[Bug classpath/22756] GregorianCalendar.getTimeInMillis() fails with large millisecond values

[gcc-bugzilla at gcc dot gnu dot org]

getTimeInMillis() converts large (magnitude) values to
year/month/day incorrectly.  Typically, it miscalculates
the month as negative and throws IllegalArgumentException.
For example:

FAIL: gnu.testlet.java.util.GregorianCalendar.conversion: uncaught exception at
"Testing setTimeInMillis(281474976710656) i = 48" number 3
java.lang.IllegalArgumentException: month out of range
   at java.util.SimpleTimeZone.getOffset (SimpleTimeZone.java:684)
   at java.util.GregorianCalendar.computeFields (GregorianCalendar.java:610)
   at java.util.Calendar.setTimeInMillis (Calendar.java:541)
   at gnu.testlet.java.util.GregorianCalendar.conversion.testMonotonic1
(conversion.java:79)
   at gnu.testlet.java.util.GregorianCalendar.conversion.test
(conversion.java:41)
   at gnu.testlet.SimpleTestHarness.runtest (SimpleTestHarness.java:254)
   at gnu.testlet.SimpleTestHarness.main (SimpleTestHarness.java:364)

The problem is that GregorianCalendar is doing some key
conversion calculations using 'int' instead of 'long', and
intermediate values are overflowing.


------- Comment #1 from from-classpath at savannah dot gnu dot org  2005-01-15 
12:40 -------
testMonotonic1 passes with current Classpath CVS HEAD and JamVM, so
setTimeInMillis() can now handle all values up until Long.MAX_VALUE.  I'm
closing this, unless someone can still see this bug with current CVS.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22756