Talk:Consolidation footway cycleway path

Unjoin designated

Isn't "unjoin designated" basically doing nothing, since designated is already defined to be only used for legally dedicated ways? ("specially designated (typically by a government)") --Hawke 16:31, 19 August 2009 (UTC)

Some consider that all highway=footway are equal to path+foot=designated (but that not all such paths are footways), some say that the path+foot=designated equals a highway=footway (not all highway=footways are designated) and still others think they're equal in both directions. The implications for "correct" tagging vary significantly with the viewpoint taken. Alv 21:40, 19 August 2009 (UTC)

Hmm. You're saying that there are things tagged highway=footway which are not footways? That's clearly a silly way to tag. From what viewpoint is that "correct" tagging? --Hawke 17:29, 20 August 2009 (UTC)

Should have been more clear; "... such paths are not equal to highway=footway" - if a path has, say, bicycle=designated along with the foot=designated, it's not a "signposted only for pedestrians" type of way so "they" claim it can't be a highway=footway. Alv 18:25, 20 August 2009 (UTC)

I could be wrong, but are you perhaps confusing access=designated with access=yes? I've only ever seen that sentiment expressed about access=yes. That is, highway=path+foot=designated+bicycle=designated is equivalent to highway=footway+bicycle=designated, but not to highway=footway+bicycle=yes. --Hawke 19:28, 20 August 2009 (UTC)

Unfortunately I'm not confusing those. Alv 10:16, 21 August 2009 (UTC)

selecting all cycleways (sql)

I have learned that sorting out cycleways from OSM data is rather complicated. I have tried to build a perfect query with some help from mailing list and I am now this far: SELECT * from OSM_LINE

WHERE highway=cycleway

OR highway=footway AND (bicycle=yes OR bicycle=designated OR bicycle=permissive OR bicycle=official)

OR highway=path AND (bicycle=yes OR bicycle=designated OR bicycle=permissive OR bicycle=official)

OR highway=bridleway AND (bicycle=yes OR bicycle=designated OR bicycle=permissive OR bicycle=official)

OR highway is not null AND (cycleway is not null AND cycleway<>no AND cycleway<>opposite)

This is not working SQL query but I hope it helps to see how messy the situation is at the moment. Anybody going to use OSM cycleway data for rendering or something else should do selection about this way. I fear it will not be possible to simplify cycleway tagging. Perhaps there should be some place in a wiki for storing sample queries for selecting cycleways or footways for Mapnik/Osmarender/Kosmos rendering or just for using the light traffic data outside OSM? Perhaps they could be even country specific. JRA , 20 August 2009

That's not the way you should get all roads where bicycles are allowed. You have to do a much more thorough analysis of the access tags, which have different meanings in different countries anyway. Path without access tags would allow cyclists for example, but path with vehicle=no wouldn't, then path with vehicle=no + bicycle=yes allows them again. Imagine we're now talking about horses, which aren't considered vehicles in one country, and are in another. Anyway, a "simple" SQL query like that is not the way to proceed with this. Better would be having a preprocessor which would take the OSM data and tell for each road if cyclists are allowed. In that case your new database could simply have a bicycle value for each road. --Eimai 14:38, 20 August 2009 (UTC)

I do not believe it is reasonable to have more than three alternatives on a printed/static map I have in my mind: bicycle only, foot only, and shared for both. Otherwise the result would be too messy. I would love to be able to filter these three very common alternatives super easily from OSM data itself. But I suppose preprocessing data inside a database will suit me best. By the way, the current Mapnik stylesheet seems to look for only two alternatives when finding cycleways: <Filter>[highway] = 'cycleway' or ([highway] = 'path' and [bicycle] = 'designated')</Filter>. --JRA , 24 August 2009

You could simplify your query though: ...WHERE highway='cycleway' OR (highway IN ('footway','path','bridleway') AND bicycle IS NOT NULL AND bicycle <> 'no') OR highway IS NOT NULL and (cycleway IS NOT NULL AND cycleway NOT IN ('no','opposite')). This though, like your query, misses out on designated cycleways along regular roads. for that you'd only need: ...WHERE highway IS NOT NULL AND ((bicycle IS NOT NULL and bicycle <> 'no') OR (cycleway IS NOT NULL AND cycleway NOT IN 'no','opposite')) --Hawke 17:20, 20 August 2009 (UTC)

New values for bicycle=*

1) I don't think we can try to use highway=path for everything and put other information in other tags. Having a few different highway tags is too useful and too well established. This was defeated in the highway=path vote, doesn't have widespread support, and should be dropped. However, highway=path is being widely used for paths in fields and forests. This should be better defined, and become it's only use. Similarly access=designated should be dropped (even though it was approved), because - while it meets a need - interpreting it is too interdependent on other tags, so it is too confusing.

2) We need to figure out how to map two paths adjacent to one another. I don't think it is acceptable to micro-map a cycleway next to a footway, and say one is exclusive to cycles, the other to pedestrians. That works at z17 and for routers, but not at z13-16. Nor is it acceptable to map the cycleway, say that is exclusive, and not map the way for pedestrians at all. Nor is it practical to try to tie the ways together using a relation. The algorithm for combining adjacent ways is just going to be too complicated. Instead I think there has to be one core way that can be used at a macro level and which expresses the combined properties of all the adjacent ways. The micro structure (if you have one) has to be an elaboration of that, not a replacement. Which is a long way of saying that the Germans have to accept that the default access for a highway=cycleway (just like a road), should be foot=yes, unless access really is exclusive to cycles, and there's no adjacent way for pedestrians. See further below for an elaboration of how the micro-structure might be recorded.

3) Danish/Dutch/German cycleways are generally well-made, clearly-defined and well-established (I probably exaggerate). The rest of the world isn't like that, so the binary distinction between footway and cycleway isn't sufficient to tag accurately. So I think we should agree to restrict highway=cycleway to ways that are well-made, clearly-defined and well-established and find a different way of recording how usable routes are by bicycle. I think this would be useful for routers and specialist renderers. It's an open question whether OCM would take this on board, but if they don't, someone else will.

So, I'd propose:

• highway=path - to be used for unpaved ways (typically in fields and forests), where access by car is private or no
• highway=cycleway - to be used for paved ways, where access by car is private or no, and access by bike is well-established. And implies foot=yes
• highway=footway - to be used for paved ways, where access by car is provate or no, and access by bike is less than well-established

and

• bicycle=prevented - physically prevented (preferred to bicycle=no)
• bicycle=forbidden - notice banning cycling that's likely to be enforced (preferred to bicycle=private)
• bicycle=tolerated - illegal, but no-one enforces it, and most people turn a blind eye
• bicycle=discouraged - partial obstruction, but negotiable without lifting the bike off the ground
• bicycle=accepted - no significant obstruction, but no signage or deliberate physical improvements (preferred to bicycle=permissive)
• bicycle=supported - some signage or deliberate physical improvement, but incomplete
• bicycle=established - well-made, clearly-defined and well-established (preferred to bicycle=yes)
• bicycle=exclusive - only to be used when the pedestrian route is also defined, or when there really isn't one

In association with that, I'd probably use some of these tags to record some of the micro details: footway:left|right=yes segregation=yes|no|line bicycle:left|right=exclusive foot:left|right=exclusive or maybe designation:left|right=(country-specific values encapsulating the legal status)

--RichardMann 13:49, 6 October 2009 (UTC)

Solution (2) vs. highway=path based equivalent for highway=cycleway/footway/bridleway

In my opinion, the fuzzy description for highway=cycleway should not be kept. Using bicycle=official, which is well described on the access=* page, highway=cycleway might be the same as highway=path + bicycle=official. This seems to fit for the situation in UK an other countries. Another option might be to abandon highway=cycleway, so ways tagged in this with it should be retagged more clearly with highway=path + some access=* tags. Are there any situations that could not be tagged with the current scheme? To keep a fuzzy tag without making it deprecated is not a good option.

There should be a solution (2a) with a specific equivalent for highway=cycleway (like highway=path + bicycle=official) and (2b) where highway=cycleway is deprecated. --Plasmon 22:31, 25 October 2009 (UTC)

moved from Consolidation_footway_cycleway_path#Matrix of Pros and Cons:

• Quite unfeasible, as that's not how most of the ways for non-motorized vehicles have been tagged. (Outside of Germany and some other countries, that is.) Alv 19:34, 25 October 2009 (UTC)

  Please explain this more detailed. What's the equivalent for most of the ways? I created a section on the discussion page for this.

  There are plenty of ways that are only for non-motorized vehicles (and pedestrians), yet they don't have any signs at all, or they are signposted with a "no (motorized) vehicles allowed". Yet in effect they are no different from the ways having a (or similar), so they should get the same highway tag, as they would with such a sign. There isn't a single equivalent tag combination in that direction (as in "cycleway is always equal to tags x+y"). Alv 00:19, 26 October 2009 (UTC)

  Hmm, I understand. In my opinion each way, that is currently tagged as highway=cycleway or highway=footway, could also be tagged with the current tagging scheme as highway=path + additional tags, e.g. bicycle=official. In some cases highway=track or any other highway=* might be more suitable. So I see two solutions: Solution A: either you declare highway=cycleway and highway=footway as a combination of other tags (something like a template). This means, that ways, where this definition doesn't match, have to be retagged. So in the meanwhile, some ways are tagged wrongly. But is "wrong" much worse than "fuzzy"?  :) Solution B: Or we do not find a equivalent and keep the tags fuzzy. Then highway=cycleway and highway=footway should either be marked as deprecated (as they are not that precise than other possibilities) or as a fuzzy placeholder like highway=road. In both cases, ways currently tagged as highway=cycleway or highway=footway have to be retagged in the medium term. Any solution (either mine or any other "good" one) would cause some retagging. That's the problem of fuzzy tags. Otherwise you try squaring the circle. --Plasmon 20:06, 1 November 2009 (UTC)

  (Yes, occasionally track could be, but hardly any other highway type.) For most cases/uses it is sufficient to know a) can I walk there b) can I cycle there without any knowledge of signposts - the information that these two convey. There's no reason to deprecate them, as the more precise information can always be added with =designated (or designation=Gemeinsamer Geh- und Radweg or traffic_sign=de:241, or similar). Alv 23:33, 1 November 2009 (UTC)

  There's no reason to deprecate them, as the more precise information can always has to be added with ... other tags. In fact, this is the basic problem of those tags, because e.g. highway=cycleway does contain some fuzzy information about bicycle usage. At least, you can use it by bicycle (what might be enough information, as you wrote). But concerning pedestrians the situation is not only fuzzy but also unclear (usage by foot allowed?). Keeping things as they are as you suggest is not a solution. Another solution might be, that highway=cycleway means e.g. highway=path + bicycle=designated. If there is a blue sign, you can set bicycle=official. In addition ways without foot=* and horse=* should lead to an error in on validation. So mappers are forced to clarify the situation for pedestrians and equestrians. This might also solve conflits and is more compatible to current fuzzy usage. --Plasmon 17:30, 2 November 2009 (UTC)

  In countries that do have a significant amount of cycleways-only-for-cyclists, it's just a matter of recommending "always add foot=designated/yes/no", possibly with tools highlighting those without the tag set. Just as with path, but with which users would always have to add (at least) all of foot, horse, bicycle. IMO official offers nothing that designated + other modes set to no couldn't describe. Alv 18:08, 2 November 2009 (UTC)

  Designated: There is a sign for a specific user (something like [1], [2], [3], [4], [5], ...), means something like "particularly suitable". Official: traffic signs by law, so the blue ones. In both cases, there is no additionally information about other users! Path implies foot=yes, bicycle=yes, horse=yes. So there is in contrast to highway=cycleway etc. no unclear situation, whether other users are allowed to use the way. --Plasmon 20:12, 2 November 2009 (UTC)

  Why should those be tagged with foot/bicycle/horse=designated? They're not legally any different from other ways where the same users are allowed, and such are rather candidates for route relations. If mappers are using the values wrong, fix the errors and educate them. But it's already described that a highway=cycleway in Germany without foot=yes doesn't allow pedestrians, but in most other countries it does allow. And if it's tagged wrong, go out fix it. Alv 07:49, 3 November 2009 (UTC)

  Because it is descriped this way on the access=* page. And it might be nice to highlight them on special maps in contrast to ways where cycling is only allowed. And those ways are often not routes, only parts of an open network, like [6] and [7]. And the situation is not consistent in the wiki. So DE:Tag:highway=cycleway has a lot of subjunctivs. You can only educate people, if there are consistent rules. What you suggest for Germany is the case, that highway=cycleway should be the same as highway=path + foot=no + horse=no + bicycle=official = . That fits to my solution (2a), as stated above, but with country based definitions. But above, you rejected concrete definitions ("There isn't a single equivalent tag combination in that direction (as in "cycleway is always equal to tags x+y")."). I am confused. --Plasmon 19:11, 4 November 2009 (UTC)

  (The replies are getting longer on each round, and far indented... I'd need to make some sort of a table presentation soonish)

  Cycling is either "allowed" (just like motor vehicles are allowed on all roads) or "legally designated"; then such network markers that don't have a legal impact should have some other tags to describe them. I believe that in solving these controversies people should look at the history of descriptions, discussions, proposals and changes, that lead to the introduction of highway=path at all.

  yes was originally the only value (relevant to this discussion), that was used for "legally accessible", no matter what the sign is. When path was introduced, designated was given as the means to describe what mode of transport the highway=path is meant for, if a sign is given - if it's not a highway=footway or a highway=cycleway (as described by that day wiki pages); surely ways would stay a highway=cycleway even when they don't have a sign but it's mainly for cycling (or cycling and walking, as "use the highest transport mode" was a consensus before the path proposal).

  Hmm, there seems to be (for quite some time now), the IMO not categorically correct "implies bicycle=designated" on the Tag:highway=cycleway page. Changing the description for highway=cycleway wasn't discussed within the proposal for highway=path, even when the alternative ways to describe signposted cycleways in the UK was given as an example. Several editors have later changed several pages and guidelines have become contradictory, but likely none of them can be changed without someone reverting... A way tagged as highway=cycleway can be either bicycle=designated + foot=designated (, UK "cycleway" sign, or similar), or bicycle=yes + foot=yes ("no motor vehicles" or no sign at all). Considering that all these ways fit the original definition of cycleway (as in "it's not for motor vehicles but I can, and I'm allowed to, cycle there"), why would they be tagged as a highway=path at all? Alv 10:07, 5 November 2009 (UTC)

Unbuilt dirt tracks

The use of path for unbuilt narrow dirt tracks was struck through. I have activated it again as a considerable number of people use it in this fashion.

But my question is: If path is not used, what is the tag for an unbuilt and unmarked narrow dirt track? --Nop 18:29, 2 November 2009 (UTC)

Something like highway=path + surface=ground + width=narrow or est_width=0.5? --Plasmon 19:11, 2 November 2009 (UTC)

I do not consider this as a good idea. I would expect that many people will not add that many tags to properly characterise a path (for various reasons) and that another group of people may inappropriately use them. Whatever the solution will be, it should also be quite easy to apply. The ideal solution would be a couple of tags (maybe three or four) such that in the majority of cases one tag alone will suffice to characterise a path. In that sense highway=path without any further qualifications should denote "an unbuilt narrow dirt track". Reducing everything to highway=path with the requirement to add a couple of further tags to completely characterise a path may not improve the situation at all: people just may not use these additional tags. In that case one would not be better off that today. --Schlauchboot 12:10, 28 November 2009 (UTC)

It’s not incorrect to use it in this way, but it is incorrect to say that it can only be used for unbuilt narrow dirt trackspaths.