GitLab

Another bug in the series of leap second and date edge cases. The call to getComponents(utc) in the test case below throws an exception:

    @Test
    public void testGetComponentsEndOfLeap() {
        // setup
        TimeScale utc = TimeScalesFactory.getUTC();
        // create date as close as possible to the end of the leap second yet still
        // within the leap second.
        AbsoluteDate t = new AbsoluteDate(2012, 7, 1, 0, 0, 0, utc)
                .shiftedBy(-1)
                .shiftedBy(FastMath.nextDown(1.0));

        // action
        DateTimeComponents dtc = t.getComponents(utc);

        // verify
        Assert.assertEquals(2012, dtc.getDate().getYear());
        Assert.assertEquals(6, dtc.getDate().getMonth());
        Assert.assertEquals(30, dtc.getDate().getDay());
        Assert.assertEquals(23, dtc.getTime().getHour());
        Assert.assertEquals(59, dtc.getTime().getMinute());
        Assert.assertEquals(60.99999999999999, dtc.getTime().getSecond(), 0);
    }

Here is the exception:

org.orekit.errors.OrekitIllegalArgumentException: non-existent time 23:59:61

	at org.orekit.time.TimeComponents.<init>(TimeComponents.java:113)
	at org.orekit.time.TimeComponents.<init>(TimeComponents.java:88)
	at org.orekit.time.AbsoluteDate.getComponents(AbsoluteDate.java:1096)
	at org.orekit.time.AbsoluteDateTest.testGetComponentsEndOfLeap(AbsoluteDateTest.java:957)

Might be related to #149 (closed) since a different TimeComponents constructor is used during a leap second.