[OSM-talk] Ditches

Steve Bennett stevagewp at gmail.com
Tue Dec 15 11:49:34 GMT 2009


On Tue, Dec 15, 2009 at 10:19 PM, Jukka Rahkonen
<jukka.rahkonen at mmmtike.fi> wrote:
> It feels sometimes ridiculous to add layer tag to ditches and roads because
> everybody knows that in majority of cases when road and ditch are crossing, the
> road is above. A very typical example is in picture:
> http://www.coquillewatershed.org/Project%20photos/pages/lampa-199-culvert-03.htm
>
> There are millions of culverts like this. Are they really worth splitting the
> way and tagging a "bridge"?  I do not bother myself, I just let road and
> waterway to  cross without any layers.

That doesn't look like a bridge by any stretch of the imagination.
Perhaps a tunnel. Perhaps nothing.

Anyway, I'm pretty sure it's already ok to have drain and road cross
(without junction) at layer=0 - they'll be rendered right by any
reasonable renderer. It should be obvious that water is the bottom
layer, and power lines are the top layer, unless any layer tags say
otherwise. The layers are just there to solve ambiguous cases like two
bridges crossing (completely ambiguous).

Btw, to mike:
>Fair points ... If it really doesn't matter to routers and other mappers and
>doesn't interfere with anything else then I am happy to accept that there is
>no fully logical solution and that it shouldn't matter to me either!

Cool! I was thinking, it may help to think of all these features as
being drawn on little postit notes. The desk is covered with postit
notes, and sometimes they overlap. "Layer=" tells you which one is on
top, when they overlap, but it's meaningless whenever there is no
overlap. And notice that a postit note could be the bottom of one pile
(layer=-2), but also the top of another pile (layer=+3). So that's why
you break the way in the middle to change the layer, without causing
any problems - or waterfalls.

(And no, the desk is not layer=0. Layer=0 is as arbitrary as any
other, it's just the convention for most roads on the ground.)

Steve




More information about the talk mailing list