[Tagging] highway=footway - Advanced definition: Distinction footway vs path

Ilpo Järvinen ilpo.jarvinen at helsinki.fi
Thu Aug 6 23:15:36 UTC 2015

On Wed, 5 Aug 2015, Greg Troxel wrote:

> Martin Koppenhoefer <dieterdreist at gmail.com> writes:
> >> Then what is the point of having path and all these other tags that 
> >> overlap?
> >
> > because path and bicycle=designated is the same as highway =cycleway
> >
> > path with horse=designated is the same as highway =bridleway 
> >
> > and you can also make combinations without having to decide for
> > footway, cycleway or bridleway. Also, without any further access tags,
> > path is neutral and open to all unmotorized means of transport (unlike
> > footway, cycleway etc.)
> I agree with what Martin said.   I also agree with previous commenters
> that redefining semantics of path or adding footpath would cause a lot
> of problems.  A few problems and random comments.
>   There's no clear default for path, footway or every cycleway about
>   surface.  The obvious answer is that surface tags should be used.
>   Similarly, a width tag should be used.  Or perhaps some tag that
>   indicates a forest trail.   But this shoudl be extra, so that data
>   consumers that don't know about it will still know that
>   foot/bike/etc. traffic works.

But that's actually exactly the dangerous approach. There's no guarantee 
that such forest trails will even be safe to use (especially on 
mountaneous terrain). It would be much more conservative approach to not 
render nor use them without intention. In theory sac_scale and various 
other tags could be used to distinguish them but it would be much better 
to have a highway type which has built-in expectation that various 
hazards/obstacles to use are more likely to exists compared to what is 
likely to occur on man-made highway=footway/path (I suppose somebody comes 
up a counterexample from a signposted footway which is why I used 
"likely". It's enough to understand that no matter what, hazards will 
always be possible and we should not try to entirely remove map user 
brains from the loop like GPS driving instructions occassionally tend to 
do to some people :-)).

In addition, giving equal prominance to forest trails and other footways 
depending on surface alone as proposed is significantly less useful to 
both map users and also to mappers (no hiking map or like is magically 
going to fix this as it's not the main source of mapper visual feedback).

The map user (I refer to a human here, not to somebody using the data) 
perspective should be pretty obvious, if both look the same there's no way 
to know that one of the equal looking "unpaved paths" is actually built-up 
recretional route whereas the others are just tiny, some even faintly 
visible, forest trails. In theory this prominance problem might be solved 
by informal=yes but in practice I expect at least the mapnik stylesheet 
guys to stonewall on this because of the extra data column that will be 
needed to make them less prominant (however, I don't endorse informal as
solution as I think a new highway=* value or =trail would be much better 
for this but one the same time some solution would be better than no 
solution). Pointing map users to "hiking maps" is not going to work since 
those people who want to use the non-trail ways are not going to identify 
themselves to need such alternative presentation when they only want to 
identify the safe routes, not the serious hiking information.

>From the mapper point of view the default stylesheets are supposed to 
help mapping (the visual feedback thing) and therefore IMHO it should try 
to provide means to distinguish details we are highly interested in, that 
is, access and surface at minimum. I don't know if such style becomes too 
messy to be implementable, however, I think that the lack of visual 
feedback on all these currently both in default mapnik and editors is one 
of the major reasons why people are not realizing where they need to 
be added. In addition, the forest/informal trails should be somehow given 
less prominance to help both mappers and map users to identify them 
visually (the former need this information to spot mistakes in order to 
correct them).

I know that there will be some (rather vocal) people who oppose any idea 
that there's need to such prominance distinction (or even that it 
would/should matter to anyone) but it's hard to understand as those people 
tend to describe their point of view as that only access tags matter which 
(to me) implies that shouldn't matter at all what the highway=* value is.

>   The default rendering is problematic in two ways; 
>     path is much heavier than footway/cycleway/bridleway, which are
>     similar except for color.

On this I agree, especially if many forest trails are mapped, see eg.:

>     I believe that a lot of the footway/path angst would go away if path
>     stopped looking like a higher class of road.

Also less prominance to forest/informal trails are needed, as they are 
more messy when mapped with high density and the distinction is 
meaningful to many map users (regardless of some people admiting it or 

>   For all of these, some notion of hierarchy is needed.  For roads, we
>   have primary/secondary/etc., which has its own issues.  For footways
>   as an example, we have sidewalks that are not particularly interesting
>   except at high zoom levels.  But a foot path that goes 3 km through a
>   forest is interesting when you can see the whole forest, just like a
>   through road.  This is not easily determinable automatically, because
>   the 10 km through path that junctions with a sidewalk does not make
>   the sidewalk important.  Importance is determined by the way being
>   useful for a long hiking route.  So probably some sort of importance
>   tag is needed.  One approach would be a tag distance=X, where X is a
>   distance that one could reasonably travel where that way would be
>   naturally a component.  Perhaps X should be rounded to 1/2/5 x 10^k m.
>   I don't really like this suggestion, but I think we need something
>   like it.

You seem to admit that there's need for some hierarchy, however, on the 
same time you seem to oppose the idea that such hierarcy would exists 
based on physical properties (man-made vs informal). I find it strange 
since it shouldn't be that hard to come up use cases where excluding or 
warning the user about informal paths is very useful thing. In order to 
make mappers to tag that differentiation, I think that the default 
stylesheet should visualize this difference somehow.

IIRC, attempt to define tags for hierarcy for bicycle network didn't go 
that well on tagging@, so I except any usage related hierarcy similar to 
what we already have with motor vehicle networks to be shot down for some 
strange reason as if such thing wouldn't exists at all for other than 
motor vehicles.

Last but not least, why I consider forest trails and other informal paths 
so important: They're great way engage mappers to do real surveys as they 
tend to be such that imagery won't let drawing them fully or even 
correctly. Also, OSM tends to be only source in quite many places where 
the trail data has a chance to be even remotely trustable/matching what's 
on the ground.


More information about the Tagging mailing list