[OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
Hi, I'm working on the cycle node network in Flanders/Southern Netherlands. I created a collection relation: http://www.openstreetmap.org/browse/relation/1726882 This should make it a lot more convenient to find all the nodes/routes/networks involved in the rcn. Fiddling with XAPI and such is not very productive... It's extremely time consuming and very error prone. Which contains a network relation for each set of numbered nodes (where each set contains only one time 01,02, etc) (I'm working without the maps of Toerisme Vlaanderen, so I had to improvise for the naming and where to make a subdivision) For the moment I'm abusing the role to add the node number. This works easier and is only temporary. I'll take them all out again, when I'm done inventorizing. The network relations contain the route relations. As far as I'm concerned, the route relations don't need to contain the nodes anymore. They are part of the ways, anyway. So the route relations should only contain (a preferably) continuous set of nodes. I try to have them start at the lower numbered node. If the forward and backward relation don't follow the same route, I use forward/backward roles. The idea I'm following is that it should be possible to go from the lower numbered to the higher numbered node by following all the ways without a role until ways with forward roles are met. Then all the ways with forward roles until a way with a backward role is found. Then skip the ways with backward roles and continue with the ways without a role. All this, because I want to be able to 'validate' the routes for continuity with a script/program and this will probably simplify the life of the people of Fietsnet as well. I'll post an example to illustrate later on. Polyglot ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
It makes perfect sense to me and I totally agree. But I assume you mean ways in stead of nodes in 'So the route relations should only contain (a preferably) continuous set of nodes' On top of that, I find it handy when the route realations have a name, so it is easy to see which are the relations of a road. http://www.openstreetmap.org/browse/way/73069100 When they have 2 or 3 meaningful letters in front (followed by a space) of the numbers, you can also easily see on the hike or bike maps which network is in place eg ZD 239-240 = Zuid-Dijleland from node 239 to 240 http://hiking.lonvia.de/?zoom=12&lat=50.779&lon=4.56392 With only the numbers, http://www.openstreetmap.org/browse/relation/14142 it is truncated and thus cluthering the map. http://cycling.lonvia.de/?zoom=13&lat=50.78855&lon=4.59216&layers=FFBTT Also when a way is part of more than one network (hike and bike) the numbers don't tell which is which. Here a good addition could be DL for Dijleland Alternative is to provide an osmc tag with the shortened name in it (probably not working for bike maps). This tagging info for prper naming should be added to the Wiki http://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Walking_Routes#Walking_node_networks and http://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Cycle_Routes#Cycle_Node_Networks and maybe also in the general remarks in conventions. http://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/Walking_Routes http://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/Cycle_Routes Regards, Gerard. Jo wrote: Hi, I'm working on the cycle node network in Flanders/Southern Netherlands. I created a collection relation: http://www.openstreetmap.org/browse/relation/1726882 This should make it a lot more convenient to find all the nodes/routes/networks involved in the rcn. Fiddling with XAPI and such is not very productive... It's extremely time consuming and very error prone. Which contains a network relation for each set of numbered nodes (where each set contains only one time 01,02, etc) (I'm working without the maps of Toerisme Vlaanderen, so I had to improvise for the naming and where to make a subdivision) For the moment I'm abusing the role to add the node number. This works easier and is only temporary. I'll take them all out again, when I'm done inventorizing. The network relations contain the route relations. As far as I'm concerned, the route relations don't need to contain the nodes anymore. They are part of the ways, anyway. So the route relations should only contain (a preferably) continuous set of nodes. I try to have them start at the lower numbered node. If the forward and backward relation don't follow the same route, I use forward/backward roles. The idea I'm following is that it should be possible to go from the lower numbered to the higher numbered node by following all the ways without a role until ways with forward roles are met. Then all the ways with forward roles until a way with a backward role is found. Then skip the ways with backward roles and continue with the ways without a role. All this, because I want to be able to 'validate' the routes for continuity with a script/program and this will probably simplify the life of the people of Fietsnet as well. I'll post an example to illustrate later on. Polyglot ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
On Wednesday 31 August 2011 15:12:05 Gerard Vanderveken wrote: > It makes perfect sense to me and I totally agree. > But I assume you mean ways in stead of nodes in 'So the route relations > should only contain (a preferably) continuous set of nodes' > > On top of that, I find it handy when the route realations have a name, > so it is easy to see which are the relations of a road. > http://www.openstreetmap.org/browse/way/73069100 > When they have 2 or 3 meaningful letters in front (followed by a space) > of the numbers, you can also easily see on the hike or bike maps which > network is in place > eg ZD 239-240 = Zuid-Dijleland from node 239 to 240 Do not give names to the route relations of cycle and walking node networks. They don't have names so you shouldn't invent one. Use the "note=*" tag for what you want to put in the name tag, and JOSM will gladly show the note to you in the relation list. Greetings Ben ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
Ben Laenen wrote: On Wednesday 31 August 2011 15:12:05 Gerard Vanderveken wrote: It makes perfect sense to me and I totally agree. But I assume you mean ways in stead of nodes in 'So the route relations should only contain (a preferably) continuous set of nodes' On top of that, I find it handy when the route realations have a name, so it is easy to see which are the relations of a road. http://www.openstreetmap.org/browse/way/73069100 When they have 2 or 3 meaningful letters in front (followed by a space) of the numbers, you can also easily see on the hike or bike maps which network is in place eg ZD 239-240 = Zuid-Dijleland from node 239 to 240 Do not give names to the route relations of cycle and walking node networks. They don't have names so you shouldn't invent one. Use the "note=*" tag for what you want to put in the name tag, and JOSM will gladly show the note to you in the relation list. Greetings Ben That is OK for JOSM, but a list like this (at the bottom) is simply not meaningful. http://www.openstreetmap.org/browse/changeset/917?relation_page=3 This is much clearer. http://www.openstreetmap.org/browse/changeset/9132576 altough a short name would also do: http://www.openstreetmap.org/browse/relation/1641610 The route name is not 'invented', everyone referr to it as the route from node A to node B So why not formalize it? I guess the province has internally a similar naming. That it is not on streetsigns is no objection for me. Tracks and paths from roads of the Atlas don't have official names either, and yet it makes perfect sense to name them as numbered in the atlas as eg Buurtweg 23, Sentier 45, etc.) http://www.openstreetmap.org/?lat=50.82162&lon=4.61208&zoom=15&layers=M Regards, Gerard. ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
On Wednesday 31 August 2011 16:09:54 Gerard Vanderveken wrote: > That is OK for JOSM, but a list like this (at the bottom) is simply not > meaningful. > http://www.openstreetmap.org/browse/changeset/917?relation_page=3 Then go complain to whoever is in charge of how that page is rendered? > This is much clearer. > http://www.openstreetmap.org/browse/changeset/9132576 > altough a short name would also do: > http://www.openstreetmap.org/browse/relation/1641610 > The route name is not 'invented', everyone referr to it as the route > from node A to node B > So why not formalize it? Because it's not its name. And because suddenly renderers have no idea whether they should render the name of a route or not. Or will you also invent another tag like 'show_name=no'? Keep the database clean and don't sacrifice correct data for a little bit of legibility on a random webpage that for all we know may soon support showing the note tag. Greetings Ben ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
> That is OK for JOSM, but a list like this (at the bottom) is simply not > meaningful. > http://www.openstreetmap.org/browse/changeset/917?relation_page=3 > This is much clearer. Then ask for that page to display a note=* tag when a name=* is absent, like JOSM does. But why would you start making collection relations when a scheme already exists? See here: http://wiki.openstreetmap.org/wiki/Cycle_Node_Network_Tagging Which as far as I can see meets your needs and relations conforming to that scheme have been in the database for over a year. -- Lennard ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
I created route and network relations as is described on that page; so far, so good. There is no way of grouping networks (yet) though. That's what I'm using the collection relation for. Kind regards, Polyglot 2011/8/31 Lennard > > That is OK for JOSM, but a list like this (at the bottom) is simply not > > meaningful. > > http://www.openstreetmap.org/browse/changeset/917?relation_page=3 > > This is much clearer. > > Then ask for that page to display a note=* tag when a name=* is absent, > like JOSM does. > > But why would you start making collection relations when a scheme already > exists? See here: > > http://wiki.openstreetmap.org/wiki/Cycle_Node_Network_Tagging > > Which as far as I can see meets your needs and relations conforming to > that scheme have been in the database for over a year. > > -- > Lennard > > > ___ > Talk-be mailing list > Talk-be@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-be > ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Fietsknooppuntennetwerk/Cycle node network
I got a bit farther in my endeavour to add all rcn route nodes and relations to network relations. This is a very big task though, and I'm getting a bit overwhelmed, so if somebody feels like joining the effort, that would be nice. The following networks are ready: Dijleland Hageland Droog Haspengouw The following are WiP (work in progress); Pajottenland Brabantse Kouters And then there are some for which I tried to add a sensible group of nodes, until I hit a node number that is used twice. What I do: For each node, determine which ways have rcn relations. For each such relation remove name, ref and created_by tags note=xx-yy (with xx < yy) order the members, such that the way in position 1 starts at xx if there is a split, because of oneway streets, a roundabout or cycle paths on both side of the road then label all roads in direction from xx to yy with a forward role, to where they join again then mark all roads back to the bifurcation with a backward role (when following the relation from xx to yy ignoring the ways with a backward role will get you there) Check that it's in fact possible to get from xx to yy. Sometimes a tiny part of road is missing, or chunk of road needs to be split off. Trying to determine wich ways belong to a specific network is probably the hardest part. In part because that task has not been made very simple by the people who assigned those numbers. Once I have a set of nodes that seem to belong together, I use a Python script I created, to add all the relevant relations automatically. This creates duplicates though, which need to be removed manually, but that is less effort than adding all the relations manually. Before uploading I also tend to check for license compliance. If the people of Fietsnet are going to be using our data, it's not done that part of the data drops into an abyss... I'm also creating a Python script which will be able to validate that continuity of the route relations, as I already did for a subset of the routes of De Lijn. OSM data is somewhat fragile in the sense that people split and recombine ways without taking into account the effect this may have on route relations and I'd like to be able to detect that automatically. Here's hoping somebody will take an interest in helping out... Jo ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be