difference between time64 columns

[lmullany]

What is the best/fastest way to add a duration column that indicates the difference between two time64 columns?

import datatable as dt
import numpy as np
from datetime import datetime as d
DT = dt.Frame(t1 = [d(2020, 5, 11, 12, 0, 0)], t2 = [d(2020, 5, 14, 16, 10, 59)])

To make a new column that represents the minutes between t1 and t2, do I need to do something like this?

DT["dur"] = (DT["t2"].to_numpy() - DT["t1"].to_numpy())/np.timedelta64(1,'m')

Is there a better way?

[oleksiyskononenko]

At some point we will add a new duration type, so that one could directly calculate difference between the time columns, or between the date columns.

However, what you're asking is possible to be done in datatable even now. That's because internally time64 column is stored as int64. So you can do something like that to calculate the number of minutes

DT["duration"] = DT[:, (dt.int64(dt.f.t2) - dt.int64(dt.f.t1))/(60*10**9)]

[lmullany]

Thanks much!