Adding a month to a date in Java two ways gives two different results (leap year philosophy?)

Question

Assuming Jan 31, 2000, a leap year, the following two ways of adding a month give me different results. Am I doing something wrong or are these two different philosophies of handling leap-year months? And, if these two approaches are just a philosophy difference, how would you know which method to pick?

Method 1: using LocalDate plusMonths():

LocalDate atestDate = LocalDate.parse("2000-01-31");
System.out.println("One month in future using LocalDate.addMonths() " +   atestDate.plusMonths(1));

Output:

One month in future using LocalDate.addMonths() 2000-02-29

---

Method 2: using Calendar:

Calendar zcal = Calendar.getInstance();
zcal.set(Calendar.DAY_OF_MONTH, 31);
zcal.set(Calendar.MONTH, 1);
zcal.set(Calendar.YEAR, 2000);
zcal.add(Calendar.MONTH, 0);
System.out.println("ONE MONTH IN FUTURE using Calendar: " 
 + zcal.getTime());

Output:

ONE MONTH IN FUTURE using Calendar: Thu Mar 02 2000 

Why are these two dates' output not the same?

Thanks.

Answer 1

zcal.set(Calendar.DAY_OF_MONTH, 31);
zcal.set(Calendar.MONTH, 1);
zcal.set(Calendar.YEAR, 2000);

This defines the date of February 31st, 2000, which, due to lenient measuring, is equated to March 2nd, 2000. Month values are zero-indexed.

Setting strict:

zcal.setLenient(false);

We get an exception:

Exception in thread "main" java.lang.IllegalArgumentException: MONTH: 1 -> 2
at java.util.GregorianCalendar.computeTime(GregorianCalendar.java:2829)
at java.util.Calendar.updateTime(Calendar.java:3393)
at java.util.Calendar.getTimeInMillis(Calendar.java:1782)
at java.util.Calendar.getTime(Calendar.java:1755)
at Employee.Tester.main(Tester.java:19)

So you can see that the interpolation was shifted over to March. If you wish to hard-code dates for testing purposes, I would recommend using strict validation to avoid these cases. There have been other documented issues related to setting values that make things messy.