If there are multiple sources of ski run properties (way & relations) these are flattened into a single representation to allow for a simpler UX.
The current approach treats all data sources equally. For example: run difficulty is merged by taking the easiest value from the available sources.
A somewhat common case for nordic ski trails is a longer trail, tagged as a relation has piste:difficulty=easy, but one of its ways may be tagged as piste:difficulty=intermediate.
In the case currently we show the whole trail as easy. However, the ways data should be closer to the ground truth, given relations work at a higher level.
The improvement would be to treat the way data as primary, and only fall back to the current strategy if the way doesn't have a difficulty value.
If there are multiple sources of ski run properties (way & relations) these are flattened into a single representation to allow for a simpler UX.
The current approach treats all data sources equally. For example: run difficulty is merged by taking the easiest value from the available sources.
A somewhat common case for nordic ski trails is a longer trail, tagged as a relation has piste:difficulty=easy, but one of its ways may be tagged as piste:difficulty=intermediate.
In the case currently we show the whole trail as easy. However, the ways data should be closer to the ground truth, given relations work at a higher level.
The improvement would be to treat the way data as primary, and only fall back to the current strategy if the way doesn't have a difficulty value.