[OSM-dev] Line segment with width primitive?

[M∡rtin Koppenhoefer]

2010/9/11 Morten Kjeldgaard <mok at bioxray.dk>:
> I think it would be extremely useful to have a line segment with a width
> primitive.


why does this have to be a primitive? You can do this with tags as well.


> As an example of where this could be of use, is a thin strip of trees lining
> a field (say, 15m wide, 500m long).


actually I'd consider this preliminary, putting nodes for each trees
seems more the "final condition" to go for (probably unreachable).
Also consider that 15 metres wide will be the average/maximum of the
upper part of the trees, surely not the width on the ground or in 2
metres height (where people or cars are moving).


 Currently, the only way to map this is
> using a very long, thin polygon.


or single nodes


 However, it would be much simpler (and
> space-saving) to store this as a simple way tagged with "width=15m".


well, see above, width=15 metres where?


The
> renderers would draw this as a rectangle, the length of the way and with the
> specified width.


in the case of trees rectangles are quite ugly.


> Another example is a long, thin stretch of grass, for
> example between buildings or between roads.


in this case a polygon is much nices, as in many cases the rectangle
will not be the best representation of this stretch (beeing not
rectangular). Also estimating the width is IMHO more complicated then
tracing the outline e.g. from an aerial image. Also your suggestion
imposes problems when it comes to topology (where to connect what,
what is actually connected in real life)


> In principle, this primitive could be used to draw all kinds of rectangular
> and square areas. A way of 100m, with a width of 100m would be rendered as a
> 100x100m square.

while technically possible, I'd see this as regression compared to
areas. It makes mapping more complicated and less intuitive for the
benefit of what exactly? processing speed?

cheers,
Martin