On Tue, 14 Aug 2012, Pieren wrote: > Personally I find the suggestion of total_tracks reasonably appealing > initially, however it would have to be repeated across all the tracks > which seems ugly and still doesn't say with confidence which tracks > are connected. > > The relation approach is clearer and makes it absolutely clear which > tracks are being grouped but the relation may not however be the same > one as the route relation which makes it difficult. > > An automated approach would involve merging all close tracks and > figuring it out. Not trivial but quite computable in my view. The > result could then be converted into relations and a bot process could > put it all back into osm, however some people don't like bots!.
The biggest problem with fully automatic approach is probably pair of tracks distancing due to terrain avoidance. But I doubt that the total_tracks=* taggers could come up very consistent values either, e.g., in the case the tracks go on different slopes of a narrow valley etc. so I would not be too worried if our machine friend cannot do it so that everyone would be pleased :-). ...I actually suspect people want more an answer to this question: how many tracks are hidden by the rendering that draws only a single track while zoomed far-away, which then cannot be solved on the data level because the the correct answer varies depending on distancing that is perceivable (it would be trivial to answer that at the render time though). -- i. _______________________________________________ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging