Along
An errand planner for Singapore public transport. Given a trip someone is already making, it finds stops that satisfy one or two errands and ranks them by the travel time, walking and transfers they add to the direct journey.
14,777outlets in its place catalogue
How it works
- The optimiser is deterministic. It generates candidate stops, prunes them against the route corridor, routes the survivors and scores each one against the direct trip. For two errands it tests a shared stop and both travel orders.
- Place search runs on a SQLite FTS5 catalogue built from OpenStreetMap, with 14,777 outlets in 27 categories, 252 malls and 5,978 transport nodes.
- A language model is optional and only interprets unfamiliar requests. It never produces routes or place results.
- Rail legs are drawn in LTA's line colours, so a local reader recognises a purple leg as the North East line before reading a word. Buses stay uncoloured because operators do not colour-code their services.
- At version 0.8.4 it had 322 backend tests and 160 browser tests, run on mobile and desktop in both light and dark themes.
- Opening hours are published for only about 20% of outlets, so the rest are shown as unknown rather than guessed.