[Tagging] Feature Proposal - RFC - area:highway

Tobias Knerr osm at tobias-knerr.de
Thu May 12 08:53:50 BST 2011

Am 11.05.2011 23:45, schrieb Stefan Bethke:
> Am 11.05.2011 um 23:01 schrieb Tobias Knerr:
>> M∡rtin Koppenhoefer wrote:
>>>> If you follow the convention that each way should be drawn along the
>>>> center of the real-world feature, then the width of e.g. a sidewalk can
>>>> still be determined at any point along the road from just the single
>>>> outline area and the way position.
>>> no, if this would be possible there would be no sense at all to map
>>> areas. You can't see sidewalks as "just another lane", because they
>>> tend to be quite irregular in certain settings (unlike lanes which
>>> usually keep their width and have no corners and other weird points).
>> I don't think this contradicts my argument. Look at the cross-section of
>> the road at any point:
>> | *  .  .  .  .    *  |
>> The vertical lines are road area outlines, the stars are sidewalk ways
>> and the dots are other "lanes".
>> If we make the assumption that each way marks the center of that "lane",
>> we can easily calculate the width of the two sidewalks at this
>> particular cut through the road: It's 2 times the width between the
>> sidewalk and the area outline.
> The last time I checked, we're mapping in two dimensions, not one :-)
> I'm not sure that mapping the actual physical extent of the various parts of roads is feasible in terms of number of mappers and their motivation, but if anybody is serious about mapping crossings and physical properties of these areas, I think mapping them as areas is the obvious and logical way forward.

Well, I'm pretty sure that I wouldn't be happy if mappers felt that they
had to draw the outline of every single lane in a road. I also wouldn't
be happy to implement support for two different mapping styles
(especially considering that these are

> We already map waterways with both a way and an area.  I'd map the road, the sidewalks, connecting areas, crosswalks, parking spots, what have you, all as areas (if I felt I had exhaused housenumbers on buildings etc.)  I'd probably add curbs as ways, not areas, unless they have multiple steps in them and approach a meter or so in width.
> Of course, that doesn't answer how anybody would be able to tell that all these features together form "the road", except for their proximity.  I'd like to learn about where that information would actually be required.

Example: A 2D rendering wants to visualize highway=residential as a way
with two :

Describe an algorithm that does that based on a bunch of ways, each with
its own area, where these areas don't even necessarily share nodes.

Any sensible rendering for applications will *not* render . You wouldn't
see anything in lower zoom levels, and the exact shape of a sidewalk is
pretty much irrelevant for most purposes. So they will draw a
fixed-width line for a highway (much wider than it is in reality), and
maybe colored casings depending on whether or not there are sidewalks.
> Stefan

More information about the Tagging mailing list