Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
> On Jun 15, 2016, at 2:48 PM, Simon Poole wrote: > > physical attributes > of ways Then why do we have 7 different tags for roads, and then add attributes such as width, surface and so on? Why don't we differentiate all roads in your "we already have enough" way? Highway=road Width=20m Lanes=5 Smoothness=excellent Oh! It's a motorway! Nope! Toll entrance to Tokyo Disneyland... We could just have a highway=motorway. Highway=path Surface=stone Width=1m Smoothness=horrible (or whatever value that tag uses). Is that a 400 year old stone path through a Japanese temple? Is that a rough path along a stream along a stone face? Is that a dangerous route over a pile of boulders? Who can tell? Certainly not a data provider. How is a data provider supposed to make assumptions of what a particular path is when there is no place to start from? All of the motorway-to-service values give a good general starting point to guess from. Path is a big mushy pile of mixed opinions that leads to inaccurate assumptions - which leads to inaccurate mapping, rendering, and eventually disappointed users. All of the tagging issues I encounter stem from the abundance of detailed tags in one area of OSM being used to justify the lack of need for the _exact_same_level_of_detail_ requested by mappers in different areas. Bikers, hikers,Trekkers, and park visitors are not going to benefit from the 7 qualitative tags (plus 5 track grades) when there is only 3 available to non-vehicle traffic - 3 to cover a giant manicured national mall-arcade space down to the most difficult and rough-hewn trail through boulders. Highway=pedestrian Highway=path Highway=footway (now same as path><) Where is the "track" for path? Where is the "track grades" to go with it? The lack of it is bewildering. Javbw. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
John Willis wrote: > How is a data provider supposed to make assumptions of what a particular > path is when there is no place to start from? I presume you mean "data consumer", and as the data consumer who probably parses path tags in more detail than any other (for cycle.travel), I do fine, thanks. highway=footway|cycleway|track, surface=*, tracktype=*, width=*, and access tags are plenty sufficient. There are two significant issues I encounter with path tagging and neither of them are a lack of highway values: - missing surface tags - highway=path, which should die in a fire I do wish people who aren't data consumers would stop second-guessing those who are, because they invariably miss the point by a country 1.609km. Richard -- View this message in context: http://gis.19327.n5.nabble.com/Subject-Feature-Proposal-RFC-highway-social-path-tp5870639p5875556.html Sent from the Tagging mailing list archive at Nabble.com. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
Yea, I meant data consumer. > On Jun 15, 2016, at 5:47 PM, Richard Fairhurst wrote: > > - highway=path, which should die in a fire Well, we are in agreement there. And since you are a domain expert, how does one go about separating mountain trails from footpaths in a park if their surface and width is the same? What "condition" tag do you use to separate them? Javbw. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
John Willis wrote: > how does one go about separating mountain trails from footpaths in a park http://wiki.openstreetmap.org/wiki/Key:sac_scale is popular for doing that. Richard -- View this message in context: http://gis.19327.n5.nabble.com/Subject-Feature-Proposal-RFC-highway-social-path-tp5870639p5875571.html Sent from the Tagging mailing list archive at Nabble.com. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
2016-06-15 11:58 GMT+02:00 John Willis : > And since you are a domain expert, how does one go about separating > mountain trails from footpaths in a park if their surface and width is the > same? What "condition" tag do you use to separate them? well, you can see from the data that a path / footway is in a park, because it is in a park. It's a geospatial database, and the park should be mapped as an area. Cheers, Martin ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
> On Jun 15, 2016, at 7:04 PM, Richard Fairhurst wrote: > > John Willis wrote: >> how does one go about separating mountain trails from footpaths in a park > > http://wiki.openstreetmap.org/wiki/Key:sac_scale is popular for doing that. > > Richard > On Jun 15, 2016, at 7:07 PM, Martin Koppenhoefer > wrote: > > well, you can see from the data that a path / footway is in a park, because > it is in a park. It's a geospatial database, and the park should be mapped as > an area. > > Cheers, > Martin So SAC scale and being outside a park polygon/relation is good enough to allow a data consumer and the folks over in -carto to render a "footway" and a "trail" differently and reliably enough? What happens when I have a strong mix of =pedestrian, =footway, and ="trail"? In the same park area? Why isn't having a footway=trail subtag (or something) seen as a much more reliable solution? When most of the trails will fall in the lowest tier of the SAC scale, is it merely the presence of a SAC scale tag that tells you it is a "trail?" Would we have to tag a cut-through with a SAC tag to get the way to render differently to show it's status as "below a sidewalk"? It seems to me - as a person who is a Kountry Kilometer away from being data consumer - that using a subtag or similar to let mappers tag trails and other rough footways (the "track" end of footway) is a much more straightforward and direct solution to get trails to render differently than more casual and easily traversed footways found in a city park or rose garden. I am really having trouble understanding the reasoning behind the resistance when it removes uncertainty and confusion while tagging. Javbw. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
John Willis wrote: > I am really having trouble understanding the reasoning behind the > resistance when it removes uncertainty and confusion while tagging. But it doesn't. You're citing your own personal hierarchy between "trails" and "easily traversed footways", which is fine. But that hierarchy is not ringing any bells with me. I honestly have no idea how any of the paths around here would be classified on such a hierarchy. We have thousands of miles of paths which are walkable as of legal right, of every quality from wide tarmac to barely discernible routes across ploughed fields, but we don't have any concept of "trails" here[1] - it's largely an American/Australasian English usage. highway=motorway/trunk/primary/etc. works when a firm, easily understood hierarchy can be established based on that road's importance in the connected network. It falls down when that hierarchy is less clear-cut, and it's very notable that road tagging is quite uniform and uncontested in some countries (e.g. the UK) where there's a clear mapping between tag values and observable characteristics, and less uniform in others (e.g. the US) where that mapping is fuzzier. For your idea of increasing the highway= options available to path mappers, such a hierarchy would need to be apparent on the paths in most countries, and to be documentable as such in a reasonably internationally consistent manner. I haven't yet seen any case made that it is, and I doubt that it could be. Richard [1] other than the very few long-distance routes known as National Trails -- View this message in context: http://gis.19327.n5.nabble.com/Subject-Feature-Proposal-RFC-highway-social-path-tp5870639p5875594.html Sent from the Tagging mailing list archive at Nabble.com. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
On 15 June 2016 at 13:10, John Willis wrote: > Why isn't having a footway=trail subtag (or something) seen as a much more > reliable solution? Perhaps more of an aside, but it may explain some people's reluctance / confusion with highway=trail: As a native British English speaker, the word I would use for what I think you're describing as a "trail" (i.e. a rough path through the countryside that's used as a route by walkers / hikers) is "path" or "footpath" (or maybe "track" if it was wider) -- which are the same words that I'd use for paths through a park. I would normally use the word "trail" to describe a route rather than the path itself, particularly if there was some other purpose/interest in the route (cultural, historic, fitness, wildlife) above than just walking the paths. But back on topic, if two paths have identical surface and width characteristics (which we can already tag), what difference does it make whether it's through a park or across more open countryside? Why would it matter to data consumers or renderers? Robert. -- Robert Whittaker ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
On 15/06/2016 13:10, John Willis wrote: So SAC scale and being outside a park polygon/relation is good enough to allow a data consumer and the folks over in -carto to render a "footway" and a "trail" differently and reliably enough? What happens when I have a strong mix of =pedestrian, =footway, and ="trail"? In the same park area? Before answering that, let's take a step back and have a think about what we're trying to do here: o People tag stuff so that the essential nature of what's on the ground is capture. That might be "is mainly designed for use by foot traffic" (highway=footway), "has a gravel surface" (surface=gravel) or "is part of some sort of route relation" (appropriate relation membership). Data consumers occasionally grumble about specific tagging choices, but usually if the data's there, they can somehow deal with it. o Renderers have a limited number of ways that they can render stuff without things becoming _way_ too complicated. If you look at osm-carto you can see that for linear features such as roads and tracks it uses: 1) feature width 2) feature colour 3) casing width 4) casing colour 5) How it's actually rendered (continuous line, dots, dashes, whatever) 6) Some text printed alongside the feature Those are basically the degrees of freedom you've got. "osm-carto" uses 6 for names, and chooses to use just lines for footpaths and tracks (we'll leave highway=pedestrian out as it's rendered - correctly in my view - as a road), which basically leaves you with 1, 2 and 5. Of those, the combination of 1 and 2 need to be carefully adjusted together so that a particular feature has the right level of "importance" in a particular rendering at a particular zoom level. "osm-carto" de-emphasises footways at the expense of other features; other styles (see the two examples below) emphasise them more. Of the available choices "osm-carto" currently uses 1 and 2 to distinguish between footway, bridleway, cycleway and track (and I believe puts "path" in one of the first two buckets based on other tags). It uses 5 to show paved vs unpaved (see http://www.openstreetmap.org/#map=18/53.21858/-1.49726 ). It doesn't show relation membership. Other styles do things slightly differently. Thunderforest Outdoor doesn't show the surface difference, de-emphasises roads but does show relation membership: https://c.tile.thunderforest.com/outdoors/15/16247/10641.png I use a style based on OSM-carto that doesn't show surface explicitly on these features (or relation membership) but does show England and Wales legal status (using colour) and track width (using dashes instead of dots): http://i.imgur.com/sjytRiy.png So data consumers have choices, and there's a limit to the differences in the data that they can render. If someone asked me 'to render a "footway" and a "trail" differently' my first question would be, apart from tags that I already understand such as surface, tracktype, smoothness, etc., what's the difference? In my view if there's a muddy flat 1m-wide path in a city centre and a muddy flat 1m-wide path miles away, then they should be rendered the same - apart from relationship membership of course, if a style renders that. It seems to me - as a person who is a Kountry Kilometer away from being data consumer - that using a subtag or similar to let mappers tag trails and other rough footways (the "track" end of footway) is a much more straightforward and direct solution to get trails to render differently than more casual and easily traversed footways found in a city park or rose garden. I am really having trouble understanding the reasoning behind the resistance when it removes uncertainty and confusion while tagging. Clearly there is confusion about tagging these sorts of features (this disussion wouldn't exist if there wasn't) but I'm really struggling to see any difference between a "footway" and a "trail" that can't be expressed in other, more frequently used tags. To take a random example, what should I infer about http://www.openstreetmap.org/way/372458198 ? How wide is it? What surface does it have? What are the access rights for various sorts of traffic? Cheers, Andy ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
> On Jun 15, 2016, at 9:56 PM, Robert Whittaker (OSM lists) > wrote: > > if two paths have identical surface and width > characteristics The issue I have is that they do not have similar characteristics, yet get rendered the same. It's like if all tracks were rendered as residential roads. Sure, there are residential roads bordering on tracks, but easily confusing them would be a nightmare. If they did that in OSM, a user would delete any App based on OSM in 3 minutes in Japan. Being tricked by a map into thinking you have a real "city road" to drive on but are instead directed to a muddy 1.2m wide hole through the bamboo or some double-track along the top of a berm in a farming field would possibly get me stuck in the mud for hours. The exact same thing is true for "foot path" and "trail" - the assumed usage for a sidewalk and a "trail" are so different! ~~ Long story ahead, not vitally important. Speaking only through my experience, "trails", paved city walkways, and rural tracks are very different. The hundred or so old paper maps I have used - all old park maps and trail guides from many sources - all these maps reliably and consistently separate such types of routes. Ones that don't are less useful and are forgotten. OSM can easily show this information because it can show the nuanced detail (3 grades of roads in GoogleMaps vs a lot in OSM-carto). My ex girlfriend in a wheelchair made me look upon our sidewalk system in the US differently. Having a mother with numb toes from chemotherapy means any balancing out on uneven surfaces is dangerous. Coming to Japan and seeing a much larger - and much more drastic (and dangerous) separation between (boggy) trails and concrete walkways through a garden - and their insane incompleteness and carelessness in their creation and maintenance of regular footways means carefully mapping where sidewalks disappear abruptly into a wall and force people into narrow shoulders on trunk roads. all of this requires attention and consideration greater than the rendering the -carto is rendering now *if your goal is to navigate via looking at a rendered map, which o still want to do. It is weird to say, but some places can be defined with less tag values. I tell All my Japanese friends that everything in America can have a "wide variance" - safety, food, cities, poverty, neighborhoods, etc compared to Japanese ones - but Japan has a wider variety of physical conditions mixed all together in very close proximity - roads drastically narrow, infrastructure piled on top of each other, land uses radically changing every 25 meters, it is a crazy complicated place. I know there are other places in the world probably even more complicated - but being able to define those differences through tagging (and rendering) seems to be a way to make a batter map. Local data providers already provide such detail because it is what has to be shown to let you properly imagine the conditions that exist. All of these things maybe can be replicated by combinations of tags about width and surface and wheelchair=* - but there is something so profoundly easy and understandable buy saying "this here is a city walkway" and "this here is a rough path" - they both imply so much. I understand that that must vary by country - but so does every country's implementation of "what is a primary road" - but that has not given rise to lumping all =trunk through =track roads into the same group. Javbw. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
On 15/06/2016 15:03, John Willis wrote: On Jun 15, 2016, at 9:56 PM, Robert Whittaker (OSM lists) wrote: if two paths have identical surface and width characteristics The issue I have is that they do not have similar characteristics, yet get rendered the same. So tag the different characteristics (surface, width, etc.), and let renderers decide whether to render the difference or not? I have to say I'm really struggling to see the problem here. Cheers, Andy ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Subject: Feature Proposal - RFC - highway=social_path
> On Jun 15, 2016, at 11:11 PM, Andy Townsend wrote: > > So tag the different characteristics (surface, width, etc.), and let > renderers decide whether to render the difference or not? > > I have to say I'm really struggling to see the problem here. Hmm.. Why is a tertiary rendered differently from residential? Is the road itself fundamentally different? Not really. Its rendering shows us what to expect, in the broadest possible terms. Why is a cycleway rendered differently? Is it reflecting its legal status or its assumed condition that it is made for bicycles? Would you expect to find random staircases in a cycle path? People have different expectations of conditions based on its classification. Would you expect to find rocks the size of footballs jutting out of a sidewalk? For it to suddenly vary in width wildly and randomly for its entire length? Perhaps a section where you have to balance on a log to cross a difficult bit to get into a grocery store? Horse manure on your jogging track around the school? Ankle deep mud on your walk between the parking lot isle and the foot court in the mall? People would complain bitterly about whoever made the path in the rose gardens have a spot where you swing your legs over an onerous Boulder in the middle they couldn't dig a trail over. All of these would be expected and routine conditions found on all but the smoothest and most well maintained of trails - and expected by users of trails. Conversely, 4cm uneven pavement can knock an old lady over. I would never have to look for blazes, spray painted markings, or ducks to make sure I am still on the walkway through a city park around a playground. All of these separate things can be defined - like the length of my fingers or the size of my feet - but none of them really define what species it is. Concrete walkways through a rose garden and a backcountry trail are different species of foot path. They may occasionally share some characteristics - but they are different as a trunk road to a residential road - massively different expectations. The cut-through in a field made by joggers along a fence may be pretty flat and easy to walk, but they are not the same species as a sidewalk or compacted path through a temple. I don't know what else to say, I have ran out of examples and analogies to illustrate my opinion. Javbw. ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
