13 Commits

Author	SHA1	Message	Date
maj00r	80379b991c	feat(editor): the work of the editor sessions not committed yet, as it stands: a curve traced on from the end of the diverging track of a switch goes on at its radius with no transition in, or from a straight right at the switch (alignment start_held, fit_going_on), the plan taking an end adjoining R 10 km or more for a straight; the switches and slips laid into curves, slid along them and laid alone on one straight (frame_path, switch slide); the route plan reading curves going on from the tracks adjoining, the kept designs taken or forgotten; the orthophoto and terrain model settings, downloads and layers, the tracks onto the terrain; the curvature of a path in the plan (bezier::plan_curvature); the Polish texts of all of it. Not built as committed	2026-10-29 22:19:06 +01:00
maj00r	ed3d6551d8	fix(editor): a straight between two curves turning the same way, traced in the mode of a curve, is laid as a straight: points along a curve, about 200 m of straight and a curve the same way (wgr.scn) were fitted as one compound curve of eight arcs, its straight two arcs run up to R 50000 and laid at R 10360 once the curve was scaled between its tangents. The two curves with a straight between them, fitted about as closely, weren't taken over it, and were refused anyway: the last curve rounded to the metre no longer fitted up to the last point. A compound curve with an arc more than 4 times flatter than its sharpest one gives way to curves with straights between them which keep to the points about as closely, where straights between curves may be laid; the curves fitted are settled to the metre, or as laid a hair under where that gives a warning, as the chain of a reverse curve was; the chain of curves takes curves turning the same way with straights between them too, three of them, and wins over the two curves where it keeps closer. Unit tests of the points of wgr.scn and of two and three curves turning one way with straights between them	2026-10-26 19:50:00 +01:00
maj00r	9b596e7ffa	feat(editor): reverse curves laid through points in the mode of a curve: points clicked along an S, with a straight between its curves or none, gave "No curve follows the points clicked along it", and a curve to and fro of three was laid as one curve running metres off the points. A stretch traced is fitted as a chain of two to four curves too: the points split where the curvature they show changes its side, a curve for each part, the lines between them, the common tangents of a reverse curve, fitted with the radii and the transitions from the tangents at the points of inflection. Two curves turning the other way may meet with no straight between them, their transitions meeting at the inflection, or their arcs; the radii which fill the room between their vertices are kept as they were laid, a hair under, so the design gives no warning of radii made smaller. Two turning the same way stay one compound curve unless a straight of shortest_between parts them. A unit test traces eight such lines, clicked exactly and 5 cm off, with other radii and the ends not parallel, and finds the radii within 3 %	2026-10-26 18:23:04 +01:00
maj00r	34bd7c853e	fix(editor): no path of a few millimetres laid by the route design, and no camera at no finite place: an arc of a compound curve squeezed to almost nothing in the plan (a share of 0.00002) made pieces of a few millimetres, which the simulation takes for paths of no length (Bad track: zero length spline, it measures a path in steps of 5 cm), laid as tracks with kinks at their joints. alignment::pieces gives such a stretch to the piece next to it, the one between the same joints if there's one, so no piece is shorter than 10 cm, and a piece of no finite length isn't split for ever. A zoom with the wheel aimed at a point which isn't a number doesn't move the camera, and a camera gone to no finite place goes back to where it last was: right after such tracks were laid the graphics driver hung (TDR) and the whole machine froze. A unit test cuts 4000 compound curves of random arcs, transitions and shares into pieces, as the plan lays them over the paths there and all over again, and finds none under 5 cm	2026-10-25 17:26:16 +01:00
maj00r	97b6a84d1b	fix(editor): a curve laid from a free end on a curve, or to one, goes on from its curvature: traced (Tab) from the end of the diverging track of a switch it begins right at P4 at the radius of the switch, not on a straight nor with a transition from a curvature of 0. The design of the line takes the curve of the adjoining track at each end (start_curved/end_curved with the radius, the side and the cant, kept with it in the scenery): the curve of the first vertex turning its way, or of the last one, begins right at the end, its transition going from that curvature and cant to those of its first arc, as between the arcs of a compound curve, none where it goes on at its radius; the radius of its arcs is the one which begins it there, solved for the tangent, the curves next to it making room on the straight. A curve traced next to such an end is fitted by its transitions alone, with none between the curves where the points keep to that as well, or laid from a straight there where it doesn't go on from it or that keeps to the points much closer: eased out of the curve of a plain track by the transition of the norms, the line beginning at its end, or right from P4, as a switch goes on with a straight. A bend of straights turning the way of the curve gets its curve going on from it with the transitions and the cant of the norms; the other way, or a straight, as before: eased out of the curve of a plain track, the jump of the curvature kept at a switch, as at its start, and not warned of. The cant of the adjoining track goes on into the curve with no ramp at that end; the norms ask for the transition between the curves only where the curvature differs by more than 5 % or the cant by 5 mm. The readout by the cursor, the status after laying and the plan tell the curve going on from the track, the plan with a box to go on or not. Dragged back to P4 from a point the curve goes on from its radius too. Selftests laywalk, laybends, curvejoint, endcurves; unit tests of a curve going on from a switch and into a free end, its cant, traced, through a transition into a larger radius, and of a bend of straights after such an end	2026-10-25 12:28:55 +01:00
maj00r	29b9b93353	fix(editor): the plan warns of the curved track adjoining a curve only where the curvature jumps there: a curve going on from the diverging track of a switch at its radius (R 300 after R 301 in wsd) no longer has it. The side the adjoining track turns to is taken with its radius into the chain and the design, and the curvature the design meets it with, signed by its turn, compared with it, within 5%; a design kept without the side is taken for turning its way. Unit test: R 300 from the very start after R 301 turning the same way gives no warning, turning the other way or starting with a straight does	2026-10-25 11:14:49 +01:00
maj00r	e43863932d	feat(editor,track): a cant ramp of its own, off the transition curve: each end of a curve may have its ramp moved and of a length of its own (ST-T1-A6 9.6 pkt 5 and 11), on the straight before the curve or on the arc, the transition keeping the full cant all along, as a switch laid on it wants it; set in the plan under "Cant ramps", stored with the vertex as "ramps", reversed with the curve, marked across the track in the view. The cant of a line is a profile along it now, the paths cut where a ramp of its own starts and ends, their rolls following it, warnings where a ramp is shorter than its cant asks for or reaches into the curve before, past the line or over the other ramp. A switch on a cant ramp is drawn following it: its rails, blades, frogs and bed made at the mean of its rolls, as before, are turned along each path to the roll of the path there, so it no longer steps at its ends against the tracks either side. The checks of a switch laid into a curve warn of the values of table 11.9 in brackets as of the others, with no word of a consent. Unit test of a ramp moved onto the straight	2026-10-25 00:54:50 +02:00
maj00r	aeef2cebb3	feat(editor): the plan shows which curve is edited: each named W1, W2... as its vertex is in the view, with its km and "Show in the view", which flies the camera over it; the part of it whose field the mouse is over or which is being edited (the whole curve, an arc of a compound one, the transition in, out or between two arcs) lit up in the view and named there with its radius, cant or length. The points clicked when a line is laid are kept with its design in the scenery ("click" lines, moved with it): the plan shows them in the colours of their departure and, while "Keep to the points clicked" is on, fits the radii of a curve whose transitions or cants are changed again to them, its transitions and cants as set; "Transitions for these cants" makes the transitions as long as the norms ask for at the radii and the cants set, an arc of a compound curve with no cant lengthening the ones either side of it. Unit test: 50 m between the arcs become 85 m, the curve 4 cm off the points clicked instead of 58	2026-10-25 00:11:58 +02:00
maj00r	2a399a92a5	feat(editor): each arc of a compound curve has a cant of its own, set in the Plan by each radius ("Cant of R1 (mm)"...), ramped along the transition curves between the arcs; until it's set, the cant of the curve in proportion to the curvature of the arc, as before, "auto" taking it back to that and "Apply recommended for V" all of them. The transitions between the arcs of the norms follow the cants set; warnings where the cant of an arc exceeds the norms, where the unbalanced acceleration does on any arc, and where a transition curve, at the ends or between the arcs, is shorter than the change of the curvature and of the cant along it needs. The labels of the curves show the cants of the arcs where they differ. Stored with the vertex as "cants", reversed with the curve; a unit test	2026-10-24 23:41:26 +02:00
maj00r	0577073926	fix(editor): a compound curve is fitted with the transition curves of the norms between its arcs from the first step, not found without them and given them after: from several starts, the problem having more than one bottom, each count of arcs from the curve fitted with one less, its longest arc cut in two, and from the curvature the points show, the fewest arcs taken which keep within 0.3 m of the points, or within 0.1 m of the closest; and the transitions fitted no longer than half as long again as the norms ask for, 30 m at least, a fit started below that	2026-10-24 22:59:47 +02:00
maj00r	f94ddb293f	fix(editor): a compound curve of more than three radii, made of bends clicked close along a curve or traced, has the transition curves of the norms between its arcs too: its radii found with the arcs meeting directly, then fitted again with the transitions between them; in abc the curve keeps within 0.12 m of the points clicked with them	2026-10-24 22:40:28 +02:00
maj00r	4602f727bb	feat(editor): bends clicked close along a curve make one compound curve, not curves with short straights between them: a row of bends turning the same way, the straights between their curves shorter than as many metres as km/h of the speed (30 m at least), is taken for points of a curve seen on the photo, one curve fitted through them between the straights before and after it, with as many radii meeting with no transitions between them as keep it within 0.3 m of the points, up to eight, and taken where it keeps within 1 m; a row which doesn't is cut in parts, two bends alone joined into one curve along the line their curves laid. A curve traced is two curves turning the same way only with a straight between them as long as that and keeping much closer to the points than one compound curve; a fit running off to the largest radius gives two transition curves meeting. The transitions fitted turn no more than the curve; arcs and spirals are found many times quicker, the curve of a compound fit made straight away, a transition integrated in 64 steps at most. The fits wait on the thread of the laying in a queue of their own, the one asked for last fitted first	2026-10-24 22:09:27 +02:00
maj00r	500f068297	feat(editor): the curves of the laying from the photo in the design of the lines: a vertex of two transition curves meeting at its radius, with no arc and no cant, kept in the scenery files; the bends of straights eased with them at a radius from 6000 down to 3000 m, or larger where they'd be too short for the speed, else with an arc of the norms as large as fits, the straights between two bends shared, the bends staying where they were clicked and a curve below the norms kept with its warnings; a stretch traced on the photo fitted with a straight, a curve with its radius and the length of its transitions, a compound one, or two curves with a straight between them, on a thread of its own while the cursor moves; the heights from point to point through all the points clicked. Unit tests of the design of the lines	2026-10-24 20:57:59 +02:00