Repository navigation
Open
Description
Part of #3096. Found while designing #3182.
Problem
At some junctions the junction face ends where a fan or a fillet of the junction has not yet met the way's carriageway edge. The face border cuts it. Then the grey shoulder ring has a notch at the border, and a fillet ends as a stub, not at its foot (the point where it touches the way's edge tangentially).
Places
Pictures: /media/timo/OSM1/tmp/ground-plan/junctions/pics/ (INDEX.md). Each picture is a top view from the crate's faces and labels (ground-plan 2013925), without lighting. Left half: the port of #3183's stop-mouth rule. Right half: the junction-record rule of #3182. Blue: route look; brown: stop look; grey: shoulder.
- c12 (census of ground-plan GP4: a stop mouth wears its unpaved arm's look #3183: service road and unpaved track, angled). Cell 34300/23098. Junction node 46.820347, 8.418890. The notch: about 46.820358, 8.418923.
pics/c12-16.png(16 m square): the shoulder has a notch where the junction face ends (about pixel 590, 380 of the left half). It is the same in both halves./media/timo/OSM1/tmp/ground-plan/gp4/m/census/zoom-c12.png: the "step and short tab" that ground-plan GP4: a stop mouth wears its unpaved arm's look #3183's census lists.
- c16 (census: unclassified hairpin, near-parallel track). Cell 34293/23098. Junction node 46.822091, 8.379356. The stub: about 46.822064, 8.379324.
pics/c16-16.png(16 m square) andpics/c16-zoom.png(5 m square): the fillet between the track and the road arm ends at the junction face border about 0.1 m wide, before its foot. Beside it, the shoulder has a notch (left half).
Why any stop look makes it visible
ground-planpaints the whole junction face in the junction's paved look. The way's own carriageway past the border has the same look. A fan or fillet that the border cuts shows only as a small notch in the grey shoulder.- A stop look (ground-plan GP4: a stop mouth wears its unpaved arm's look #3183, and TrailView ground plan: stop mouths follow the route's true edge (junction record data) #3182's rule) paints the fillet between the route and the unpaved arm in the unpaved arm's look, up to the route's edge. At the junction face border this look meets the way's own paved face. A cut that does not end at the fillet's foot leaves a step or a stub of the other look.
- So the fault is in the faces, and neither look rule can remove it.
What is known
- At the seams, the paved outline lies within 5 mm of the carriageway edge (centre ± half-width) for 95 % of the 9,126 seam sides of the fixture. 300 seam sides (3.3 %) do not cross the paved outline within the seam cut: another face lies there. These are candidates for this fault (probe
jprobe,/media/timo/OSM1/tmp/ground-plan/junctions/measure/). - Guess, not proven: the face ends at a seam (
faces/points.rsregion) or at abeyondcut of a fan that crosses a shared seam or a next turn (faces/mod.rs:938-1030). The fan's shoulder ring (band_rings) ends at the same cut.
Acceptance
- At c12 and c16, the shoulder has no notch at the junction face border, and each fillet ends at its foot.
- A count over every fixture cell: fans and fillets that a junction face border cuts before their foot. Target 0, or each remaining case named with its rule.
- The owner's geometry rules hold: no special case for a junction kind; one general rule for the whole fixture; faces and checks as today elsewhere.
Generated by Claude Code
Metadata
Metadata
Assignees
Labels
No labels
Activity