[pgrouting-users] Bug in TRSP?

[Stephen Woodbridge]

Hi Ashraf,

Can you look at this and respond when you have a chance. Thanks!

Also, I have your enhancements to the TRSP, I started to merge them but 
got buried in work. I will get to them as soon as I have a free day to 
tackle it. I'm very excited to get them added to the code base.

Thanks so much.

Best regards,
   -Steve

On 2/11/2014 4:13 PM, Fabien Ancelin wrote:
> Hello all,
>
> There is a piece of logic in the trsp algorithm that I suspect is a bug.
> It happens in the file GraphDefinition.cpp, line 426
>
>
> if(m_lStartEdgeId == m_lEndEdgeId)
>
> {
> if(get_single_cost(1000.0, path, path_count))
>
> {
> return 0;
> }
> }
>
> *err_msg = (char *)"Path Not Found";
>
> deleteall();
>
> return -1;
>
>
> Is there any reason why the get_single_cost function get passed an
> absolute value. Even more is there any reason for the following
> clause conditions in the function get_single_cost?
>
>
> GraphDefinition.cpp, line 493
> if(start_edge_info->m_dCost >= 0.0 && start_edge_info->m_dCost *
> (m_dEndPart - m_dStartpart) <= total_cost)
>
>
>
> GraphDefinition.cpp, line 506
> if(start_edge_info->m_dReverseCost >= 0.0 &&
> start_edge_info->m_dReverseCost * (m_dStartpart - m_dEndPart) <= total_cost)
>
>
> Shouldn't we have the following clause instead ?
>
> if(start_edge_info->m_dCost >= 0.0)
>
> if(start_edge_info->m_dReverseCost >= 0.0)
>
>
> The issue is : it is possible to have the a route that starts and ends
> on the same segments and that the cost associated with the used portion
> of the segment is superior than 1000 (especially if you route on long
> segments and your cost is in meters).
>
>
> I have changed the line 493 ad 506 in my version of trsp so that it
> doesn't test against the total cost and it seems to work fine. Is what I
> report here a bug, or I am missing something in the way TRSP works?
>
>
> Regards,
>
>
> /Fabien/
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> Pgrouting-users mailing list
> Pgrouting-users at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/pgrouting-users
>