Richard Mann wrote: >then yes they probably will get converted into tags on the road, just as >soon as that renders properly. Rendering gain trumps notional information >loss. The Danes are just ahead of the curve.
I think they have been too eager to discourage drawing the cycleways separately. The national officials here are allegedly constructing a database for a online routing service for cycling and they have concluded that the information can not be described with sufficient detail for very accurate routing "as tags on the roads". Other authorities already made such a service available for the capital region, but now they wan't to expand it to cover the whole country. (Sorry, the document was only in Finnish). Just try describing this intersection (which isn't that complex, even) solely with tags on the "proper" roads: (make sure you have the bicycle layer visible, if the link gets truncated) http://elanor.mine.nu/daeron/kartat.php?zoom=17&lat=60.20853&lon=24.94616&layers=0B000000000 Consider especially a cyclist arriving from the south on the eastern cycle track, and how to guide them towards other directions... As most of those are shared use tracks, also think how a wheelchair user/ visually impaired user gets as good instructions as possible. It would have been very sufficient in the beginning to add cycleway=track only on the roads, but as the level of detail has increased, removing the later added highway=cycleway ways would be outright destructive. Additionally, I feel a point that hasn't been mentioned here, for only expanding the area covered with separate highway=cycleway ways, is that the transitions from cycleway=track to a highway=cycleway next to and along the road either introduces a nonexistant 90 degree curve (or two) away from the road centerline, or where the cycleway then continues away from the road at a shallow angle, the highway=cycleway way starts at an incorrect location. "Keeps straight things straight". Linking the cycle track to the road next to it could use some explicit information; some have mentioned using relations, but wouldn't a simple tag sidewalk=this, on the cycle track separated by a curb stone or a thin green patch, suffice? "The nearest road segment (roughly) parallel to this way, is the road that most would call this cycleway part of". The same idea works with addressing; node and a tag "this belongs to the nearby way (with this name)". There's no support in software, yet, and I don't know if osmosis could even reasonably find such ways, but that shouldn't be an excuse to delete the more accurate data. It didn't take long for authors to implement checking for Karlsruhe schema addressing, as in finding the point on the correct road, so if it were urgently needed, someone could. And most of the time it wouldn't even be needed, since surely a cycleway/footway few meters from a parallel road _is_ the sidewalk, unless there's a building in between. Only where such a cycleway is roughly midway between two different roads and very close to both of them, one couldn't deduce it - but they hardly could dedice between the two when mapping, either, in such border cases. -- Alv _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
