Just to follow up the ignore-turn-restrictions discussion. I have done many more compilations with several different arguments to find out if there is any specific reason why this happens. But I couldn't find any. To me it seems now that the rules given in the stylefile (lines) are not really parsed like they should. This happened with or without the --not-sorted-roads argument (I first suspected that there might be an error, but this argument makes no changes). I assumed that if i give an argument like oneway=yes {set oneway=no} would only affect lines listed below the argument. This seems not to be the case. This is an abrigded lines style-file that does not work like it should (it should not touch the primary, secondary, tertiary oneway settings, but delete them from residential, living_street and : highway=primary [0x10 road_class=0 road_speed=2 resolution 19] highway=secondary [0x0d road_class=0 road_speed=2 resolution 19] highway=tertiary [0x06 road_class=1 road_speed=4 resolution 20] oneway=yes { set oneway=no } oneway=true { set oneway=no } oneway=1 { set oneway=no } oneway=-1 { set oneway=no } oneway=0 {set oneway=no} oneway=false {set oneway=no} highway=residential [0x07 road_class=1 road_speed=4 resolution 22] highway=living_street [0x07 road_class=1 road_speed=3 resolution 22] highway=unclassified [0x07 road_class=1 road_speed=4 resolution 22] Effectively what I get is that for some roads oneway attribute will be set to no, while on most roads the oneway attribute will be lost. Another error to verify: If I put oneway=* {set oneway=no} map generation takes literally ages (2 hours instead of 5 minutes). Using the alternatives as set above map rendering time does not increase significantly from the 5 minutes. the same happens if I put maxspeed=* {set maxspeed=no} into the beginning of the stylefile.