Closed lucb1e closed 8 years ago
I have varying experience with this, but also have observed some situations where I would say we should systematically observe and evaluate this ...
Yes, probably my observation fits here: Suppose we expect OsmAnd to take a U-turn on a dual carraigeway road.
Logically, OsmAnd should use the closest link that joins the two carriageways. But practically OsmAnd ignores any link that results from a lower-raking road crossing the dual carraigeway road.
For example, if we are on a primary dual carriageway road, and if a secondary road crosses the road without any turn restrictions at the junctions, this link is a good candidate for taking a U-turn. But OsmAnd ignores this lower-ranking cross-link. It will try to find a primary link that joins the carriageways; thus taking a far longer route unnecessarily.
This behavior is very consistent (repeatable), and this bug can be checked very easily.
Yes this problem is frecvent. I reported many times, but lately i didnt reporty this problem , because i fear they will destroy the entire routing again. In this moment the routing, generally is good enough. PS: This software is old enough to have 3D vector rendering engine until now. But they "are stick in the mud" with basic function like routing...
Closing this one as it is very general with no specific input - we have issues elsewhere o tweak the routing.
The old routing engine was next to prefect. The newer one, which is probably a year old by now (I don't remember) can often be improved. It tries to take bigger roads all the time, making the route simpler but going around where I need to be.
It's probably a good idea to prefer bigger roads, but not this much.