Quote from: jamespetts on May 13, 2020, 03:19:43 PMI am not sure that I understand the point that the routes "seem to be gone entirely" - can you elaborate?
When a motorway diverges, more generally, wen any oneway road diverges, private cars will always prefer a road that is parallel to the x- or y-axis.
Further, attaching such roads to an existing diagonal road will immediately cause all cars on that road to turn onto the straight road.
Attaching straight roads to another straight road (both non-diagonal) will catch some cars, but that behavior is more randomly.
Due to the above behavior, I'd assume these cars are in heuristics mode, so I assumed routes were reset and not yet recalculated since the last update.
And there seems to be a bug, or at least quite unexpected bahavior in heuristics mode related to diagonal one-way roads.
I have split this from the topic about the losses of synchronisation, since this is a separate issue. The problem is not that the routes are not present: they are confirmed present and the cars at least mostly follow them.
Are you able to put together a precise reproduction case for this so that I can look into it in more detail?
You can observe this on stephenson-siemens, on all motorways there were routes stored. Now it feels like there are none stored anymore (as described above)
A preciese reproduction case is difficult. I don't know, what exactly should that case show? A new map without any calculated routes?
Quote from: Freahk on May 13, 2020, 05:43:15 PM
You can observe this on stephenson-siemens, on all motorways there were routes stored. Now it feels like there are none stored anymore (as described above)
A preciese reproduction case is difficult. I don't know, what exactly should that case show? A new map without any calculated routes?
You may have misunderstood: my apologies. I have conclusively determined (using a debug build that shows the actual routes in the GUI) that the routes are indeed present. If you are seeing odd behaviour, this is a separate issue to the absence of routes. I will need a reproduction case specifically of odd behaviour in a specific location with a specific description of what cars do in that specific location that is incorrect and what about that specific behaviour is incorrect.
Thank you again for the report.
Please could you add debug builds to the nightly section?
Server with debug info, so that it will print a backtrace when it crashes, and clients with this route info shown?
There are many places where cars take the wrong way, and seeing what is really stored and where would help debugging.