fix(map): draw transit routes along real roads, not straight lines - #802
Draft
smmariquit wants to merge 1 commit into
Draft
fix(map): draw transit routes along real roads, not straight lines#802smmariquit wants to merge 1 commit into
smmariquit wants to merge 1 commit into
Conversation
Seven of the nine transit routes had no entry in jeepney-geometries.json, so Map.svelte fell back to deriveRouteLineFromStops and joined their raw stop coordinates. On the intercity services that meant 4-to-6-point straight lines running across Laguna de Bay and open country. Geometry now comes from OSM route relations where one maps the service, and from OSRM road routing otherwise: - uplb-to-upd / upd-to-uplb: relations 17932247 / 17932246 (DLTB, end to end) - lb-to-sta-cruz / lb-to-san-pablo: relations 2582682 / 1856207, the mapped Calamba-origin corridors trimmed to the Los Banos leg - lb-to-calamba: relation 1856209 reversed; the return relation 1856216 is tagged disused - lb-to-buendia / buendia-to-lb: routed, no relation covers these - kaliwa-kanan / forestry: untouched. The campus relations 337589/337590 stop at the campus boundary and would drop the Olivarez Plaza town leg Provenance ships with each line. A routed line says so in the route modal, and a route with no geometry at all now draws dashed and faded with an explicit "not mapped" note instead of a confident straight line. Sourced-but-qualified lines (reversed relation, trimmed corridor, and the SLEX/C-5 vs Skyway disagreement on the DLTB bus) carry their own caveat. Geometry stays in the JSON keyed by route id rather than moving to a column: it is generated from OSM, not editor-maintained, so a migration would buy nothing. Lines are simplified to 3 m and written one route per line, which keeps the bundled file at 44 KB.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| await sleep(1500); | ||
| const body = await run(); | ||
| const parsed = JSON.parse(body); | ||
| writeFileSync(path, body); |
| await sleep(1500); | ||
| const body = await run(); | ||
| const parsed = JSON.parse(body); | ||
| writeFileSync(path, body); |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
Seven of the nine transit routes had no entry in
src/constants/jeepney-geometries.json, soMap.sveltefell back toderiveRouteLineFromStopsand connected their raw stop coordinates. With 4–6 stops over 60–75 km, that drew straight lines across Laguna de Bay, Mt. Makiling and private land. Onlykaliwa-kananandforestryhad real geometry.Where the geometry came from
uplb-to-updosm-relationupd-to-uplbosm-relationlb-to-sta-cruzosm-relationlb-to-san-pabloosm-relationlb-to-calambaosm-relationdisused:lb-to-buendia,buendia-to-lbroutedkaliwa-kanan,forestryrouted(unchanged)Why the campus loops were not switched to OSM. Relations 337589 / 337590 (Kaliwa / Kanan) exist and are gap-free, but they stop at the campus boundary. Our route includes the Olivarez Plaza and Robinsons town leg, so adopting them would have removed a correct part of the one route that already worked. Forestry relations 1954831 / 1954832 carry a
disused:prefix and 2020 data.Honesty over coverage
resolveRouteGeometryreturns the source alongside the line, and it drives the UI:osm-relation— solid line, no caveat, unless the route carries a specific one.routed— solid line plus a note that it is road-routed between stops rather than traced from the operator's published route.stops-only— dashed, faded, and labelled "the exact path of this route is not mapped yet". No route ships in this state today; it is the safety net for any route added without geometry, and it replaces today's confident straight line.Three routes are sourced but still qualified, and say so in the route modal:
lb-to-calamba: the line is the mapped return trip drawn backwards.The generator also refuses a relation whose stitched line misses either terminal by more than 1.5 km, and asserts the mapped direction matches what the config claims — so a bad pairing fails loudly instead of shipping as a nicer-looking wrong line.
Where it lives, and why not a column
Still the JSON keyed by route id. The geometry is generated from OSM, not editor-maintained, so a
jeepney_routes.geometrycolumn would have cost a migration, a PGlite regeneration and sync changes to buy nothing. No migration in this PR; nothing to hand-apply to prod.scripts/generate-transit-geometry.tsreplaces the stale rootfetch-routes.ts(which only knew the two bundled constants routes). It reads stops read-only, caches every upstream response underdata/transit-geometry-cache/and serialises requests, so re-runs cost upstream nothing.Lines are simplified with Ramer-Douglas-Peucker at 3 m — narrower than a lane — which takes the bundled file from 466 KB to 44 KB. The file is excluded from Biome formatting because pretty-printing coordinate pairs quadrupled it.
Data bugs found, not fixed here
Read-only pass over prod; no database writes. Three stop coordinates are wrong and now visibly so:
14.1542, 121.1979) is ~2 km off the highway, up the Makiling foothills, on three routes. The routedlb-to-buendialine visibly detours to reach it.lb-to-san-pablois 7 km off-corridor; the real jeepney runs Bay → Calauan → San Pablo.lb-to-sta-cruzsits 2.9 km off the mapped corridor.forestry's CPAf stop hassort_order9 while sitting geographically mid-route, so the stop list is misnumbered.Worth a follow-up data pass; deliberately out of scope here.
Verify
bun run test(439 pass),bunx vitest run(155 pass),bunx biome checkclean on touched files,bun run buildgreen.dist/server/entry.mjsand inspected each. Campus loops confirmed unchanged.Notes for review
Draft, and deliberately no
run/e2e— the serial CI queue is running and concurrent e2e runs exhaust the shared pooler.