plus 250 ft; transitioning to a lower bin requires crossing it minus 250 ft. Net effect: 500 ft of deadband at each of the five bin thresholds (1000 / 10000 / 25000 / 35000 / 45000 ft). Smooth color transitions where the aircraft actually changed altitude bands; no flicker at boundaries where it didn't.

Behind the scenes

• The new rendering replaces dotsLayer in _posMapState with tracksLayer, an L.layerGroup containing one or more L.polyline instances plus any orphan L.circleMarker fallbacks. The Leaflet canvas renderer (L.canvas({ padding: 0.5 })) handles polylines as efficiently as it handled circle markers, so the change is performance-neutral or slightly favorable: typical flight tracks produce 4-8 polyline segments versus N circle markers for N samples.
• The altToColor() function that mapped altitudes to colors for the dot renderer is removed; its replacement is the BIN_THRESHOLDS + BIN_COLORS arrays inside loadPositions(), which the new hysteresis-aware binning uses directly. The .map-altitude-legend gradient in the page CSS retains the same six color stops it has since v3.4.x — that's the visual contract between the rendering and the legend, and it didn't need to move.
• Each segment's color is determined by its ending point's altitude bin, so a sharp color transition lands at the point where the aircraft reached the new band. The polyline-builder closes the current run and starts a new one whenever the bin changes between consecutive points, sharing the transition vertex between the two polylines so the visual line stays continuous across the color change.

Operational notes

• No config migration; no API changes; no data-storage changes. The rendering swap is entirely client-side in templates/aircraft.html.
• The receiver marker, map theme toggle, and expand control are unchanged. The track is the only visual element that changed.
• For very dense histories (10k+ sample windows), the page's existing even-sampling truncation behavior still applies — the line is drawn from the sampled subset, with the same (sampled) annotation in the meta line under the map.