How TrailSplits ships its map data: releases, checks and safe activation
Updated 13 October 2026. This post first described the move from Switzerland to the Alps; coverage has since grown well beyond both, and the method below is what carried it.
TrailSplits rebuilds its data from open sources such as OpenStreetMap and open elevation models. Nothing is overwritten in place: each build is an immutable release, checked first and then made live.
The contract: releases + activation
- Build the artifacts on a build server.
- Check them: endpoints answer, metadata is complete, and sample routes and searches still give the expected results.
- Activate the release: the live map, search and routing move to it one by one, and if any step fails, all of them go back.
- Keep the previous release, so a switch can be undone.
This makes a refresh routine rather than a risk. The map data follows OpenStreetMap on a weekly cycle.
What gets built
- Terrain tiles (for elevation and the 3D view)
- Terrain overlays (slope, aspect, hillshade)
- Contour tiles
- The search index
- Routing data
Each artifact can be published on its own, and each is traceable to the release it came from.
Why this matters
- Traceable: the map, search and routing data on a page come from a known release.
- Reliable: production changes only after the checks pass.
- Scalable: a new region is just another release.
The sources and their licences are listed on Data sources; the public endpoints these releases power are described on the API pages.