Re: [sumo-dev] NaN-Check in TraCI

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

I fully agree that this should be handled with a friendly error message rather than a crash.

Please create an issue on github.

If you could put together a small sample input for reproducing the crash and attach it to the ticket, that would be great.

Thanks for reporting!

Jakob

Am Di., 11. Dez. 2018 um 09:32 Uhr schrieb Bernd Lehmann <sys.bernd@xxxxxxxxx>:

Hi,

we heavily use polygons to visualize data within sumo. During the
development of the visualization we run into an error which terminated
sumo with an error message similar to: "pure virtual function called".

The reason was a error in one of our calculations for the shapes used
as polygons. The position of the shapes was filled with NaNs which we
provided to sumo via TraCI. I didn't locate the code where the error
ultimately arises, but i guess it has something to do with the used
RTree which sorts the polygons by their positions. Since relational
operators for NaNs always fails, i guess that this cause the error
somehow.

Due to the error message i suspected an memory handling error within
sumo. I wasted a lot of time before i noticed the NaNs. Therefore as
an enhancement i would suggest to add NaN checks in the implementation
of TraCI. A proper error message would prevent other developers to run
into the same issues.

What do you think about it? Shall i create a issue on github?

Kind regards,

Bernd
_______________________________________________
sumo-dev mailing list
sumo-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-dev