<div class="gmail_quote">On Thu, Jul 30, 2009 at 12:00 PM, Ben Supnik <span dir="ltr"><<a href="mailto:bsupnik@xsquawkbox.net">bsupnik@xsquawkbox.net</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<snip><br>
Second: double-decker bridges I think open a bucket of worms:<br>
<br>
- Do the two layers share the colocated nodes? Is this even legal?<br>
- How well do the editing programs let you manage such a beast?<br>
- How well do the maps render it?<br>
<br>
In Boston I see that the double-decker part of I-93 does not overlap itself...in practice such a map would be pretty unreadable so it doesn't surprise me that the data is how it is.<br>
<br>
So far I think any candidate heuristic I have looked at for building a 3-d network from the data would manage a double-decker road in a reasonably sane way.<br>
<br>
cheers<br>
Ben<br></blockquote></div><br>I know you're trying to deal with the reality of the planet file as it is now, but I think you should target the way the data should be, and let the crowd fix the errors. The two layers of the bridge should not share any nodes unless you can physically navigate between the two ways at that point using the primary mode of transportation for the ways (i.e., for cars, no shared nodes unless you can drive from one to the other at the shared node). Even if there were an elevator, it should be represented by a (very) short connecting way with appropriate tagging and access restrictions. Ideally, each way comprising the double (or more) decker bridge would be tagged with a layer tag to indicate the vertical ordering. If you're trying to build a 3-D model, you could use the value in the layer tag to space the ways vertically by 5-10 meters. Not ideal, but it might work.<br>
<br>Karl<br>