> Volker Schmidt <vosc...@gmail.com> het volgende geschreven: > >> On Mon, 19 Aug 2019 at 15:40, Peter Elderson <pelder...@gmail.com> wrote: >> Ideally, you should not have to create gpx-s from them and you should need >> no ordering or routing at all, because they ARE the routes. An app or >> gps-device should use them as is, just tell the user what to do next. Since >> no app currently does that (future still has to arrive) we resort to >> transferring the route to them as tracks, i.e. gpx. > > Now we are getting closer to the point. You are correctly saying "no app is > currently doing that". So why should we sort topologically non-sortable > route-relations members? We have a solution that works with existing tools on > unsorted hiking/cycling routes, and that is routing with strong preference on > the use of ways that are part of cycling/hiking routes. > I see the problem from the mapper's perspective (as I map a lot) and from the > end-users perspective (I very often design bicycle tour routes from OSM > data). > I am not a data consumer in the sense I do not write software thta uses OSM > data, I am an end usere, eclusivley using the software produced by others) > and I acknowledge that my experience is limited to cycling/hiking routes. I > am sure there are routes that have different problems and may need sorting, > One such category are most likely public transport routes, which are used in > a completely different way.
Routes in osm describe closely the routes found on the road. They are described as the chain of ways you follow. If they don’t, that should be fixed so they again reflect the situation on the road. The osm route IS the route, and it should be usable as is, without redoing the routing. If you find it necessary to do the routing again, no matter how cleverly you do that, you don’t fix the problem, you are just fixing errors at the wrong end, leaving the osm data as flawed as before. (Question: where does your weighted routing start and end?) If you have a method of fixing the data, I would like it better if you make that avaible as a tool to enhance/ensure the quality of the osm-data, so the recorded routes can be used as the ready-to-use routes they are supposed to reflect. Once routes are reliable for direct use, they will be used much more frequently in real applications. Osmand could use it, they now demand a linear gpx even for routes that are already known and displayed because you just can’t rely on OSM routes. If they could, exact turnbyturn instructions could be given to the user, complete with other map info, without any routing algorithm, because the route is already there. In the current state, reliability is simply too low for direct use. If that remains the case over the next few years, mark my words, you can stop putting routes into osm, because other sources will offer far better functionality based on actually reliable data. The people who mark the ways will record changes realtime, and that will be available for users with apps near realtime. Then osm mappers who now often are faster with updates, will always be slower and they will be recording things that are already done better elsewhere (and shown on the OSM of course, just leaving out the messy data) > ____________________________________________ > Tagging mailing list > Tagging@openstreetmap.org > https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging