I would say that regardless of the presence of segregated=yes, the presence of a truthy sidewalk value means that cycleway:surface and footway:surface are incorrect and should be surface and sidewalk:surface instead. A way with a sidewalk implies segregated=yes. Only without a sidewalk (or negating sidewalk values) and segregated=yes should it be considered a combined path.
3 Likes
The problem I can see with that is the highway=primary/secondary etc. does NOT indicate who can access the road, therefore sidewalk=* makes sense. On the other hand highway=cycleway does indicate that the road is specifically designed for bicycles. If there is an infrastructure for pedestrians, is it still a road designed for bicycles only?
So in my opinion, even though thisv s signed as cycleway, this should be tagged as highway=path in “osm-like sense”, as this is not really a cycleway if an infrastructure for pedestrians is present. If this is possible then, technically, motoroways can have sidewalks too?
EDIT as I just noticed that. Does the third option even work in navigations? If there is no foot=yes on the line, navigations would assume that the road is for bicycles only, sidewalk=* does not indicate access.
Either the cycleway is segregated or it has a sidewalk, the tags ought to be mutually exclusive. If it has a sidewalk, then StreetComplete should be tagging surface=* and sidewalk:surface=* rather than cycleway:surface=* and footway:surface=*
I think any mapping activity should have a practical use: we are mapping for the map user, and if they’re not interested in the data we provide, we shouldn’t collect it just for the sake of collecting data.
If there is no legal reason to map Dutch-style cycle&footways differently, then we should look for another practical reason to map them differently. One such reason could be that the kerb separating the cycleway from the footway is so high that it is an obstacle for wheelchair users. I am most familiar with the cycling network in Lelystad (where I grew up); there the kerbs are usually very low so they don’t form a barrier for wheelchair users (this one, for instance https://maps.app.goo.gl/2LqV4sdiAf4QfgdBA ). We could use the"Dutch" tagging scheme to express that there is a kerb between the cycleway and the footway, and that it is high enough (>3 cm, see Key:kerb - OpenStreetMap Wiki ) to be a significant barrier for wheelchairs. We could recommend that the situation you posted above should be tagged according to 3. in my initial post, and the situation I posted in this post should be tagged according to 1. or 2. This would create a practical reason to differentiate between the tagging schemes. That is assuming that wheelchair users do sometimes need to cross these kerbs (can’t think of good examples now)… Edit: situations like this maybe? https://maps.app.goo.gl/9XbZvohJfNadLrEB8
Semantically a sidewalk implies segregation of the two.
Yes, regardless of the presence or absence of segregated=yes.
highway=primary: general road for road users (cars, bicycles, etc.), may have sidewalks.highway=cycleway: way for bicycles (and depending on the tagging,mofa,moped), may have sidewalks.
Both allow pedestrians, unless otherwise excluded, and in the case of the Netherlands use of the sidewalk is mandatory if present.
Yes, a dedicated cycleway with a sidewalk alongside it, is still a dedicated cycleway.
See the photo I posted. The sidewalk there is, in OSM terms, an attribute of the cycleway, not part of it. This affects the placement of the mapped ways. Sidewalks often have lowered kerbs where they cross the cycleway they are built alongside, which is another distinguishing feature from a combined foot/cycleway. Legally the sidewalk of a cycleway is also distinct (RVV 4.1, 4.2), but that is not relevant here.
The distinction lies in whether the sidewalk is considered a section of the cycleway (which holds true for combined cycleways/footways) or an attribute of the cycleway like it is for streets. This distinction can be subtle and may even be country-specific, but for most Dutch cycleways falls squarely in the ‘attribute’ category due to the high level of segregation of traffic flows where needed.
4 Likes
I think this part was missed as I added it later, but are additional access tags added to those cycleways? If there are not, they are tagged as cycleways with sidewalks that noone can use, which doesn’t make sense (highway=cycleway indicate it is only for bicycles unless addidtional access tags are added).
If there are access tags, then 2nd and 3rd method are the same, only one use segregated=* and one sidewalk=*. We can simply just use one of them in case.
And unless the national default access values say otherwise. This is addressed here. A cycleway in the Netherlands tagged with only highway=cycleway will route pedestrians unless the router completely ignores the documented defaults. There are several countries where the default foot value is yes (or another permissive value) for highway=cycleway.
Again, wrong. For method 3 the centre line is the exact middle (or the centre line if present) of the cycleway surface, exlcuding the sidewalk. For method 2 it is the middle of the combined footway and cycleway.
Method 3, sidewalk as attribute of cycleway (branching off and becoming a separate footway on the right)
For pedestrians (in The Netherlands) it is important to know if the sidewalk is on the left or right of the bike path or on both sides. For example, a bike path coming from the left or right side has implications like bikes having priority and crossing the curb. So only the tag segregated doesn’t cut it if you want good quality maps or navigation for pedestrians.
That said, in combined bike/foot paths there is not really a sidewalk. It is one path where one part is used by bicycles and another for pedestrians. I guess pedestrians always walk on the right side. Otherwise a tag would be needed to indicate where they walk.
That’s correct. Then 2nd and 3rd method show the same thing. Cycleway with segregated part for pedestrians and cycleway with sidewalk are the same thing. The lane would be in different place but it doesn’t matter - the lane is just a visualition for routing. Lanes are not always in the middle anyway. Again, different method to show the same thing.
I am not telling which method is correct. Maybe we should re-tag every segregated=* to sidewalk=* if the other is more specific. But I see the problem with many ways to show the same on the map.
Of course that matters! We don’t just draw ways in OSM at random. 
1 Like
Does not matter, or does not matter to you?
For a detailed renderer it is quite usefull to have the sidewalk on the actual side that it is on.
Does not matter in the case that this is our conception we agreed on. If we change tagging, we can also change the position of the lane. Or do not if we decide so. With roads we sometimes do not draw a lane in the middle, eg. if right lane is used for entering the highway.
What I meant is that the lane is a simple way to show the road. While it’s important to use the same rule the same time, it’s not something that we can’t change or make an exception.
This is something we do everyday. In any other cases, the existence os sidewalk=* tag does not change the position of eg. highway=primary. What’s more, if the sidewalk if draw as the different lane, this doesn’t also affect the position of highway=primary as well. Using the same logic, if we use highway=cycleway we can still draw the lane in the middle of the cycleway, regardless of sidewalk=, segregated= or none of them.
It’s actually very good practice to walk on the left side of the road. As this way you are facing the traffic that is closest to you. Walking on the right side of the road, you don’t see the traffic when it hits you. (of course in the UK this is the other way around)
Segregated merely says that a pedestrian should not use the cycleway to walk on it. They could be using a sidewalk (as an attribute to the cyclway); but they could also use a seperate path that is paralel to the cycleway. This seperate path might or migth not be explicitly drawn.
Drawing the sidewalk as a different path; I would recommend against; as it holds very little value and clutters the map.
There is a placement tag for that actually.
This is problematic because highway=cycleway + foot=designated + segregated=yes (and variants) is documented as and is interpreted as “segregated foot- and cycleway” by software, with everything it implies.
It implies that it is two features drawn as one way (footway and cycleway), and hence, to refer to properties of either one of them, one uses the cycleway:* and footway:* prefixes, e.g. for surface or width:
https://wiki.openstreetmap.org/wiki/Tag:highway%3Dcycleway#Examples
In other words, there is no “main feature” anymore, to which tags without prefix would apply to. E.g. surface in the above picture would apply to the whole way, in which case it is neither fully asphalt nor paving stones, but at least paved.
The tagging situation is different for cycleways with sidewalks (highway=cycleway + sidewalk=yes). Even though the actual physical difference between the above and this tagging may be just the presence of the curb, referring to properties of the cycleway and the sidewalk is different, e.g.
highway=residential
surface=asphalt <- refers to road
sidewalk=right
sidewalk:right:surface=paving_stones <- refers to sidewalk
Now, if you mix the two segregated=yes + sidewalk=yes, you end up with foot- and cycleways that additionally have a sidewalk and this is certainly not what you wanted to express, but it is interpreted as such.
3 Likes
Only for cycleways where the segregation actually is a sidewalk.
But this example is a very German road design. Even the wiki is ambiguous is it a path where bikes are designated or is it a cycleway where foot is designated. 
In The Netherlands these don’t exist. Either it:
- is a cyleway without any space for pedestrians; but they are allowed to walk on the cyleway (foot=yes; segregated=no);
- there is a sidewalk (foot=yes; sidewalk=both|left|right; segregated=yes)
- There is a seperate path that follows the cycleway:
- If seperately drawn: foot=use_sidepath
- If not drawn: foot=yes, segregated=yes
In general it is a very bad idea wanting that tags are interpreted differently in different regions.
Hm, anyway, I am not sure if you got my point. cycleways with sidewalks are fine, I don’t care. It is the combination of sidewalk and segregated that is causing problems. I described this in the second part of my post.
2 Likes
I interpret segregated=yes + sidewalk=yes* exactly as sidewalk=yes* alone.
segregated=yes is redundant in this case.
- or a value that amounts to yes
The problem is that there is a gap in the tag’s definition. These mutually exclusive statements are both applicable according to who you ask or what you refer to:
- A:
segregatedshould only be used for the main ‘carriageway’ of a shared foot/cyclepath, and has nothing to do with any sidewalk tagged withsidewalk. - B: The presence of a
sidewalknext to a cycleway, impliessegregated=yes
Given a cycleway with only a sidewalk and the presumption that all pedestrians must use that instead of the cycleway surface:
Position A means that this is valid:
highway=cycleway
foot=yes
sidewalk=left
segregated=no
Position B means that this is valid:
highway=cycleway
foot=yes
sidewalk=left
segregated=yes
This stems from the tag’s use and promotion over the years. See for example this comment in JOSM’s issue tracker.
To be clear: no one is to blame here, but to resolve this segregated would have to be (re)defined as either strictly pertaining to the (shared) ‘carriageway’ of a (shared) cycleway, or as meaning ‘pedestrians can follow this OSM way segregated from the rest of the traffic on it’. I wouldn’t mind seeing a proposal with two vote ledgers on this if you really want a clear answer.
It doesn’t help that as a tag name, segregated is really too generic, which tends to lead to broader use than envisioned. If you were to ask someone unfamiliar with the project what segregated=yes means and only tell them that is has to do with pedestrians, position B is likely to be what they consider it to represent.
4 Likes