Measure and range & bearing design pass

Status: Built (2026-10-31), for #1130: slices 1–3 in #1197, #1201 and #1202; slice 4's results are in §6.5 of the S-164 conformance review, whose finding 5 this follows.

The S-164 Part 4 accuracy tests (Accuracy1–15) ask an ECDIS to measure between typed positions, to give the position at a typed range and bearing, and to draw long great circles, rhumb lines and geodesic circles. SoundCharts has a measure tool, but it only measures rhumb lines by clicking, and nothing turns a range and bearing into a position. This note audits what's there, proposes numbered decisions for a better measure tool and a new range & bearing tool, and splits the work into slices.

On 2026-10-31 #1130 was rescoped to the tools only. They use the spherical geodesy the viewer already has. Moving to the WGS-84 ellipsoid is #1189, with its own licensing decision, so no decision here depends on it. The sphere is already within the S-164 tolerance: 1% or 30 m, whichever is greater, for distance and 1° for bearing. It's off by 0.04–0.12% on distance and about 0.2° on bearing across Accuracy1–9. (#1189 has since moved the tools to the ellipsoid; see §6.5 of the S-164 conformance review.)

The mock for this round is the Measure & range/bearing · design pass artboard on the Navigation & time page of the SoundCharts design canvas (https://claude.ai/artifact/2hWbgQ9wKM1gC2gM51mSCs): today's measure tool, the measure card in great-circle mode in light and dark, the range & bearing tool with a 2,000,000 m ring, and parts for the fields, locks and toolbar.

1. What the tests ask

Paraphrased from S-164 2.0.1 Part 4. The manual is IHO copyright, so it isn't quoted here and mustn't be committed.

Tests	The mariner…	Reads
Accuracy1–3	measures a great circle between two given positions, about 33 km apart in Accuracy1	distance, forward bearing, reverse bearing
Accuracy4–6	enters a start position, a distance and a bearing	the end position
Accuracy7–9	measures a rhumb line between two given positions, about 10.7 km in one of them	distance, forward bearing, reverse bearing
Accuracy10–12	draws great circles of about 3,000 km, and a geodesic circle of 2,000,000 m radius	the line against the chart; the circle against range rings at 2,000,000 m intervals
Accuracy13–15	draws rhumb lines of about 3,000 km	the line against the chart

Accuracy1 on the sphere, from 32°35.224'S 061°17.710'E to 32°27.436'S 060°58.609'E: 33,154 m (17.90 NM), forward 295.7°, reverse 115.9°. The test expects about 33.19 km, so the sphere is 36 m short, well inside the 332 m the tolerance allows.

2. What's there today

#	Observation	Where
A1	The measure tool only measures rhumb lines. Its tooltip says so.	MeasurePathState.ComputeLegs, Tooltip_MeasureMode
A2	It shows the forward bearing only. There's no reverse bearing, which differs from forward + 180° on a great circle.	MeasureLeg
A3	Distances show to 0.1 NM and bearings to the whole degree ({0:0.0} NM @ {1:000}°T). Rounding alone can use up to 93 m of the 107 m the 10.7 km test allows.	Status_MeasureLeg, Status_MeasureLegLabel
A4	Distances are always NM. The Distance unit setting (NM, metric, miles) only changes the scale bar. The tests give their values in km and m.	DistanceUnit, ScaleBarView
A5	Results only appear in the status bar and as labels on the map. Waypoint positions are never shown, and nothing can be copied.	MeasureTool.PushSummary
A6	Points can only be clicked, so a test position can't be placed exactly. The route editor's waypoint rows already accept typed and pasted positions in many forms (LatLonFormatter.TryParse).	RoutesPanelViewModel.CommitWaypointPosition
A7	No tool gives a position from a range and bearing. GeodesicHelper.DirectProblem in Core does it for great circles, but nothing in the UI calls it, and there's no rhumb-line version.	GeodesicHelper
A8	No geodesic circle or range rings. GeodesicHelper.TessellateArc makes circles for S-100 ArcByRadius (72 segments a circle).	GeodesicHelper
A9	Great-circle route legs are drawn in 24 segments however long they are. A 3,000 km leg gets 125 km chords, which bow away from the great circle at large scales.	RouteOverlayScene.GeodesicSegments
A10	There are two earth radii: 6,371,000.4 m in MarineGeodesy and 6,371,008.8 m in GeodesicHelper. They differ by 1.3 ppm, which is harmless, but measuring a path and then running the range & bearing tool back along it should agree to the metre.	both
A11	No MCP tool drives or reads the measure tool, so an agent can only check results from a screenshot.	McpTools/
A12	Promoting a measurement to a route makes rhumb legs, whatever was measured.	MainViewModel.PromoteMeasurementToRoute

Keys today: Cmd/Ctrl+M toggles Measure, I toggles Pick, Cmd/Ctrl+E toggles the route editor, Esc leaves the active tool, Enter finishes and Backspace removes the last point. Cmd/Ctrl+R and G aren't bound anywhere. Only one map tool is active at a time (MapToolController).

3. Decisions

The measure tool

D1. A measure card. While Measure is on, a card sits at the map's top left, under the time stamp (12, 44), in the slot the route editor's chip uses. The two tools can't be on together, so they never collide. The card is 296 px wide on MapOverlaySurfaceBrush with the usual border and radius. Its header holds the title, the geometry control (D2), a copy button (D8) and close. Its rows are From, To, Distance, Bearing and Reverse, then Total when the path has more than one leg. Before the first click the card shows the existing hint ("Click on the map to start measuring…").

D2. Rhumb or great circle. A two-part segmented control, Rhumb | Great circle, sets the geometry of the whole path. G switches it while the map has focus. The map draws great-circle legs as curves, and the leg labels and card follow at once. The tooltip loses "(rhumb line)".

D3. Default geometry: remember the last choice, starting at Rhumb. Rhumb is what the route editor makes by default and what an ECDIS shows by default. Over the short distances people usually measure, the two differ by less than the display precision. The choice persists in settings (MeasureGeometry), so someone working through the great-circle tests sets it once. Needs the user: whether great circle should be the first-run default instead.

D4. Forward and reverse bearings. Bearing is the initial true bearing at From. Reverse is the initial true bearing at To back towards From. On a rhumb line that's always forward ± 180°. On a great circle it isn't, and the tooltip says why ("The bearing from To back to From. On a great circle this isn't the forward bearing plus 180°."). Both are labelled °T.

D5. Precision. Distance to 0.01 NM, 1 m, or 0.001 km (D6). Bearings to 0.1°. Positions in the app's degrees and decimal minutes to 0.001' (LatLonFormatter), about 2 m. Map leg labels and the status bar use the same precision: 17.90 NM · 295.7°. At this precision display rounding takes at most 9 m and 0.05° of the tolerance.

D6. Units follow the Distance unit setting, with a second line. The primary distance uses the existing Distance unit setting: NM, km (m below 1 km) or statute miles. Under it, in muted text, is the other system: km and m when the primary is NM or miles, NM when it's metric. The tests give values in km and m and mariners think in NM, so both are always on screen. There's no new setting. The scale bar is unchanged.

D7. Typed positions. From and To are text fields. They show the position in the app's format and accept anything LatLonFormatter.TryParse does: degrees and decimal minutes, degrees, minutes and seconds, signed decimal degrees, and hemisphere letters before or after. That's the same parser the route editor's waypoint rows use. Enter or leaving the field commits. A position that doesn't parse gets a destructive border and an inline hint, "Not a position. Try 32°35.224'S 061°17.710'E or -32.587, 61.295", and the leg keeps its old point. Esc reverts the field. Typing From with an empty path places the first point. Typing To then makes the leg. On a longer path, From and To are the ends of the selected leg (D9). A committed point off screen pans the map to it, without zooming.

D8. Copy. The header's copy button puts the measurement on the clipboard as text: positions, geometry, distance in both unit systems, bearings and the earth model (D17). It's the evidence a test run records. Cmd/Ctrl+C with the map focused does the same.

D9. Multi-leg paths stay. Clicking keeps adding legs, as today, and the card shows the last leg. Clicking a leg's label on the map selects that leg, and the card's leg rows follow it (Leg 2 of 3, with ‹ › buttons). Total always covers the whole path.

D10. Promoting keeps the geometry. Promote to route gives each new leg the measure geometry: a great-circle measurement makes geodesic legs.

The range & bearing tool

D11. A new map tool next to Measure. The toolbar gains Range & bearing (Map.RangeBearingMode) between Measure and Route edit, with a compass-style icon. Cmd/Ctrl+R toggles it; nothing uses it today. View › Appearance gets a matching menu item. It's the ECDIS EBL/VRM pair with an origin you choose: a bearing line and a range ring from one point.

D12. Clicking. The first click sets the origin. After that the bearing line and range ring follow the cursor, and the card's Range, Bearing and End follow too. A second click fixes the end. Another click after that starts again from a new origin. Enter fixes the end at the cursor, Backspace returns to placing the origin, and Esc leaves the tool. Dragging still pans.

D13. A range & bearing card with locks. The card takes the same slot and style as the measure card. Its rows are Origin (a position field, D7), Range and Bearing (number fields), End (read-only, copyable) and Reverse. Typing a range locks it: a lock icon appears in the field, and from then on the cursor only swings the bearing, like a VRM set to a range. Typing a bearing locks the bearing in the same way, and the cursor only changes the range, like an EBL. With both locked, the end is fixed and the cursor does nothing. Clicking a lock, or clearing its field, unlocks it. Accuracy4–6 is then: type Origin, Range and Bearing, read End.

D14. Range entry. A number on its own is in the primary unit (D6). A unit suffix overrides it: 2000000 m, 2,000 km, 1079.9 NM, 1.2 mi. NM is case-insensitive. A bare M is refused with a hint, because charts use it for nautical miles and SI uses it for metres. The field shows the value back in the primary unit, with the other system underneath, as in D6. Bearing accepts 0–360 with optional ° or °T. ↑/↓ nudge the bearing by 1° (10° with Shift) and the range by one unit of its last digit.

D15. Geometry, and the ring is always a geodesic circle. The card has the same Rhumb | Great circle control and G key as Measure (D2, D3; its own remembered setting). It decides the bearing line and how the end is computed: along a great circle (GeodesicHelper.DirectProblem) or along a rhumb line (a new rhumb direct helper). The range ring is always a geodesic circle: every point on it is the range from the origin along a great circle. On a rhumb-line bearing the ring passes through the end only approximately, and the tooltip on Range says so.

D16. Range rings. A Rings row, off by default, adds rings at a set interval (every [2,000,000 m] × [3]), drawn as geodesic circles like the VRM. Accuracy12 sets the range to 2,000,000 m and rings every 2,000,000 m: the VRM and the first ring must lie on each other. Each ring is labelled with its range at its northern point.

Both tools

D17. Name the earth model. Each card's footer says, in muted text, Sphere · R 6,371 km, with a tooltip: "Distances and bearings use a spherical earth. That's within the S-164 accuracy tolerance. WGS-84 is tracked in #1189." It changes to WGS-84 when #1189 lands. One radius constant replaces the two (A10). Since #1189 the footer reads WGS-84 ellipsoid, and both tools measure on it.

D18. Long lines and circles are densified by length. Great-circle lines, geodesic circles and the bearing line are split into segments of at most 50 km, and at least 64 per circle. Each line is cut at the antimeridian (MarineGeodesy.SplitAtAntimeridian) and drawn at world copies like other chart data. A circle round a pole is drawn as an open line, never filled. Rhumb lines are straight in Mercator and need cutting only. Great-circle route legs get the same densification (A9).

D19. Coexistence. One map tool at a time, as today: turning on Measure, Range & bearing, Pick or the route editor turns the others off, and a card only exists while its tool is on. Switching away clears the tool, as today. The pick panel and the route panel are unaffected. With the route editor on, its chip takes the slot instead. Neither card hides the overscale or warning notes: they move down below the card while it's open.

D20. Keys.

Key	Measure	Range & bearing
Cmd/Ctrl+M	toggles the tool	switches to Measure
Cmd/Ctrl+R	switches to Range & bearing	toggles the tool
G	switches the geometry	switches the geometry
Enter	finishes the path	fixes the end at the cursor
Backspace	removes the last point	back to placing the origin
Delete	clears the path	clears the origin
Cmd/Ctrl+C	copies the measurement	copies the result
Esc	leaves a field (reverting it), then the tool	the same
Tab	moves through the card's fields	the same

G, Enter, Backspace and Delete act on the tool only while the map or the card's buttons have focus. In a text field they edit the text (RouteEditorKeys.IsTextInputFocused already decides this for ⌘E).

D21. Automation IDs. Measure: Measure.Card, Measure.Geometry (with Measure.Geometry.Rhumb, Measure.Geometry.GreatCircle), Measure.From, Measure.To, Measure.Distance, Measure.DistanceOther, Measure.Bearing, Measure.ReverseBearing, Measure.Total, Measure.LegPrevious, Measure.LegNext, Measure.Copy, Measure.Close, Measure.Model. Range & bearing: Map.RangeBearingMode, RangeBearing.Card, RangeBearing.Geometry (.Rhumb, .GreatCircle), RangeBearing.Origin, RangeBearing.Range, RangeBearing.RangeLock, RangeBearing.Bearing, RangeBearing.BearingLock, RangeBearing.End, RangeBearing.ReverseBearing, RangeBearing.Rings, RangeBearing.RingInterval, RangeBearing.RingCount, RangeBearing.Copy, RangeBearing.Close, RangeBearing.Model. Each value field's accessible name includes its value and unit ("Distance 17.90 nautical miles").

D22. Light and dark. The cards use the theme tokens (MapOverlaySurfaceBrush, BorderColor, the foreground brushes) and MonoFontFamily with tabular figures for every number and position. On the map, lines are the accent colour as today. The measure path and the bearing line are solid. The rubber band is dashed as today. The VRM is solid, extra rings are thinner at 60% opacity, and the origin and end are discs like measure waypoints. Labels use the measure tool's existing light and dark chips. An invalid field uses the destructive token in both themes.

D23. MCP tools. Four tools, named like the rest:

• set_measure takes points (each a position string the card accepts, or {lat, lon}), geometry (rhumb or greatCircle) and optional selectedLeg. It turns the tool on and returns what get_measure does. An empty points clears the path.
• get_measure returns the geometry, the earth model, and per leg the from and to positions (decimal degrees and the app's text), the distance in metres and NM, the forward and reverse bearings, and the total.
• set_range_bearing takes origin, range (metres as a number, or a string with a unit as in D14), bearing, geometry and optional rings: {interval, count}. It turns the tool on, locks what it's given, and returns what get_range_bearing does.
• get_range_bearing returns the origin, range, bearing, locks, end, reverse bearing, rings and earth model.

Values are unrounded in the results, so a test checks the maths and a screenshot checks the display. set_viewport then frames the result for a capture.

4. Out of scope

• The WGS-84 ellipsoid: #1189. D17 is the only place the UI mentions it.
• EBL/VRM from own ship, relative bearings and floating EBLs. Those belong with route monitoring (#1131). D13's origin can later be "own ship" without changing the card.
• More than one range & bearing at a time. ECDIS usually has two pairs. One covers the tests; add a second when someone asks.
• Parallel index lines, magnetic bearings and measuring along an existing route.
• Keeping measurements across sessions or exporting them other than by copy.
• A s100 CLI command for the same maths. The MCP tools cover the automation the tests need.

5. Slices

1. Geometry core. One earth radius (D17, A10). A rhumb direct problem and reverse bearings (D4, D15). Densifying lines and geodesic circles by length, cut at the antimeridian, also for great-circle route legs (D18, A9). A small formatter for D5 and D6 and a range parser for D14. Unit tests with the Accuracy1–9 positions and tolerances.
2. Measure card. D1–D10, D17, D19–D22 for Measure. set_measure and get_measure (D23).
3. Range & bearing tool. D11–D16 and the rest of D19–D22. Toolbar button, Cmd/Ctrl+R, the overlay with rings, and set_range_bearing and get_range_bearing.
4. Conformance. Run Accuracy1–15 through the MCP tools with captures, update finding 5 and the Accuracy rows of the conformance review, and close #1130.

Slices 2 and 3 both depend on 1 and can then go in parallel.

6. Decisions on the open questions

1. Great circle isn't the first-run default. Measure remembers the last choice, starting at Rhumb (D3).
2. Cmd/Ctrl+R opens Range & bearing (D11).
3. The second unit line always shows, on the cards only (D6). Map labels and the status bar show the primary unit, and Copy includes both (D8).