Must footpaths always be linked to main roads?

I am reworking a section of designated footpaths in a public walking path. At the point I found the map, it was incomplete - probably reflected a partially constructed state from some months ago. I have expanded the route (unsubtle plea: would appreciate any feedback on the changeset here)

I am reasonably confident that I have the details correctly mapped now, but one thing that concerns me is that I get a warning that the footpath elements do not link to anything but themselves. The footpath in question starts from a carpark so there is a “virtual” bit - the carpark links to the main highway.

Should I
a) add a section of footpath that explicitly goes through the car park and links to the highway? It would narrowly fix the warning in the changeset but technically the footpath doesn’t start in the car park
b) leave it as is, potentially making the footpath elements unroutable

Any help or suggestions gratefully received. Thanks.

Hi tflex, welcome to our community!

As you already pointed out, isolated footways are not routable. Generally it is good practise to connect footways to any surrounding way from where access to the footway is given. See

The … highway=footway should join with another kind of way for routing porposes.

Tag:highway=footway - OpenStreetMap Wiki

Besides that highway=footway implies foot=designated and that no vehicles are allowed, unless explicitely signed/tagged, see

OSM tags for routing/Access restrictions - OpenStreetMap Wiki

So in your case foot=yes and motor_vehicle=no are already implied.

If the way is open to be used by bicycle, carriage or horse it would be better to tag it as highway=path and add foot=designated, bicycle=designated (or yes) and so on, as described on the wiki page for highway=path.

Finally the value “mixed” is not very helpful for data consumers. Better to use the documented values and split the way where necessary.

1 Like

It it not enough that paths link to something, they should also be connected to provide reasonable routing.

Otherwise you will guide poor users on strange routes such as:

Note, the carpark Way: 1346115768 | OpenStreetMap also doesn’t link to anything, so maybe you could add a service road to connect it to the main road.

Unfortunately, many routers don’t route through areas; next to the technical questions are issues around incomplete mapping. You can use https://wiki.openstreetmap.org/wiki/Tag:footway%3Dlink in those cases so signify that the way is only for routing purposes. Ideally trace a way that someone would reasonably use. It should then connect to the service road.

There is also Tag:service=parking_aisle - OpenStreetMap Wiki that can be used inside the parking lot. This might be a cleaner solution.


Looking over your changeset, I’ll add to make sure that connected paths are connected. E.g. complete this circle. How are these two ways separated (e.g. draw a fence between the ways)? This barrier should be a node on the connecting way between them, even when permanently locked. Is the foot=yes based on physical size? It could help to specify Key:maxwidth:physical - OpenStreetMap Wiki. In any case, foot=private is probably more appropriate, as foot=yes means that anyone can walk through, which conflicts with locked=yes.

Thanks for your contribution!

1 Like

thank you all. This has been tremendously useful feedback, and (as I had hoped) I got a lot of information on how I can improve the very rough first cut.

To summarise for anyone who may follow - yes, footpaths should be connected to other paths, and thereby be routable. The suggestion to validate routability by checking directions between two points in a complex route network was very useful for me, have incorporated this into my workflow