Inconsistency between table and viaroute solutions

[ZupoLlask]

Dear Community,

I've stumbled upon a routing inconsistency between viaroute and table solutions.

Although all 3 locations are considered routable with:

• http://router.project-osrm.org/table?loc=41.143505,-8.621610&loc=41.146953,-8.622990&loc=41.148046,-8.623843

The same 3 locations produce "{"status_message":"Cannot find route between points","status":207}" with:

• http://router.project-osrm.org/viaroute?loc=41.143505,-8.621610&loc=41.146953,-8.622990&loc=41.148046,-8.623843

Can you please help?

[emiltin]

difference with the z parameter?

[ZupoLlask]

No difference at all...

[ZupoLlask]

@emiltin Maybe this may be somewhat related with #930. What do you think?

[TheMarex]

Yeah the behavior of viaroute and table is currently not the same. This will probably change in the future to be more consistent.
That said. I'm not sure why viaroute fails in this example. #930 should have fixed most "No route found" errors.

[DennisOSRM]

Looks like a minor regression

[ZupoLlask]

When there's a fix to test this regression please let me know.

[MoKob]

Was able to reproduce (and fix) this locally.

The problem here is that the via-node is mapped to a street that requires a u-turn at the via-node. The default search does not allow via-nodes at u-turns.

Passing http://router.project-osrm.org/viaroute?loc=41.143505,-8.621610&loc=41.146953,-8.622990&u=true&loc=41.148046,-8.623843 (allow a u-turn at the via location), the plugin can find the correct route.

The difference here is that the table plugin does not have to worry about u-turns (point-to-point).