DARP: Maximum ride time constraint between pickup and delivery

[Rinocastellano]

I'm implementing a Dial-a-Ride-style problem using cuOpt's pickup & delivery feature. In the DARP literature, a critical constraint is the maximum ride time (or maximum in-vehicle time): for each order, the elapsed time between the actual pickup service time and the actual dropoff service time must not exceed a given threshold T_max[i]. It cannot be expressed with static task_time_windows alone, because tightening the dropoff window only approximates the constraint: if the pickup is delayed within its own time window, the effective ride time may still exceed T_max.
Is there any native or planned support for a per-order maximum ride time constraint in cuOpt's routing solver?

Workaround I'm currently considering:
Setting dropoff TW as [earliest_pickup + min_travel, latest_pickup + T_max], which gives a conservative upper bound but doesn't enforce the ride time exactly when pickup timing is variable.
Is this a known limitation? Has anyone found a cleaner workaround, or is this on the roadmap?

[np96]

Hi @Rinocastellano
This is not implemented yet and workarounds won't enforce what you need — either the solutions will be too strict/too relaxed.
You may open an issue and, if you want, implement on your own. AFAIR it was never mentioned on GitHub.

There's a chance that I'll have to implement it until the end of this year. My case is food delivery — a courier needs to deliver hot food and time windows alone won't enforce this.