Links to OSM objects sometimes incorrect in OpenMapTiles preview map

[jacobwhall]

In the following screencast, the first link is correct but the second is not. From this page

Screencast.From.2025-11-09.23-30-12.mp4

[jake-low]

Thanks for documenting this. The root cause is Planetiler's Feature Merging. By default, Planetiler assigns numeric IDs to features based on their OSM type and ID (the formula is osm_id * 10 + (1 if node, 2 if way, 3 if relation)). When you call FeatureMerge.mergeNearbyPolygons() or similar, Planetiler combines any polygons in the tile that have the same attributes (even if they have distinct IDs) into a single multipolygon, but that multipolygon is assigned an ID from some arbitrary member of that input set (possibly the lowest ID value? I'm not entirely sure). For applications that decode the ID from the tile feature and expect it to point to an OSM element, this gives misleading/wrong information.

You can see this in the screenshot below; all of the highlighted buildings have been merged into one MultiPolygon feature in the tile, and have been given an ID number that corresponds to an arbitrary member of the input set (relation/4476350), which can be seen on the far left.

Feature merging is essential for keeping tile sizes manageable, so this behavior is typically desirable (unless you're trying to create links from map features back to OSM, like we're doing here). But IMO it would be better if Planetiler could assign merged features an ID that indicates that they do not correspond to a single specific OSM feature, so that we could at least display something like "composite feature" in the popup instead of linking to the wrong OSM element. ID numbers ending in 0 are already reserved for this purpose (e.g. if you're tiling data from both OSM and non-OSM sources together), so perhaps the right fix is to modify FeatureMerge.java to assign an ID ending in 0 to each merged feature. I've been meaning to open an issue for Planetiler that suggests this feature; thanks for giving me the necessary nudge to do so.

[jacobwhall]

Ah, I see. Thanks @jake-low for the detailed explanation. The direct links to OSM features made me think it was intended to be a one-to-one relationship

[jake-low]

This is now fixed: Planetiler v0.10 (which we updated to in 7d1ed5a) gives composite features an ID that ends with 0 to distinguish them, and de58206 adds a tooltip in the viewer explaining what these IDs mean.

[jacobwhall]

Awesome, thanks @jake-low!