[Tagging] As the crow flies

Ronnie Soak chaoschaos0909 at googlemail.com
Tue Feb 26 18:17:47 UTC 2013

2013/2/26 A.Pirard.Papou <A.Pirard.Papou at gmail.com>

> The specification I'm trying to suggest is exactly that.
> There is a gap in an OSM route and the sole idea is to bridge it.
> We must indicate "go from here to there in an unspecified way".
> It is just to
>    - make sure that those who follow the route will go "there" and not
>    somewhere else
>    - indicate to validators that there is no mistake and that the route
>    is connected and maybe looped
> That there are paths in between or not, what those possible parts are
> called, that the route may exist and just be unknown, that there should be
> paths but that there is a map bug, or any other reason for a gap, all that
> is very good for a note=literature but is totally irrelevant for the
> attempted specification.
> They were mentioned because the idea evolved from a path feature to a
> relation feature.
> Thanks for the clarification. Now I understand the problem.
The order of ways in the relation is of course giving you a hint on where
to continue after a gap, but a router might now now on which end of the
next way to continue.
The route might not follow the direction of the way in the OSM-sense. Nor
is it always the 'unconnected' end, as it might be a gap on both ends of
the next way.
Also the nearest end might not always be the right one. Imagine a path
below and on top of a steep cliff. They might be quite near, but you can't
go there directly.

When encountering a gap, routers will now probably switch to their standard
routing algorithm (either 'follow road' or 'straight line') to lead you to
the next point.
What the next point is will be determined by the router or the conversion
software that brings the OSM format into a format understandable by the
I imagine it will mostly be 'nearest endpoint'.

You now search for a way to give those tools a hint to influence their

One easy solution would be: include (of course in the correct order) the
start- and endpoints of the ways in the relation. (Maybe with a different
That's all the information the router needs. Either their is a way between
those points that's also included in the relation, then the router can use
it, or there isn't, than the router can do it's standard routing between
those points.
The user can choose the type of standard routing on the routing

Is this to much 'mapping for the router' or aceptable in the database?

Best regards,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20130226/61eb0720/attachment.html>

More information about the Tagging mailing list