Inconsistency of duration between route and table service for identical coordinates

[hklarner]

I have noticed that routing along three coordinates frequently results in a different total duration when compared with the same coordinates in a table call and sum of durations.

Example:

# table service
http://router.project-osrm.org/table/v1/car/13.401527133953232,52.49691894523965;13.397105630079224,52.49764574610592;13.400278578988658,52.498914605838394
duration = 167.2

# route service
http://router.project-osrm.org/route/v1/car/13.401527133953232,52.49691894523965;13.397105630079224,52.49764574610592;13.400278578988658,52.498914605838394
duration = 216

Can anyone briefly explain why that happens?

Also I have compared distances between pairs of coordinates and they seem to always be equal.

Final question: we use table for vehicle routing and route for displaying solutions and it is a bit counter intuitive that the routes we compute actually have different lengths when we draw them, so is there a parameter setting that guarantees that route and table result in identical durations?

[jcoupey]

When considering 167.2 = 104.7 + 62.5 from the durations matrix you're looking at trip pairs independently, not an actual route through the middle point, and what you get is a route with a U-turn. The route request does not have this U-turn enabled by default. You'll get the same consistent result if adding ?continue_straight=false.

[hklarner]

Makes sense and can confirm that ?continue_straight=false fixes the inconsistency. Thanks!