[OSM-talk] Highways with highway=steps
osm.list at randomjunk.co.uk
Fri Sep 28 14:57:25 BST 2007
On 28/09/2007, Martijn van Oosterhout <kleptog at gmail.com> wrote:
> On 9/28/07, Dave Stubbs <osm.list at randomjunk.co.uk> wrote:
> > the cycle map  just renders anything with "ncn" as NCN and anything
> > with "rcn" as regional routes. It lets you do other things too such as
> > "ncn => proposed", although nothing clever is done yet when you add
> > that. If you have the "ncn_ref" tag there then you can just leave off
> > the route tag completely as far as the renderer is concerned.
> While I agree it's simpler, it's also completely useless for anywhere
> outside the UK. We need a proper way of handling cycling routes that
> doesn't have artificial restrictions on the number of networks.
I wouldn't have thought it was *completely* useless. It's most
definitely UK-centric which is hardly surprising given what we've been
trying to do with it.
But you basically have three defined levels of cycle route.. national,
regional, and local.
For refs of these you have ncn_ref, rcn_ref, lcn_ref... to point out
specifics about them you have ncn, rcn, lcn.
The tag space isn't restricted so there are virtually infinite
possibilities for *cn. ie: no "limits", artificial or otherwise if you
follow the scheme through to it's logical conclusion. It's possible
that's not such a great idea though.
As we'll be getting relations shortly thanks to some great work from
many developers, I think routes will be getting easier and all this
tag hacking can go away (to some degree at least).
A relation along the lines of cycle_route=ncn can then happily
co-exist with a relation cycle_route=rcn with a nice common name= and
ref= tags on them too, while renders, routers etc have a generic
fall-back if they've never heard of that route type before.
I'm sure there will be a discussion on the exact dynamics of the
scheme for that.
More information about the talk