-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am 18.07.2010 04:52, schrieb Alan Millar:
>>
http://maps.cloudmade.com/?lat=45.418666&lng=-122.321563&zoom=16&directions=45.41881686582581,-122.32349395751953,45.4324308667588,-122.36778259277344&travel=car&styleId=1
>
>>
>>
A good example of a place where dual carriageways should be used.
>>> Notice the double-double yellow lines separating the two
>>> directions of traffic.
>
>> Definately. The road should never have been modeled as only 1 way
>> in the first place. The router was absolutely right. The road was
>> wrong.
>
> No, I don't agree; this is not an obvious dual carriageway.
>
> You CAN turn both left and right leaving the highway; the middle
> lane is a left-turn lane in both directions, crossing the opposite
> side.  You just can't turn left entering the highway.

Why isn´s it?
I see a dual-carriageway with oneway-links on both sides
and 2 oneway-links in the middle. As all these ways are
oneway in reality just connect or don't connect
the middle lane to the carriage-way they are crossing
if and only if making U-turns using the middle lane
is forbidden.

Marcus



>
> An that is just this one crossing.  The very next intersection at
>
> http://nroets.dev.openstreetmap.org/demo/index.html?lat=45.40876&lon=-122.30162&zoom=17
>
>
>
is just a simple meeting of two roads, as are most of the rest of the
> crossings as the road heads east.

Ok, so turn-restrictions it is.

> That doesn't make any more sense to me than adding "no left turn"
> in a place where the link approaches at a small angle of a few
> degrees and you are supposed to merge left onto the highway.  That
> may make sense for routing restrictions, but it is going to really
> confuse a human map reader.
>
> Am I really asking something unreasonable of a router that when an
> xxxx_link way meets an xxxx way at a very low angle, the router
> should know to go forward and not almost reverse?

Yes because
a) such intersections with very sharp angles exist in many cities
    and are valid. (I already said that I know real live examples.)
b) It is not the place of the routing-algorithm to change the
    topology of the well defined graph it is routing on.
c) Routers don't care for angles or even for any kind of physical
    location most of the time.  Metrics do but most metrics would
   act incorrect if they where to associate extreme cost with sharp
   angles. (A shortest route is no longer shortest if it cares about
angles,
   fastest- and fuel-efficient and motorbike-friendly- already have their
   own set of rules for angles,...)
d) You would need to add your heuristic not to one but to ALL our
   routing-engines and all their routing-algorithms, document them, test
   them, tune them, make them well-behaved.  As opposed to fixing the map.

Whould it not be easier to write a trivial tool to search for such
sharp angles
and offer them to users for checking if a turn-restriction would be
apropriate? Maybe even prepare a changeset, so the offered
turn-restriction
can be added with one click?
That´s even less work then changing one and only one router considering
the testing and documentation involved as the router still needs to work
correctly on all other roads.

Marcus
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkxCoj0ACgkQf1hPnk3Z0cQgAQCeKshSgy0jfkSkbK6IZ8V0wgHS
MfgAoL1bRkOH6ab3byiABBaAp0u2CTpZ
=lN1l
-----END PGP SIGNATURE-----


_______________________________________________
Routing mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/routing

Reply via email to