You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The ocean and land layer of the shortbread style use ST_SimplifyPreserveTopology() in their layer queries. This function considerably slows down tile generation for me. Some tiles take more than a minute to generate because of that.
Given that the ocean layer relies on external data anyway, the MVT geometries could simply be pre-generated. Doing so for zoom 0-12 takes less than 4 hours on my machine and results in about 3GB of data. That means the data could even be pre-generated once a night when new coastlines arrive. Zoom 13 and 14 should probably better off holding the original OSM data. If that is still too much data, a global simplify over the entire (unified) water_polygon table should only take a couple of minutes.
The land layer is a bit more tricky because it needs to be minutely updated. Just speculating but how about a generated column which is filled with a simplified geometry, but only iff the geometry is too large to reasonably handle (ST_MemSize is a great function for quickly estimating how big the polygons are). Then use coalesce(simplified_way, way) to get reasonably sized data.
The text was updated successfully, but these errors were encountered:
The ocean and land layer of the shortbread style use
ST_SimplifyPreserveTopology()
in their layer queries. This function considerably slows down tile generation for me. Some tiles take more than a minute to generate because of that.Given that the ocean layer relies on external data anyway, the MVT geometries could simply be pre-generated. Doing so for zoom 0-12 takes less than 4 hours on my machine and results in about 3GB of data. That means the data could even be pre-generated once a night when new coastlines arrive. Zoom 13 and 14 should probably better off holding the original OSM data. If that is still too much data, a global simplify over the entire (unified) water_polygon table should only take a couple of minutes.
The land layer is a bit more tricky because it needs to be minutely updated. Just speculating but how about a generated column which is filled with a simplified geometry, but only iff the geometry is too large to reasonably handle (ST_MemSize is a great function for quickly estimating how big the polygons are). Then use
coalesce(simplified_way, way)
to get reasonably sized data.The text was updated successfully, but these errors were encountered: