Talk:Proposed features/sidepath 2.0

From OpenStreetMap Wiki
Jump to: navigation, search

the information is already there

Actually this information is already implicitly in the db, in theory there's no need to tag this explicitly, the spatial configuration should be sufficient. On the other hand, for data consumers that want to be close to realtime osm, timeconsuming preprocessing is often not an option, although routing always (AFAIK) requires preprocessing steps anyway. So in short: I agree it would be nice to have this choice as a "routing customer" (according to your context you might even decide to take a longer route with less streets nearby), but it only has to be done, no need for an additional tag. --Dieterdreist (talk) 19:04, 15 June 2017 (UTC)

I sort of agree. Using a tag like cycleway=sidepath is only suitable for very basic evaluations. --Hubert87 (talk) 16:54, 16 June 2017 (UTC)
I agree... for the first part: don't duplicate info. The routing engine should use the highway=primary to increase the cost of using a way along it. And a tool with own maps like OSMAND can perfectly do the job if they need too. Once per map generation. More over it doesn't describe a feature but it's neighborhood. --Nospam2005 (talk) 19:58, 23 June 2017 (UTC)

would instantly improve bicycle routing (for example with Osmand)

Dear OSM community, Dear Dieter,

just an addition from my side: In case of doubt, please help to make this possible. The tag cycleway=sidepath is already being used for a long time. It would just be refined. This would instantly improve bicycle routing, because I would integrate this into the Osmand routing engine right away and start tagging some cycle paths along primary roads accordingly.

The three new tags would be easily available to any routing engine, without the need to write a code that interprets spatial relations. (And without the need to run this code anew each time the routing map / dataset is updated).

In my special case (improving Osmand), it would help to improve Osmand bicycle routing right away, as opposed to 'we have to wait for many months". The reason: In the team, we have a lot of volunteers (like me) and the team working on the calculation of routing data is heavily engaged with other important projects (some of which will also improve bicycle routing.)

This is no case of "tagging for the routing engine / the renderer", because the new tags are precisely defined and describe a real feature.

Of course I could also just tag cycleways with the existing cycleway=sidepath, and I have even started to do so (with some cycleways along primary roads in Germany). But just while doing this, I thought "would not cost me more time to tag it more precisely, not just as a sidepath, but as a sidepath of a primary road".

Thanks to all


that is an interesting (as it is completely at odds with what is actually the current state of affairs) statement considering that cycleway=sidepath has nearly no use, and was never discussed, and was documented in a wiki page late 2016, which is not a "long time" either SimonPoole (talk) 08:01, 16 June 2017 (UTC)

You are right, the documentation for cycleway=sidepath is from 2016. But when we consider that fact, it seems quite impressive that the tag has been used more than 2000 times up to now. And there is also the more precise cycleway:right=sidepath (and cycleway:left) which has been used another 2000 times up to now. So more than 4000 uses altogether. And the cycleway:right=sidepath is more than five years old now: It is part of the "Luebecker Methode" of cycleway mapping, and has been documented there since Mai 2012 Best wishes, Malte
WARNING: cycleway:right/left/both=sidepath and cycleway=sidepath are being used very differently right now.
Very simplistic spoken: The first one says that this road HAS a cycleway that is mapped separately. The second one say this way IS a cycleway parallel to a road. --Hubert87 (talk) 16:41, 16 June 2017 (UTC)

Yes, I got confused - you are right. The cycleway:right=sidepath is applied to the road (according to the "Luebecker Methode"). Only the cycleway=sidepath is applied to the cycleway itself. Malte --Posteinwurf (talk) 23:14, 17 June 2017 (UTC)

separation distance

would it make sense to define some minimal separation distance between the road and the cycleway? For me, it's a difference if it is 0 (like a sidewalk) (or 0.5m in the SVZ example) and more then 2 meters of green, often with trees between road and cycleway.

rough definition of distance makes sense

Yes, a definition would make sense.

There might be difference between cities and highroads, though. Outside of cities, the cycleway running alongside a primary very often has a distance between 2 and 5 meters.

In these areas, I would still tag it then. For cyclists this still seems very close, because you got the pollution, the noise and the danger of accidents.

People that do not mind cycling next to a road with heavy traffic, can normally change that in the routing engine's settings. In Osmand for example, there is one mode that will not avoid cycling next to main roads at all. The next mode ("balanced") would only accept a slight detour to avoid this. Only the third mode would make a longer detour to avoid cycling next to a main road.

Some reasons why not, and an alternative proposal

  • As said above, cycleway=sidepath is not a mainstream cycling tag: it's not mentioned on the "bicycle" wiki page, it is used nearly exclusively around Munich in Germany, it was never discussed, and was documented in a wiki page only late 2016.
  • the proposal includes a double tagging: the road category would be tagged in the highway tag of the road and, independently, another time in the cycleway tag of the cycleway that parallels the road
  • Adding these tags would require huge retrofit work, that cannot be automated
  • A more feasible alternative would be to add a tag to those cycleways that are not parallel to roads. ("cycleway=not_sidepath" or something to that effect). This could be added gradually by hand, and maybe also by a (major) Maproulette, where an algorithm selects candidate cycleways by analysing the distance from roads.

I would even be willing to do a huge part of the mapping work myself. Actually it is not too much work, and can be done semi-automatic: I have already programmed an overpass code that shows any cycleway with a distance of 0-5 meters from a primary road. With that code, you can easily spot all the cycleways along main roads. Then I would check manually if they are really close, and add the tag manually. Afterwards, I could do the same for cycleways along primary roads that are tagged as "path with bicycle=designated" (quite frequent combination). Maybe there is still a faster way with JOSM, but even with my method I reckon that I could add the tag to all cycleways along primary roads in Germany in some hours. During the next weeks, I could then go on with other countries. So I would myself try to apply it systematically during the coming weeks. Even if there are just 4000 uses now, it will steadily increase.

Double tagging is true, but that is also the case with footway=sidewalk, a tag that has been used 500.000 times worldwide and makes sense. I think cycleway=sidepath is a quite similar case.

Spotting the cycleways far away from main roads would work as well, but I think it would be more work, because theere are also many cycleways tagged as "path and bicycle=yes / designated". So you would also have to mark these paths as a non-sidepath. And I think there are many of them. I think tagging the cycleways along primary roads would already be a big step forward for bicycle routing, and if I first concentrate on this, it would involve much less work.

Best wishes Malte

You can sort of spot standalone cycleway using this tag, already. It's those how are not tag this way. Granted, it doesn't with the use numbers yet.
But the way I understand your proposal, your are looking for a more reprise tag anyway --Hubert87 (talk) 16:58, 16 June 2017 (UTC)

Why want poeple avoid such routes?

What are the reasons for 'disliking' such routes? Noise, polution, obstacles, danger ... imho, it is more important to bring these information into the map rather than some local distance to a primary or secondary highway. Not every primary or secondary highway has a lot of traffic - esp. in rural areas far from big cities, where primary/secondary sometimes has less traffic than a unclassified near a city.

But such information is difficult to collect and to map, so the approach with a tag 'cycleway is nearby a primary' may work quite good. But this should also bei used for tracks along primary/secondary. Sometimes these tracks are used as cycleway.