Hello,
I have been wondering if it would pose a big performance-problem to add multiple (many) possible time-frames for a shipment.
My use case is the following:
- we are transporting people
- we can change the start-time for a transport +/- 15 minutes
- drivers have to take breaks (30 minutes a day)
- we wouldn't want drivers to take their break while people are being transported
This leads to a lot of "shipments" being defined like this:
ordered: 8:00 o'clock, 20 minute drive
-> start between 7:45 and 8:15 (8:00 - 0:15; 8:00 + 0:15)
-> end between 8:05 and 8:35 (8:20 - 0:15; 8:20 + 0:15)
So, for a 20 minute drive, we end up with a maximum time-window of 50 minutes - for that reason vroom often schedules the break while we are transporting people. That is something we do not want.
My proposed workaround would be to define multiple time-windows for the shipment, such as:
7:45 -> 8:05
7:50 -> 8:10
7:55 -> 8:15
...
(+/- some small margin)
Would that introduce a huge additional performance loss? Or is there a better way to model our problem?
BTW: thank you very much for making such a powerful library open-source
Hello,
I have been wondering if it would pose a big performance-problem to add multiple (many) possible time-frames for a shipment.
My use case is the following:
This leads to a lot of "shipments" being defined like this:
ordered: 8:00 o'clock, 20 minute drive
-> start between 7:45 and 8:15 (8:00 - 0:15; 8:00 + 0:15)
-> end between 8:05 and 8:35 (8:20 - 0:15; 8:20 + 0:15)
So, for a 20 minute drive, we end up with a maximum time-window of 50 minutes - for that reason vroom often schedules the break while we are transporting people. That is something we do not want.
My proposed workaround would be to define multiple time-windows for the shipment, such as:
7:45 -> 8:05
7:50 -> 8:10
7:55 -> 8:15
...
(+/- some small margin)
Would that introduce a huge additional performance loss? Or is there a better way to model our problem?
BTW: thank you very much for making such a powerful library open-source