It's just my opinion (fwiw) but it does seem to align with similar ones expressed by others in this thread: creating a specific path to cross an area seems superfluous. Logic as to whether "you" (and/or your vehicle) can cross a beach properly belongs in a router, especially as whether you are travelling by foot, bike, horse, ATV, family sedan, 4WD... and whether the surface is loose sand, hard-packed sand, pebbly rocks, or treacherous crags that are virtually impassible even by mountain goat are all possibilities. Combinations of access, vehicle/mode of travel and surface are likely to properly arrive at quite different answers as to whether such a navigable path is routable in any given case.

Can or should a router be expected to know that a "sandy beach" (perhaps natural=beach is enough, perhaps surface=sand or other must be included) is navigable by foot across the polygon (without area=yes) and without an explicit way? Well, I'd certainly call that router "sophisticated" if it did so, but that seems to be where this sort of logic belongs, as opposed to a datum like a specific highway=path way across a beach. Yes, it is tempting to add such a path to "make things work (for now)" but as was already said, that truly is "coding (data) for a router." Along with coding data for a renderer, I believe we want to discourage that.

There will continue to be (excellent, in my opinion, but nevertheless necessary) discussions like this on whether our project puts logic into our database or into our code/tools that use data from our database. I believe most of the time, the correct answer to emerge (as it has many times -- see numerous Telenav examples) is a small amount of well-defined and well-structured data which "adhere" to a smart chunk of logic (in a renderer, in a router...).

If/as we keep up that good work, both our map and our tools become better and better.

SteveA
California

_______________________________________________
Talk-us mailing list
Talk-us@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-us

Reply via email to