[OSRM-talk] Island hopping

Florian Lohoff f at zz.de
Tue Oct 14 12:27:23 UTC 2014

On Tue, Oct 14, 2014 at 12:45:34PM +0200, Zenon Panoussis wrote:
> Hi
> > How can you detect whether there is an intersection at sea or in
> > the harbour? 
> If it is an intersection between route=ferry and route=ferry,
> it is almost certainly at sea. The only possible exception is
> on the port quay itself, but there you should also have an
> intersection between route=ferry and natural=coastline. Hence
> the proper logic would be something like

An intersection is NOT a problem. If a clueless moron sets a shared
node between ferry routes is the VERY CLEAR intention that
a ferry can take ANY way ... And in the case of the Greek Routes
its not changing the ferry in mid sea but the ferry using the route
into the harbour and out the harbour again. Dont assume all ferry
routes are in one piece. There are ferry routes which are made
up of 20-30 ways. So your code needs to analyse the degree of
the tree node to exclude it from routing via.

> This depends on the main purpose of OSRM. If it is primarily
> intended as a debugging tool for OSM map data, then that is the
> correct approach. But if OSRM is primarily intended as a routing
> tool, it is more important that it does its job correctly and
> does not add its own errors on top of OSM errors.

There are probably 30-80 routing capable applications out there.
On my mobile i have locus maps, osmand, mapfactor navigator. I am 
using a GPSMap 60 with mkgmap generated maps.

Are you telling me i need to fix all those 3 Open Source and 1 Commercial app
just to fix this issue? There are TONS of such issues in the OSM Dataset. We
need to fix the data. OSRM is one of those tools which turn up such
inconsistencys now which is A VERY GOOD THING. So go out and fix the
data. I am using OSRM myself for Quality Assurance. I am calculating
~50k routes every 4 hours on the OSM Dataset and monitor route length
variantions which shows broken data.

> But you would see it anyway. In this case, if OSRM refused
> mid-sea changes of ship, you would see a route that only goes
> to the entrance of the Piraeus port and then stops there. This
> way it would be even easier to identify the OSM error just by
> zooming in on the place where the route stops.

Somebody told OSRM its okay for the ship to enter the harbour and
directly leave it again. 

Dont set joint nodes between ferry routes in mid sea.

> Example: start: Rue des Terre-Neuvas, Miquelon-Langlade, end:
> the road in Seal Cove about 50 km north-northwest. There is no
> route, OSRM fails and the startpoint is moved to the middle of
> the sea.
> Start like this: http://s8.postimg.org/t3oc1dbrp/miquelon0.png
> Click the end on Seal Cove: http://s28.postimg.org/rs7pa9dh9/miquelon1.png
> Boom, error. And see, now we found another ferry route that is
> broken just where the green pin moved.

The ferry route is broken in the Data - Fix it.

Florian Lohoff                                                 f at zz.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: Digital signature
URL: <http://lists.openstreetmap.org/pipermail/osrm-talk/attachments/20141014/2ba042c2/attachment.sig>

More information about the OSRM-talk mailing list