Issue with TimeInterval Mod Function #349
Description
In some cases, the mod function overloaded for TimeInterval class produces incorrect results.
Below is a list of the incorrect result cases.
These are simple cases where I call mod(interval1, interval2). In each case, interval1 has a single unit set to nonzero and interval2 has a single unit set to nonzero. I am using the standard argument names to indicate which unit is set and what the value is. In all cases, I am using ESMF_KIND_I4 values.
Interval1 |
Interval2 |
mod(Interval1, Interval2) |
|---|---|---|
| mm = 1 | s = 17 | ZERO |
| mm = 1 | m = 17 | ZERO |
| mm = 1 | h = 7 | ZERO |
| mm = 1 | d = 8 | ZERO |
| yy = 1 | s = 17 | ZERO |
| yy = 1 | m = 17 | ZERO |
| yy = 1 | h = 7 | ZERO |
| yy = 1 | d = 4 | ZERO |
where ZERO is a TimeInterval with all units == 0.
As you see, in each case Interval2 does not divide Interval1 evenly, and yet I found that mod(Interval1, Interval2) == ZERO.
In the cases where Interval2 is n days (n=8, n=4), it still returns an incorrect result. So it appears that the issue is not limited to a time-only duration dividing a date-only duration. It works correctly for (yy, mm), (d, h), (d, m), and (d, s), where (a, b) is (Interval1, Interval2). I did not test combinations with two time-only durations or combinations with durations have the same unit, and I did not test intervals where more than one unit is set.