It’s not so easy to separate function from physical characteristics (see highway=steps).
I think that’s why the discussion seems circular. People are generally in agreement about the principle that the highway= tag should reflect function or purpose. But there is disagreement about whether the function varies depending on the type of vehicle that can use the way.
If highway=residential generally represents streets where people can drive a standard size car to houses, does a street that is only for smaller vehicles have the same function? If highway=track generally indicates ways used by cars or tractors, do ways used for argiculture, but only by motorcycles, have the same function?
There is a kind of breakpoint in our highway tagging where the idea of classifying by function breaks down. The top level tag does not tell us whether a cycleway is for accessing local houses or cycling between cities. It does not tell us whether a footway is for walking between houses or across a mountain pass. The highway hierarchy from residential and service upwards is implicitly defined by reference to motor vehicles. Because it is implicit, it has perhaps never really been made clear what kind of motor vehicle: standard size cars, or any motor vehicle no matter how small?
But that’s not our definition of highway=residential. Our definition simply is:
The highway=residential tag is used on roads that provide access to, or within, residential areas but which are not normally used as through routes. Most traffic on a residential road will be for the access to, or from, residential properties, but may have also some transit traffic.
Nothing about whether a normal sized car can use it or not.
But something must be missing from that definition. Taken literally, a cycleway or footway that provides access to, or within, residential areas would be highway=residential. But I don’t think that in any way matches the reality. That’s why I think there are implicit assumptions behind highway tagging, and I don’t know if those assumptions relate to cars or to any kind of motor vehicle.
To be clear, are you suggesting that:
a way with the principle function of giving access to houses for pedestrians is highway=footway
a way with the function of providing access to houses for cyclists (and possibly also pedestrians) is highway=cycleway
a way with the function of providing access to houses for any kind of motor vehicle, no matter how small, is highway=residential?
Yes by definition if their function (or primary/intended use) is to provide access to points within farmland or managed forests, or similar areas like those used for outdoor recreation. In most places, highway=track is assumed to be public (access=yes).
I would add the common case of variable width ways, typically one or more narrow sections interspersed with wider sections. service=alley is used in many such cases today.
Arguing for this equals arguing against distinguishing cases #2 and #3 in highway classification. But the way service=alley has been used by various mappers is already an attempt at distinguishing at least case #2.
I think paths like these would be ok for agricultural motorcycles, but not for agricultural 4-wheel vehicles. Would it be a case of defining a highway=track exception for countries where agricultural patterns are different? Would this change anything in countries where this is not the usual pattern?
I expect that for the authors of this wiki text, simply using the word “road” heavily implies that a normal sized car can use the way. It probably seemed so obvious that it wasn’t necessary to state explicitly. The word “road” is sometimes used for ways that standard cars can’t use (old road, abandoned road, etc) but this is usually qualified with a second descriptor word to make this clear. Perhaps this is a cultural assumption, and the word “road” is not interpreted this way worldwide. If that is the case the wiki should certainly be clarified with more explicit statements.
That’s mainly due in the western countries, where this definition is coming from, agriculture is carried out with two-track vehicles and in those areas it’s an easy differentiation with path.
I think the golden rule of OSM is “any absolute statement is always wrong”**. The raster map style I look after nudges wide paths and narrow tracks down a category, and also tries to do a decent job with unpaved roads. I currently just host maps of Ireland and the UK, but it’d be pretty straightforward for someone to host Thailand or somewhere else. There’s also a Garmin map style with similar logic in it.
I suspect that there are other map styles that incorporate this sort of logic too.
** yes, I know that that is also an absolute statement
My point is rather, if highway=residential and highway=unclassified must be wider than a certain threshold, we need a replacement highway-class for ways with the same function but having a width below the threshold. AND all renderer need to support all those new types ( just hinting to the success-story of highway=busway ).
Isn’t it easier to convince the renderer to simply evaluate width? Since with all our existing tags we can describe all those situations.
Anything with dual rear wheels should be considered the widest vehicle that can enter a service alley. This would include a production pickup and box or good truck. Both are about 6"+ wide with limited turning radus. Anything larger would have difficulty navigating the effective lane and half wide road.
If we request support for width=*, then it will be a fixed width threshold everywhere on the globe. I expect developers will immediately refuse, as this would lead to many requests for adjusting the threshold. If we request support for a new highway type, local communities can adapt the width threshold to local conditions and can refine the use case with additional rules (track width, total way width, presence of certain obstacles, etc.).
I guess this thead was the cause of the service=driveway change by @Duja in the Wiki removing the access to buildings from the usage for service, which i dispute:
It was only intended as a cleanup and simplification, not as a change in meaning. The original used to read:
The {{tag|highway|service}} [[tag]] denotes ways used for vehicle access to a building, service station, […] It is also used for access to parking, driveways, and alleys.
The whole second sentence was just mangled. Driveways and alleys are by definition used for vehicle access to a building. So why was there an “also”? It can be easily read as “access to driveway” which is confusing; and not many L2 English speakers are instantly aware of meanings of driveway and alley. So I moved “parking lot” to the first sentence and got rid of that one.
).