Versions impacted by the bug

>=5

What went wrong?

When subsetting a coverage's discontiguous time axis, the initial index will not be set unless the start time is contained within a time window and the subset will fail. For example, if the discontiguous time axis values look like this:

index 0: ( 84.000000,  96.000000) == (2013-12-15T12:00:00Z, 2013-12-16T00:00:00Z)
index 1: (108.000000, 120.000000) == (2013-12-16T12:00:00Z, 2013-12-17T00:00:00Z)
index 2: (132.000000, 144.000000) == (2013-12-17T12:00:00Z, 2013-12-18T00:00:00Z)

and the requested subset time is 2013-12-16T11:00:00Z, it would be expected that the the subset would start with index 1. However, the current code fails with no points in subset: lower 107.0 > end 144.0.

Fix with test incoming.

Code of Conduct

• I agree to follow the UCAR/Unidata Code of Conduct