[mkgmap-dev] Detecting, reporting and highlighting dead-ends

Sat Jan 5 20:40:23 GMT 2013

Recently, a fellow mapper (FinWonder) came up with the idea of 
highlighting dead ends: 
http://forum.openstreetmap.org/viewtopic.php?pid=299968

He posted some Python code here:

https://github.com/KonstantinShemyak/no-dead-ends-OSM

python markOSMBridges.py example.osm > marked-example.osm

Unfortunately, the code is a bit memory-hungry; it failed to process the 
entire finland.osm (about 3 gigabytes, uncompressed from 
finland.osm.pbf) within 8 gigabytes of RAM.

I think that this is a very interesting idea. I would even refine it 
further and split the 'dead-end graphs' for each mode of transportation: 
car, bicycle, foot, etc. For example, a way that ends in highway=steps 
is not a dead-end for pedestrians, but it is for everyone else. A 
highway=service leading to a fuel station along a highway=motorway is 
not a dead-end for cars, but is for everyone else. A way that ends in 
highway=path,access=no,bicycle=yes is dead-end for everyone except 
bicyclists.

Short dead-ends (say, 10-meter or shorter driveways) should probably be 
ignored. While computing the dead-ends, the code could warn about 
'major' highways that are dead-end, possibly because a T crossing with 
another way is incorrectly tagged (the junction node is not linked to 
both ways). Currently, mkgmap only complains about dead-end oneways, 
because that is much easier.

Would this be feasible to implement in mkgmap? Or maybe better in a 
separate tool? I would guess that the dead-end way sections would be 
useful to be in separate layers that can be enabled or disabled on 
demand.

	Marko