Blog
Architecture, self-hosting, MapLibre platform design, vector tiles, geospatial APIs, and operations from the Planisfy codebase.
A self-hosted map platform needs more than unit tests: it needs product loops, real storage, browser rendering, dataset activation, restart persistence, and restore evidence.
Why map platforms need both movable publication aliases and immutable resource URLs, and how that model supports caching, promotion, and rollback.
A successful Planetiler build is only one stage in the lifecycle from OSM source data to a validated basemap running in Martin.
Why liveness, dependency health, dataset readiness, and product readiness need different signals in a self-hosted map platform.
Following the complete lifecycle from uploaded source data through background processing, PMTiles storage, publication, style creation, and MapLibre rendering.
A published MapLibre style starts a chain of dependent requests for sources, TileJSON, tiles, glyphs, sprites, credentials, and cacheable resources.
A technical overview of what Planisfy manages today, which workflows have been validated, and where its current product boundaries remain.
MapLibre GL JS provides an excellent open renderer, but production map products still need data services, publication workflows, access controls, observability, and an operational platform around it.