Consuming highway=path

n76 · 2024-08-29T04:11:39.455Z

I wish I could be as brief as @tekim

Basically OSM tagging is not clean when it comes to trails. I wrote up some thoughts on that over four years ago when I was trying to improve my hiking maps.

Since I wrote that some have proposed highway=scramble to help in some situations but I don’t think that has gained consensus yet.

At the very least a surface=* tag ought to be put on any highway=footway or highway=path. Additional tags like width, mtb_scale, sac_scale, visibility, etc. all can help a data consumer muddle through and give better routing for the specified traffic (foot, wheelchair, mountain bike, etc.). Even if you are not an avid hiker or mountain biker you can probably make an estimate of the width, what the surface is.

alan_gr · 2024-08-29T07:18:44.977Z

Hungerburg:

but e.g. Minutely updated vector tiles demo goes the same way.

I agree with your points in general, but bringing the vector tiles demo into the discussion may be a distraction. Currently it renders highway=footway the same as highway=path, as a relatively wide white line. So I don’t think the specific problems of highway=path are the issue here.

Peter_Elderson · 2024-08-29T07:20:10.719Z

highway=path is used for a wide range of features.
All are slow connections for foot, many also for bicycle and/or horse, some also for small motorized vehicles.

Differentiating is done using attribute tags for access and for physical characteristics.

I think it is important to agree on, and explicitly describe, the (world wide) defaults, not only for access but also for the physical characteristics. To be used for rendering, routing, and issue-detecting QA-tools.

Mappers will know what will apply when only highway=path is tagged. Everything that is significantly different from the defaults, should be tagged with extra tags.
Especially when people state “but in our country/region/terrain, paths are always/legally/practically …, so that is the default and we don’t tag it”, that’s when it should be tagged explicitly, because it differs from the world wide OSM defaults.

Woazboat · 2024-08-29T09:39:35.495Z

It also doesn’t render highway=track at all yet because it’s just a tech demo, so whatever it does is irrelevant to this discussion.

Hungerburg · 2024-08-29T09:59:18.882Z

tekim:

Yes, we need to do better.

I did start this to gather ideas, how to do better.

alan_gr:

bringing the vector tiles demo into the discussion may be a distraction

It is just exemplary, consumers start with an assumption, streets.gl is another. This then stays forever. Mapbox outdoor, Maptiler outdoor – all paths the same in them. They are not new.

whb · 2024-08-29T10:41:11.465Z

Woazboat:

It also doesn’t render highway=track at all yet because it’s just a tech demo

highway=track are rendered, but many are missing because the import was buggy at the beginning. However, modified or newly created highway=track should be displayed:

Minutely updated vector tiles demo General talk

It has been fixed and will start appearing. I’m not immediately doing a reload.

SomeoneElse · 2024-08-29T11:18:37.473Z

Hungerburg:

It is just exemplary, consumers start with an assumption, streets.gl is another. This then stays forever. Mapbox outdoor, Maptiler outdoor – all paths the same in them. They are not new.

I don’t think that this is an example of that. It’s mostly a tech demo of “minutely updated vector tles” (which is “the hard bit”). It’s not supposed to be a tech demo of an OSM schema and a display style (very much the “easy bit” in comparison).

If you want to have a look at the schema and display style used in this demo, have a look at a diary entry I wrote a while back. That doesn’t describe minutely updated vector tiles, but instead is using Tilemaker to create a static set of shortbread-schema tiles.

It’s written a bit like the switch2osm guides so that hopefully other people will be able to follow what’s written there, and it’d be really useful if people did experiment - they’d then have experience of what some of the trade-offs are, such as “more tags, bigger tiles”.

Hungerburg · 2024-08-29T16:14:20.349Z

alan_gr:

demo

Woazboat:

demo

SomeoneElse:

demo

Others consider the demo much more ambitious

Busways not rendered on main openstreetmap-carto for years, PRs are ready but not accepted General talk

Paul Norman is developing a vector tile-based system to replace Mapnik based rendering, which will make OSM-Carto obsolete as well.

Apart from that, the highway=path documentation shows a dirt path for ages, and consumers still render any old path as paved - The “demo” mimics mapbox style, in my eyes. The line with casing can only be understood as representing something constructed.

BTW OSM-Carto has three signatures for pedestrian paths. They are based on surface: paved, unpaved, unspecified. Differences are hard to spot though.

Curiously, americanamap does not show any pedestrian or cycling ways, nor tracks at all.

A big thank you to @n76 – great primer on the intricacies of consuming pedestrian information.

In the area where I map, footway is reserved for mostly urban municipal infrastructure while back-country it is all path. But as @alan_gr notes this does not hold for all of the world. Title of topic needs change?

alan_gr · 2024-08-29T16:35:28.068Z

Hungerburg:

In the area where I map, footway is reserved for mostly urban municipal infrastructure while back-country it is all path. But as @alan_gr notes this does not hold for all of the world. Title of topic needs change?

Did I note that somewhere? In this topic I merely mentioned that the vector demo map doesn’t distinguish between path and footway, which doesn’t seem to be the same point.

As it happens though, where I map it is much the same as what you describe. Even though the wiki documentation for footway doesn’t really confine it to urban infrastructure, that seems to have become the convention in many places (I am unsure if there are regions that take a systematically different approach). Perhaps this happened partly because none of the photos in the footway wiki show rural situations, or because footway is often linked to signage (blue circles).

Generally when I map highway=path myself there are no signs at all and it’s hard to know if the path is intended for, and used by, road bicycles, MTB, and/or horses. So I use a vague tag to map vagueness in the real world. (I do try to add surface and other tags though).

(Incidentally you have quoted three of us as using the word “demo”, but you were the first person to use the word - I literally copied it from your original post).

SomeoneElse · 2024-08-29T17:25:54.473Z

It sounds like you’re assuming that what that tech demo shows (minutely updated vector tiles in Shortbread schema, interpreted on a web page by Versatiles Colorful) is what “vector tiles on osm.org” would look like.

I don’t think that anyone has ever said that, and I’d be surprised if that was the case. Exactly what is planned I don’t know, and that’s a matter for (I suspect) the OSMF and the EWG with regard to what money is made available for what work.

Peter_Elderson · 2024-08-29T18:50:15.939Z

alan_gr:

Even though the wiki documentation for footway doesn’t really confine it to urban infrastructure, that seems to have become the convention in many places (I am unsure if there are regions that take a systematically different approach).

Nederland has many non-urban footways, e.g. in managed nature reserves and in private areas open to the public. Most of these foot paths have a footway sign or comparable, some are only recognizable as footways because you pass a kerb or a fence/gate that is intended to let only pedestrians pass. (Usually they grossly underestimate cyclist’s skills to pass these “barriers” … been there, done that). Alternatively, one might see Nederland as one middle-sized urban area; then the “confined to urban infrastructure” rule of thumb holds up.

Not very relevant to consuming highway=path, I’m afraid.

Hungerburg · 2024-08-29T19:06:49.662Z

alan_gr:

I literally copied it from your original post

I linked to the forum topic, and the term is in the title. There is no emoticon “blush” available, so I put up a bright smile. While demo still, held with high hopes by some.

If I did misquote you on the footways, please apologise. I was thinking, you are right, this not only applies to path, this applies to footway just the same. If I remember correctly, in Australia footway is used outside urban sphere, a bit of overpass did yield Way: ‪South Coast Murramarang Coastal Track‬ (‪696854537‬) | OpenStreetMap

Hungerburg · 2024-08-29T19:37:35.572Z

tekim:

I would definitely map what is pictured as highway=path, with the appropriate other tags.

I looked up how single-trails are mapped here: path plus mtb tags. And yes, bicycle=designated - on the official ones.

All three bicycle routers available on the openstreetmap.org website route cycles over there. Some more, some less.

This does not mean, the detail/niche tags are of no use, e.g. cycle.travel does not route there. At least somebody we all know has spent the necessary effort in consuming highway=path.

ezekielf · 2024-08-29T20:51:46.209Z

Here are two foot paths. One is in a very rural area. The other is in a park in middle of a very dense urban area. Does that make one highway=footway and the other highway=path? Can you tell which one is which just from the images?

Hungerburg · 2024-08-29T21:57:09.924Z

My first thought - make this a poll. On the other hand, it is way too easy, the question gives away the answer. On yet another hand, perhaps a third alternative, namely track might win the rural location.

My money quote from @n76, and that is where I come in, and perhaps @Dmytro_Klymenko

I am helped in this case because mappers are not actively trying to fool my filter: They are simply as confused as I am.

alan_gr · 2024-08-29T22:25:06.287Z

Hungerburg:

it is way too easy,

I’m glad it’s easy for somebody. It’s not at all obvious to me (at least if there are no signs). The one thing I would say is that the urban/rural distinction is probably not
very important. When I refer to urban/rural in this context, it’s a short cut for certain kinds of ways found mostly in population centres and other kinds of ways found more commonly outside built-up areas. It doesn’t mean that I see the simple fact of being urban or rural as changing the classification.

ezekielf · 2024-08-30T00:52:28.865Z

alan_gr:

I would say is that the urban/rural distinction is probably not
very important

Yes. This is the point of my previous post. Both examples could be tagged as either highway=footway or highway=path + foot=designated|yes. The meaning is the same: “a path for walking”. Nothing more.

Peter_Elderson · 2024-08-30T07:07:31.837Z

ezekielf:

Can you tell which one is which just from the images?

From the images you can’t even say they are foot paths.

bagis · 2024-08-30T10:04:29.265Z

Recently I was out using StreetComplete and it asked me for the surface of “the path”. I was very confused as this way was clearly signed as designated for bicycle and foot. Further investigation led to learning that this is how the JOSM preset work.

The preset “Dedicated cycleway” will make a highway=cycleway.
The preset “Dedicated footway” will make a highway=footway.
The presets “Combined foot- and cycleway” and “Segregated foot- and cycleway” will make a highway=path + foot=designated + bicycle=designated.

For me highway=cycleway and highway=footway just adds confusion, deprecating them would direct mappers and consumers towards handling highway=path together with other tags. But I might be ignorant of some important detail, and double checking my ignorance made me find the page on Duck tagging.

Peter_Elderson · 2024-08-30T10:56:25.296Z

bagis:

signed as designated for bicycle and foot

So, the path is not specifically a cycleway (path for bicycles) and not specifically a footway (path for pedestrians), it’s a combined path, and StreetComplete wants to know the surface. I don’t see the problem here?