|
I have noticed certain areas have many ways mapped as the footprints of the physical roads, such as this in Indio, California. However, this polygon way itself comprises an ordered list of nodes in its perimeter. Any routing queries passing through this way must traverse the perimeter nodes in order. In other words, in the real world I can traverse this way as a straight-line path, but in OSM I must traverse this way as a twisting, curling path along its perimeter. This causes shortest-path routing problems when I use this way data in other applications. Similarly, certain areas such as this in Folsom, California represent houses' driveways as disconnected way polygons. My question is: are these polygon representations of ways correct? If so, why? |
|
Tagging it in the way that you have linked to is bad practice and not recommended by most OSM mappers, for exactly the reasons you describe. The more accepted tagging is In any case, you should supply a routable centreline with a standard |
|
Currently none of the main-stream routing engines support routing over areas properly (not that it can't be done, if just hasn't been). The camps are slightly divided on if additional ways should be added to improve routing till the point in time that area routing is supported more wide spread, or if such ways should not be added. Doing the first naturally removes pressure on the routing engine developers to actually fix the issue, doing the later makes current apps work better. As you can see, there is not going to be a hard and fast answer to your question. PS: it should be noted, as Richard has pointed out, that the tagging of the driveways is broken in any case, landuse residential should be used for the whole area in question not for individual driveways. And if you are mapping the driveways as areas (with better tagging), you should likely be mapping the actual roads that way too. |
All those abstract and unsquare polygons in the Folsom area look like a cubist painting!