[OSM-talk] Layer transitions

Lambert Carsten lhc.osm at solcon.nl
Wed Aug 12 11:36:09 BST 2009


On Wednesday 12 August 2009 11:59:36 Jochen Topf wrote:
> Hi!
Thanks for your quick input.
>
> I have amended the bridge and tunnel pages with the reason for those rules:
> Rendering breaks if you have different layers on junction nodes. 
This is tagging for the renderer which I thought was generally considered a 
big nono.
> But whats 
> more important, in real life bridges don't start in the *middle* of
> junctions so there *is* a little bit of non-bridge roads between the
> junction and the bridge.
This argument has been made previously and doesn't really hold up. An 
X-junction between a large primary road and a connecting cycleway (for 
example) does not either get a small bit of primary inserted in the cycleway 
to cover the width of the larger primary road.
>
> For very tight features this is a problem, I agree. But I haven't seen
> a good solution for this yet. The rules on the wiki work in most cases and
> give good guidance to newbies. But they should never be read as gospel,
> so if you have a solution that works better for your case, just use it.
Of course, personally I let a bridge connect to a junction without changing 
the layer tag of the connecting roads.
The 'problem' has shown up because keepright is now showing these 
intersections as possibly incorrectly tagged and people wanting to 'clean up' 
their area.
>
> For canals in Dutch cities, I'd consider not tagging the roads special but
> giving the canals a layer=-1 and maybe use tunnels where they cross under
> the roads. That might be a bit of a hack, but sometimes its easier to tag
> something as a tunnel even if it might technically not be one.
Changing the layer of the water to -1 will still generate an 'error' from 
keepright at the bridge junction because the bridge is  layer=1. So  in those 
cases the bridges would need to be tagged layer=0 as well (virtually all the 
bridges in Amsterdam and similarly built cities).
That would work to 'get around the T-junction rule', but then the question 
arises: What did we achieve with all this? 
The renderer still needs to figure out to draw the bridge correctly when 
connecting to a junction. Newbies now not only need to 'understand' the rule 
but also how and when to get around it.

Lambert Carsten




More information about the talk mailing list