Hi Marc, That analysis site is wonderful! Maybe we should start a 'weekly assignemnt' to solve all the problems reported. Most of them are not very hard to solve. Add a way, sort the ways once again, split a way and readd one of the parts.
Do you think you (we) could create something comparable for public transport? find interrupted/broken route itinerary relations locate route_masters where there are too many or too few route relations for the variations find routes which don't serve all stops find lines which aren't mapped at all yet find lines with route relations but route_masters still missing Jo 2015-05-15 21:34 GMT+02:00 <m...@vmarc.be>: > Jakka, > > I cut the roundabouts in pieces highway to highway. Is this the properly >> way? >> But gives the tester prg like http://osma.vmarc.be/ ... an alert? >> > > The logic in osma.vmarc.be is intended to work OK independent of whether > the roundabout is mapped as a single way in a closed loop and tag > "junction=roundabout", or cut into different ways, each with or without tag > "junction=roundabout". > > For hiking networks the analyzer is intended to choose the shortest path. > For bicycle networks the analyzer is intended to follow the direction of > the ways in as mapped in the round-about. > > Roundabouts you must take at the right hand.(Europe) >> > > The analyzer assumes the actual direction of the ways adhere to this rule > and currently has no logic to verify if this is really the case. According > to the wiki (http://wiki.openstreetmap.org/wiki/Tag:junction=roundabout) > this should be a fair assumption: quote: "The OSM ways of the roundabout > itself must be drawn in the direction the traffic flows." > > For an example of a roundabout that is not split into different ways you > can look at the roundabout close to the start node in route > http://osma.vmarc.be/en/route/1029070. If you zoom in on the roundabout > in the map and hoover over the different parts of the roundabout, you will > see that one part of the roundabout is chosen for the forward path of the > route (from low node number to high node number), and the other part for > the backward path (from high node number to low node number). The other > two roundabouts in this route are cut into different ways. Here also the > different parts are assigned to the proper forward and backward path. > > Ways that are closed loops are handled similar to roundabouts. Route > http://osma.vmarc.be/en/route/1173862 has such a closed loop without tag > "junction=roundabout" (way 6849252) at the end node of the route. Only the > parts of the way that are actually used in the forward and the backward > paths of the route are used (note that the bit at the north side is not > used in forward nor the backward path). > > > With respect to route 2718669: the analysis result in > http://osma.vmarc.be/en/route/2718669 currently thinks this route is not > OK because it does not find an end node for the route in any of the ways in > the route relation. The start node 1442170312 (which is part of way > 202508685) does have a tag rcn_ref=5, but there is no node with such tag in > any of the other ways of the route, also not in the last way 163973463. > When I add rcn_ref=99 in node 1489841045, then the analyzer logic thinks > this route is completely OK (apart for an informational message > RouteNodeNameMismatch because the analyzer does not expect the text > "Verlaine-sur-Ourthe" in the route name). I do not see any ways with tag > "junction=roundabout" in this route though. > > a large square, marketplace, which cannot be tagged as roundabout ... >> > > An example in a walking-network of a marketplace that is mapped with a > single way in a closed loop can be found in the last way of route > http://osma.vmarc.be/en/route/3145182. This way is recognized by the > analysis logic as a loop and treated the same as a roundabout. Because this > is a walking network, the analyzer chooses the shortest path to the > destination node. In route http://osma.vmarc.be/en/route/3941922 another > part of the of the same way is chosen as shortest path. > > > To summarize, it should be OK to map roundabouts either as a single way or > cut into separate ways. Either way of mapping a roundabout will have its > pros and cons. The logic in osma.vmarc.be is intended to understand > roundabouts and loops, and do the right thing, independent of how they are > mapped. The intention is to not have "alerts" or "false positives" for this. > > I am sure the analysis logic in osma.vmarc.be is not OK yet in all > possible situations. If you have examples where you have doubts or see > analysis results that are just plain wrong, I would like to hear about > this. I think there are still a lot things to be improved. > > Regards, > Marc > (osma.vmarc.be developer) > > > _______________________________________________ > Talk-be mailing list > Talk-be@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk-be >
_______________________________________________ Talk-be mailing list Talk-be@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-be
