Hi Andrzej, I looked at your patch and it seems reasonable to me but as a cyclist I'd prefer to use a different tag instead of mkgmap:good_road, maybe mkgmap:fast_road? Do you know how this patch influences bicycle routing? The special handling of highway=mini_roundabout is somewhat problematic because that tag should not be used on ways. I also would not use the type 0x0c for them, so I suggest to remove that rule so that our "mop up" rule catches them: # Mop up any unrecognised highway types highway=* & area!=yes [0x07 road_class=0 road_speed=0 resolution 23] Gerd popej wrote
Hi Bodo,
I think Garmin is using some kind of hierarchical routing algorithm. It looks like GPS uses only class 4 and 3 roads for long distances (at least in case of recent nuvis, hiking GPS can be more flexible). For proper routing GPS needs a continuous and reasonably dense notwork of high class roads.
My patch address 2 problems in default style.
First are links and roundabouts, which often get lower road class than connected roads. This breaks continuity of high class network and breaks routing. My patch simply increases road class for many links and roundabout types.
Second problem is low density of good highways in some regions. To make it better, my patch increases road class for international roads. These roads are detected by "ref" and "int_ref" tags, both in objects and relations.
I have added some small general corrections for "highway=mini_roundabout", "access=hov" and "oneway=reversible".
There is no direct change for "inc\roadspeed". Maybe patch overrides some of your changes in "lines" file? Anyway, I don't think road speed breaks routing, you can tune it to your liking.
-- Best regards, Andrzej _______________________________________________ mkgmap-dev mailing list
mkgmap-dev@.org
-- View this message in context: http://gis.19327.n8.nabble.com/Strange-long-distance-routes-on-Nuvi-tp589982... Sent from the Mkgmap Development mailing list archive at Nabble.com.