You can clone with
HTTPS or Subversion.
http://osrm.at/7zs - no route in OSRM, althougth data in database is normal.
another link to 1st point: http://osrm.at/7zt
also: http://osrm.at/7zu and http://osrm.at/7zv
Thanks, will have a look.
The 'only_straight_on' restriction is probably involved..?
yes, but not as 'only_straight_on'. The issue seems to be related to a situation where there are several TRs tagged at one piece of road.
sorry, not sure i understand? what's TR..?
a turn restriction.
http://osrm.at/7zJ and http://osrm.at/7zK are more examples for this bug.
In each case there is also a only_straight_on-TR involved (387023 and 387023). These restrictions explicitly allow travelling the direct route from startpoint to endpoint.
There are no other TRs involved on the same road-segments.
Thanks for the examples. Please note that the kind of restriction, e.g. only_straight_on, is most probably not related. Rather, the issue occurs when two restrictions are tagged onto the same road segment.
Got a solid lead on this one now. Appears to happen when a way in OSM data is the source segment of one turn restriction and also the target segment of another restriction.
One more if it will help:
hope this helps: http://osrm.at/7Jc
Note that some weeks ago the routing worked like a charm in these situations, so IMHO this bug was introduced (quite) recently
Pushed a fix w/ commit f68af08 that may help here but needs more testing/investigation.
Just came here because I wanted to report presumably the same bug, which can be seen at http://osrm.at/7JP. May I ask when this fix is applied to the demo website and can be verified there?
Thanks everyone for posting links to where this issue occurs. We now habe more than enough of these links.
@beart-github in about 10-12 hours from now. But please note that this issue may need further work.
Seems like this bug is still active.
Yes, we are still working on this issue.
Some more examples: http://osrm.at/7OU http://osrm.at/7OV http://osrm.at/7OW
There is some progress on this issue, e.g. here:
fixes #1041, some only_ turn restrictions are inverted under certain …
Added a fix for this. Will take roughly 18--24 hours to see this reflected on the demo site.
do we have any tests tracking this?
@emiltin I'll engineer one if this goes away with the next update
The fix is live on the demo site. Everyone, please check your examples. I looked at a few and they all appear fixed.
for the record, all problems i spotted seem to be fixed, thanks :)
thanks for the feedback @richlv. closing here.
Yes, all bugs I found are fixed too. Thanks!
Really weird situation, @NikolasHuan... @DennisOSRM
Indeed weird. Investigating the issue with http://osrm.at/7QI and will have a fix ready later today.
Fix will be deployed in a matter of minutes and appear on the demo site within the next 12--24 hours.
what kind of test would cover this?
one could get the OSM data for this issue and try to replicate the situation and then check that there is no alternative route.
there isn't a specific type of situation that can be reproduced in a small test scenario?
Certainly there is but this is hard to reproduce manually.