chore(prototype): log lane connections at prototype segment nodes

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

docs/research/vehicle-probe.md

@@ -46,3 +46,65 @@ Ticket "Probe vehicle data and native passing in game" (part of map "Map: TpF3 o
 ## Decision
 
 Overtaking on two-way roads (one lane per direction) is dropped (decided by the user, 2026-10-05). The effort continues on multilane roads only: a held-back vehicle should switch to a lane the slower vehicle isn't in.
+
+## Segment rebuild (ticket "Can rebuilding a road segment move a passer past its blocker?")
+
+- The script sent 3 `makeWorldBuildProposalCmd` proposals (remove segment + re-add a copy of its `BaseEdge`/`BaseEdgeStreet`, `ignoreErrors = true`) for segments 29701 and 29699 (`country_new_medium`). **None took effect**: both segment ids stayed in use by vehicles until the end of the log, and 29699 was "rebuilt" twice. The game logged no error, and there's no callback in engine state. The copied `BaseEdge.distance` read `0.0`, so the copy is probably incomplete. A next attempt would need `api.engine.util.proposal.makeProposalData` to see the proposal's errors.
+- The human's manual segment deletion left no trace in the probe data: every vehicle stayed `tvState=1` (en route), with only normal path trimming.
+- Related find: `api.cmd.makeVehicleSendToDepotCmd(vehicle, sellOnArrival, jumpToDepotEntity)` "teleports" a vehicle to a depot. The engine can relocate vehicles instantly, but only to depots.
+
+## Lessons from downloaded mods and the game folder (2026-10-05)
+
+Scanned: the 9 mods in `C:\Users\Public\mod.io\10640\mods\` (e.g. `devers_1`, `move_it_probe_overpass`, `zhenya_auto_assign_terminals`, `celmi_timetables`, `burntnuggie_line_priority_1`), and every `.lua`/`.tl` in the install's `base`, `mods/release` and `dlcs`, including inside the zips.
+
+- **Callbacks only work on the GUI thread.** Mods send commands from a `guiUpdateScript`/`guiHandleEventScript` (declared in the same `.gs.lua`) and get `function(res, success, resultEntities)` back. Errors are readable via `res.resultProposalData.errorState.messages`. From `update` (engine thread), callbacks fail with `Error decoding argument #3`, which is what we hit.
+- **Rebuilding a segment:** `api.engine.util.proposal.replaceSegment(entity, streetTemplate?)` returns a ready `Proposal`. The official campaign (mission 03 `init.script.tl`) sends it from engine state with `makeWorldBuildProposalCmd(proposal, nil, true, true)`. `devers_1` uses it the same way. It keeps lane links, edge objects (signals) and electrification. `api.engine.util.proposal.makeProposalData(proposal)` validates a proposal (`.costs`, `.errorState`) without sending it.
+- **Teleport:** only `makeVehicleSendToDepotCmd(vehicle, sellOnArrival, jumpToDepotEntity)`. Mission 05 uses it for a ship, then `makeVehicleSetLineCmd(vehicle, newLine, 0)` and a restart. No coordinate or lane-position teleport exists for line vehicles. `makeCustomVehicleCreateOrUpdateCmd` is aircraft-only ("carrier currently must be Air").
+- `api.engine.util.transport.calcPosition(edge.geometry, t)` gives a world position along an edge geometry.
+- `burntnuggie_line_priority_1` sets line **reservation** priority via `makeLineUpdateCmd`. That's path reservation for rail, not road traffic.
+- `makeVehicleSetLineCmd` to the vehicle's own line: the lanes planned ahead change, but the engine re-plans them on its own all the time (the leader's changed too, without a command). It doesn't force a lane change, and with `stopIndex=0` it re-targets the line's first stop.
+
+## Segment rebuild via `replaceSegment` (run 2026-10-05 ~20:55 UTC)
+
+- `api.engine.util.proposal.replaceSegment(seg)` plus `makeWorldBuildProposalCmd(p, nil, true, true)` from engine state **works**: segments 29702, 21304 and 29012 never appeared again after their rebuild (they got new ids).
+- **After all 3 rebuilds, the follower and leader ended up in different lanes** (`country_new_medium`, 2 lanes per direction). Each time the *leader* moved from lane 4 to lane 3, either right away or one segment later, and the follower stayed in lane 4.
+  - #1 (seg 29702): follower 29251 at 50.4 m behind leader 29852 at 76.7 m. Afterwards, on seg 29699, 29251 is at 29.1 m (lane 4) and 29852 at 20.2 m (lane 3). 29251 drives 18.6–19.6 m/s, above its earlier 16.67, and stays a segment ahead. **The follower got past.**
+  - #2 (seg 21304): 29851 (leader) moved to lane 3 on seg 27833. The lanes split, but the order was unchanged.
+  - #3 (seg 29012): 29251 (leader) moved to lane 3 on seg 29322.
+  - Without rebuilds, the same pair stayed together in lane 4 for 6 segments (first probe). The engine also switches lanes by itself at segment ends sometimes, so 3/3 is suggestive, not proof.
+  - Likely cause (human's hypothesis): the vehicles' `MovePath` refers to lanes of the removed segment, so the engine must re-attach them to the new segment and re-plan, and a vehicle can land in a different lane.
+- **Debug `MovePath` exchange is unusable from engine state**: `api.cmd.sendCommand(api.cmd.debug.makeComponentExchangeCmd(v, movePath))` fails with `Error decoding argument #3` even **without** a callback. So the earlier error wasn't caused by the callback. The only untried route is the GUI thread.
+- Not yet tested: a rebuild on a two-way (one lane per direction) road.
+
+## Rebuild on a two-way road (run 2026-10-05 ~21:12 UTC)
+
+- 3 `replaceSegment` rebuilds on `country_new_small` (one lane per direction), segments 30213, 29776 and 30215, follower 25.8 m behind leader. **No effect**: the next snapshot had the same gap (22.6/48.4 → 49.3/75.1), both vehicles stayed in lane 2, same speeds, same order for 10+ snapshots. Rebuilding does not move vehicles past each other on single-lane roads.
+- How the multilane lane change happens is still uncaught: snapshots every 2 s only show before (both lane 4) and after (the leader in lane 3 two segments later). The probe now traces the two vehicles on every update for ~30 s after a rebuild (`TRACE` lines), and rebuilds only multilane segments.
+
+## Segment rebuild dropped (2026-10-05)
+
+The human confirmed in game what the log showed: rebuilding the segment under a held-back vehicle does not get it past the blocker on a single-lane road. The rebuild code is removed from the probe (it's still in git history, last in commit 5d3a254). The lane split after rebuilds on multilane roads (see above) is left as an observation for the multilane ticket, not pursued here.
+
+## Position flip via `MovePath` exchange on the GUI thread (run 2026-10-05 ~21:23 UTC)
+
+- On the GUI thread (`guiUpdate`, with a callback), `api.cmd.sendCommand(api.cmd.debug.makeComponentExchangeCmd(follower, leaderMovePath), cb)` fails with the **same** `Error decoding argument #3: The lua object can't be cast to desired type`, and the callback is never reached. Debug mode was active.
+- So the error is not about callbacks or threads: **the exchange refuses a `MovePath`**, consistent with its doc comment (`EmissionEmitter` or `ModelInstanceList` only).
+- **Conclusion: no script API can write a road vehicle's position or route.** Moving a vehicle past another (jump, flip, teleport) is impossible for line vehicles. The only exception is `jumpToDepotEntity`, which can only target a depot.
+- The probe retried every frame (5951 errors) because the error came before the counter was incremented. The experiment is removed, and the probe is logging-only again.
+
+## Overlaid passing lanes (ticket "Can a street type with overlaid passing lanes fake overtaking on single-lane roads?")
+
+- Prototype: `mod/content/streets/overtake_proto_{a,b}.street_template.lua`, copies of `country_new_small` with a 0.5 m passing lane per direction at the centre line, offset ±2.5 into the oncoming lane (A and B use opposite signs). A mod street template sits anywhere under `content/` and refers to base resources with `::/…`.
+- First in-game run (human, 2026-10-05): the template loads, the lanes are visibly offset as intended, and vehicles can drive on the passing lanes. **Town (non-player) cars sometimes choose the passing lane.** **A faster bus behind a slower vehicle does not switch to the empty lane**: the switching logic is missing.
+- Changes for the next run: passing lanes allow only `BUS`/`TRUCK` (keeps private `CAR` town traffic out), and the probe logs the road template per vehicle (`road=`), since lane indices 2/3 also mean ordinary lanes on `country_new_medium`.
+- Hypothesis for the switch: the engine only re-chooses lanes at segment boundaries (`GetNextEdgeTrafficPenaltyAndAlternatives`), so short segments give a blocked bus more chances to pick the passing lane. Fallback trigger: `replaceSegment` ahead of the pair (made the leader re-plan lanes on multilane roads).
+- Second run (2026-10-05 ~21:50 UTC, variant B, passing lanes BUS/TRUCK only). The human judged **variant B correct** (backward passing lane offset -2.5, forward +2.5), so variant A was removed. On the prototype road, lane 4 is the normal forward lane and lane 3 the forward passing lane.
+  - **An overtake happened**: fast 29251 (lane 4, 22.22 m/s) got past slow 29851 over segments 29707 → 30211 (side by side) → 30215 → 29708 → 29042 and stayed ahead. But the *slow* vehicle was in the passing lane (lane 3), and the passer went past in the normal lane.
+  - The engine treats both lanes as equal: some vehicles cruise in the passing lane almost the whole way (29796: 28/28 samples in lane 3, 27775: 29/29). Others mix (29251: 33 in lane 4, 14 in lane 3).
+  - Vehicles slowed to 9–10 m/s and briefly stopped on segment 27290 (presumably the end of the prototype stretch): a merge from two lanes into one.
+  - Visually the lane change is abrupt (human: "the lanes need a smoother transition"): the switch happens at the node, which has almost zero length on a straight, and the jump is about 5–6 m sideways.
+  - Next ideas: a lower speed limit on the passing lane (prefer the normal lane unless it's blocked), a smaller offset or transition segments for a smoother change.
+- Third run setup: passing lanes at 20 m/s (normal lanes 22.22 m/s), everything else unchanged. Question: does a lower limit keep vehicles in the normal lane and send them to the passing lane only when the normal lane is blocked?
+- Third run (20 m/s passing lanes): **no change.** Slow buses still sat in the left passing lane (29849 and 29851: 29/29 samples in lane 3); fast 29251 split 8/11. The engine doesn't pick lanes by speed limit (the slow buses top out at 16.67 < 20 anyway). Observed pattern: a vehicle gets a lane when it enters the stretch and keeps it; it doesn't move over when blocked; passes happen only when fast and slow happen to be in different lanes.
+- Lane connections are editable: `BaseNodeConfig.laneConnections` (`LaneConnection{segment0, lane0, segment1, lane1, withRoad, withTram}`, lanes zero-based left-to-right seen from the node) plus `userModifiedLaneConnections`, scriptable via `SimpleStreetProposal.nodeConfigsToAdd`. The game also has a native **crossing tool** ("allows to change the road and tram lanes at crossings", `UI::LaneModifier` in the executable). Next test (option a, by hand): at the stretch's entry junction, connect the normal road only to the right lane, to see whether entry control works and whether a blocked fast bus moves left inside the stretch.
+- Fourth run (after the human's crossing-tool edits): **entry control not effective.** Slow buses 29849, 29851 and 29852 still entered prototype stretches (segments 21304, 29012) in the left lane 3 and stayed there (12–14/12–14 samples). 29850, 29848 and the fast 29251 entered in lane 4. No vehicle switched lanes inside a stretch. It's unknown whether or where the edits were saved, so the probe now logs the lane connections (`NODE … userModified= conns=`) at both nodes of every occupied prototype segment.

mod/_metadata/modinfo.json

@@ -9,7 +9,9 @@
     "summary": "Vehicle overtaking (WIP)",
     "name": "Overtake",
     "tags": [
-        "script"
+        "Script Mod",
+        "Car",
+        "Road"
     ],
-    "url": ""
+    "url": "https://git.arthurerlich.de/haylan/tpf3-overtake"
 }

mod/content/overtake_probe.script.tl

@@ -1,10 +1,11 @@
--- PROTOTYPE (ticket "Probe vehicle data and native passing in game"): throwaway, lives only on branch prototype/vehicle-probe.
--- Every INTERVAL it dumps every player-owned road vehicle, then groups them by street segment
--- so lanes (edge index) and oncoming traffic (opposite direction / turnaround edge) are visible in stdout.txt.
--- ponytail: module-local timer, not GameScriptState; fine for a probe, resets on reload.
+-- PROTOTYPE: throwaway, lives only on branch prototype/vehicle-probe.
+-- Every INTERVAL it logs every player-owned road vehicle (segment, lane, position, speed).
+-- All experiments (timeToIgnore, jump, set-line, segment rebuild, GUI position flip) are in git history and
+-- docs/research/vehicle-probe.md; none moved a vehicle past another on a single-lane road.
+-- ponytail: module-local timer, not GameScriptState; fine for a probe (state resets between calls anyway).
 
 local TAG = "[overtake_probe] "
-local INTERVAL = 5.0
+local INTERVAL = 2.0
 local elapsed = 0.0
 local ticks = 0
 
@@ -25,76 +26,53 @@ local function turnaround(edge : EdgeId) : integer
 	return tn.turnaroundEdges[edge.index + 1]
 end
 
--- EXPERIMENT: does api.cmd.debug.makeComponentExchangeCmd accept a MovePath, and does a positive
--- dyn.timeToIgnore make the follower ignore the vehicle ahead? Hypothesis only; nothing documents either.
-local IGNORE_SECONDS = 10.0
-local MAX_GAP = 40.0
-
-local record Car
-	v : integer
-	pos : number
-	speed : number
-	dir : boolean
-	mp : Engine.Component.MovePath
+-- short street template name of a segment, so passing-lane use on the prototype road can be told apart
+local function roadOf(seg : integer) : string
+	local be = api.engine.getComponent(seg, api.type.ComponentType.BASE_EDGE) as Engine.Component.BaseEdge
+	if be == nil then return "?" end
+	return (tostring(be.roadTemplate):match("([^/]+)%.street_template$")) or tostring(be.roadTemplate)
 end
 
-local function tryIgnore(follower : Car, leader : Car)
-	local mp = follower.mp
-	mp.dyn.timeToIgnore = IGNORE_SECONDS
-	p(string.format("IGNORE_TEST send veh=%d leader=%d gap=%.1f v=%.2f leaderV=%.2f", follower.v, leader.v, math.abs(leader.pos - follower.pos), follower.speed, leader.speed))
-	local cmd = api.cmd.debug.makeComponentExchangeCmd(follower.v, mp as Engine.Component)
-	-- engine-state scripts can't pass a callback ("Error decoding argument #3"); the next dump shows the effect
-	api.cmd.sendCommand(cmd)
+-- lane connections at both nodes of a segment, to check whether crossing-tool edits exist where vehicles enter
+local function nodeLog(seg : integer)
+	local be = api.engine.getComponent(seg, api.type.ComponentType.BASE_EDGE) as Engine.Component.BaseEdge
+	if be == nil then return end
+	for _, node in ipairs({ be.node0, be.node1 }) do
+		local cfg = api.engine.getComponent(node, api.type.ComponentType.BASE_NODE_CONFIG) as Engine.Component.BaseNodeConfig
+		if cfg == nil then
+			p(string.format("NODE seg=%d node=%d no BaseNodeConfig (default connections)", seg, node))
+		else
+			local conns : {string} = {}
+			for _, lc in ipairs(cfg.laneConnections) do
+				if lc.segment0 == seg or lc.segment1 == seg then
+					table.insert(conns, string.format("%d:%d->%d:%d", lc.segment0, lc.lane0, lc.segment1, lc.lane1))
+				end
+			end
+			p(string.format("NODE seg=%d node=%d userModified=%s conns=%s", seg, node, tostring(cfg.userModifiedLaneConnections), table.concat(conns, " ")))
+		end
+	end
 end
 
 local function dump(dt : number)
 	local player = api.engine.util.getPlayer()
 	local vehicles = api.engine.getEntitiesWithComponent(api.type.ComponentType.TRANSPORT_VEHICLE, { requireOwnedByPlayer = player })
 	p(string.format("tick=%d dt=%s playerVehicles=%d", ticks, tostring(dt), #vehicles))
-
-	local bySegment : {integer : {string}} = {}
-	local byLane : {string : {Car}} = {}
+	local protoSegs : {integer : boolean} = {}
 	for _, v in ipairs(vehicles) do
 		local lv = api.engine.getComponent(v, api.type.ComponentType.LAND_VEHICLE) as Engine.Component.LandVehicle
 		local mp = api.engine.getComponent(v, api.type.ComponentType.MOVE_PATH) as Engine.Component.MovePath
+		local tv = api.engine.getComponent(v, api.type.ComponentType.TRANSPORT_VEHICLE) as Engine.Component.TransportVehicle
 		if lv ~= nil and not lv.isTram and mp ~= nil then
 			local edge, dir = edgeAt(mp)
-			local info = api.engine.system.landVehicleMoveSystem.getVehicleInfo(v)
-			local seg = edge and edge.entity or -1
-			local line = string.format(
-				"veh=%d seg=%d lane=%s dir=%s turnaround=%s pos=%.1f speed=%.2f accel=%.2f blocked=%s state=%s standing=%.1f timeToIgnore=%.2f timeUntilAccel=%.2f occ=%s..%s brake=%.1f power=%s traction=%s",
-				v, seg, tostring(edge and edge.index), tostring(dir), tostring(edge and turnaround(edge)),
-				mp.dyn.pathPos.pos, mp.dyn.speed, mp.dyn.accel, tostring(mp.blocked), tostring(mp.state),
-				mp.dyn.timeStanding, mp.dyn.timeToIgnore, mp.dyn.timeUntilAccel,
-				tostring(mp.occupiedBackIndex), tostring(mp.occupiedFrontIndex), lv.dyn.brakingTimer,
-				tostring(info and info.availPower), tostring(info and info.traction))
-			p(line)
-			if edge ~= nil then
-				local key = seg .. ":" .. edge.index
-				byLane[key] = byLane[key] or {}
-				table.insert(byLane[key], { v = v, pos = mp.dyn.pathPos.pos, speed = mp.dyn.speed, dir = dir, mp = mp })
-			end
-			bySegment[seg] = bySegment[seg] or {}
-			table.insert(bySegment[seg], string.format("%d(lane=%s,dir=%s,v=%.1f,blk=%s)", v, tostring(edge and edge.index), tostring(dir), mp.dyn.speed, tostring(mp.blocked)))
-		end
-	end
-	for seg, list in pairs(bySegment) do
-		if #list > 1 then
-			p(string.format("SHARED seg=%d: %s", seg, table.concat(list, " ")))
-		end
-	end
-	for _, cars in pairs(byLane) do
-		for i = 1, #cars do
-			for j = 1, #cars do
-				local a, b = cars[i], cars[j]
-				-- ponytail: assumes pos grows in driving direction when dir=true; the log shows whether that holds
-				local ahead = (a.dir and b.pos > a.pos) or (not a.dir and b.pos < a.pos)
-				if i ~= j and ahead and math.abs(b.pos - a.pos) < MAX_GAP then
-					tryIgnore(a, b)
-				end
-			end
+			p(string.format(
+				"veh=%d road=%s seg=%d lane=%s dir=%s turnaround=%s edgeIndex=%d/%d pos=%.1f speed=%.2f state=%s tvState=%s line=%s standing=%.1f",
+				v, edge and roadOf(edge.entity) or "-", edge and edge.entity or -1, tostring(edge and edge.index), tostring(dir), tostring(edge and turnaround(edge)),
+				mp.dyn.pathPos.edgeIndex, #(mp.path.edges as {any}), mp.dyn.pathPos.pos, mp.dyn.speed, tostring(mp.state),
+				tostring(tv and tv.state), tostring(tv and tv.line), mp.dyn.timeStanding))
+			if edge ~= nil and roadOf(edge.entity) == "overtake_proto_b" then protoSegs[edge.entity] = true end
 		end
 	end
+	for seg, _ in pairs(protoSegs) do nodeLog(seg) end
 end
 
 local ret : GameScript<table> = {

mod/content/streets/overtake_proto_b.street_template.lua

@@ -0,0 +1,43 @@
+-- PROTOTYPE (ticket "Can a street type with overlaid passing lanes fake overtaking on single-lane roads?"):
+-- throwaway, lives only on branch prototype/vehicle-probe.
+-- Looks like country_new_small (same style, one visible lane per direction), but each direction gets a narrow
+-- extra passing lane at the centre line, pushed by `offset` over the oncoming lane. Passing lanes run at 20 m/s
+-- (normal lanes 22.22) to test whether a lower limit keeps vehicles in the normal lane unless it's blocked.
+-- Offsets B -2.5 / F +2.5 put the passing lanes over the oncoming lanes (variant B; the opposite signs, variant A,
+-- were wrong and are removed).
+function data()
+return {
+		availability = { yearFrom = 0, yearTo = 0 },
+		cost = 180,
+		country = true,
+		defaultWithCrosswalk = false,
+		description = {
+			description = "PROTOTYPE overtaking road, variant B: country small road with overlaid passing lanes (offset B -2.5 / F +2.5).",
+			icon = "::/infrastructure/street/country/country_new_small.tga",
+			name = "Overtake proto B (passing lanes B -2.5 / F +2.5)",
+			previewIcon = "::/infrastructure/street/country/country_new_small_preview.tga",
+		},
+		emissions = { noise = 0, pollution = 0, pollutionRadius = 0, radius = 0 },
+		fillGroundTex = "::/infrastructure/street/shared/street_fill.gtex",
+		laneConfigs = {
+			{ forward = false, height = 0, offset = -1,   speed = 22.22, transportModes = { "PERSON", "CARGO" }, width = 2 },
+			{ forward = false, height = 0, offset = -0.5, speed = 22.22, transportModes = { "CAR", "BUS", "TRUCK" }, width = 5 },
+			-- backward passing lane, meant to sit over the forward driving lane (BUS/TRUCK only: keeps town cars out)
+			{ forward = false, height = 0, offset = -2.5, speed = 20, transportModes = { "BUS", "TRUCK" }, width = 0.5 },
+			-- forward passing lane, meant to sit over the backward (oncoming) driving lane (BUS/TRUCK only)
+			{ forward = true,  height = 0, offset = 2.5, speed = 20, transportModes = { "BUS", "TRUCK" }, width = 0.5 },
+			{ forward = true,  height = 0, offset = 0.5,  speed = 22.22, transportModes = { "CAR", "BUS", "TRUCK" }, width = 5 },
+			{ forward = true,  height = 0, offset = 1,    speed = 22.22, transportModes = { "PERSON", "CARGO" }, width = 2 },
+		},
+		maintenanceCost = 30,
+		menuCategory = { categories = { { category = "roads_country", filterCategories = { }, order = 442 } } },
+		modifiers = { maxSpeedModifier = 0, noiseModifier = 0, pollutionModifier = 0 },
+		pedestrianWalkPenalty = true,
+		priority = 200,
+		roadType = "STREET",
+		sidewalkFillGroundTex = "::/infrastructure/shared/none.gtex",
+		simBuildable = false, -- towns must not build the prototype
+		streetStyle = "::/infrastructure/street/country/country_new_small.street",
+		builderAudioRes = "::/gui/construction/sound/buildoze_street.builder_audio",
+	}
+end