[mkgmap-dev] special cases in restriction relations, r3178 still not perfect

Fri Apr 11 16:30:29 BST 2014

Hi Gerd,

 > The problem is that the to-way is a loop, both start and end point
 > are connected to the from way.

Loop gives you 2 possible choices. It should be relative easy to find 
which one fits description in "restriction=". I mean, if you already 
have this complex code, then you probably could go a bit further, 
calculate 2 angles and make a decision. But I understand, that some 
generic algorithm would be preferable.

 > Or maybe "start/end"  already means "either start or end, not both" ?

IMO this statement doesn't exclude loops.

Maybe we could propose additional requirement, that form-via-to set of 
members should constitute one unambiguous way? This would exclude loops.

-- 
Best regards,
Andrzej