(tz) Wrong behavior JDK 1.3.1 in GregorianCalendar at DST boundary time

Customer Problem Description:
-----------------------------

FULL PRODUCT VERSION :
java version "1.4.0_01"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.0_01-b03) Java
HotSpot(TM) Client VM (build 1.4.0_01-b03, mixed mode)

FULL OPERATING SYSTEM VERSION :
Microsoft Windows 2000 [Version 5.00.2195]

A DESCRIPTION OF THE PROBLEM :
If I initiate a GregorianCalendar to the time at DST
boundery I got a different output with JDK 1.3.1 and 1.4.


REGRESSION. Last worked in version 1.3.1

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
Just run this small programm with JDK 1.4 and then with
1.3.1



EXPECTED VERSUS ACTUAL BEHAVIOR :
JDK version utc time
1.4.0_01 returns 1017539999000 (Sun Mar 31,2002 3:59:59
Europe/Berlin)
1.3.1 returns 1017536399000 (Sun Mar 31,2002 1:59:59
Europe/Berlin)


REPRODUCIBILITY :
This bug can be reproduced always.

---------- BEGIN SOURCE ----------
// DST switch on 31 March 2002 at 2:00:00
//
// |-------|-------X-------|-------|
// 12:00 01:00 03:00 04:00 5:00
// ^
// |
// daylight switch from 2:00 to 3:00 by Europe/Berlin

Calendar cal = new GregorianCalendar(2002, Calendar.MARCH, 31, 2, 59, 59);
cal.setTimeZone(TimeZone.getTimeZone("Europe/Berlin"));
long time = cal.getTime().getTime();
System.out.println(time);

How to reproduce it:

  // DST switch on 31 March 2002 at 2:00:00 by Europe/Berlin
  //
  // |-------|-------X-------|-------|
  // 12:00 01:00 03:00 04:00 5:00
  // ^
  // |
  // daylight switch from 2:00 to 3:00

To reproduce you have just to run this with JDK 1.3.1 and JDK 1.4 to see the
difference:

  Calendar cal = new GregorianCalendar(2002, Calendar.MARCH, 31, 2, 59, 59);
  cal.setTimeZone(TimeZone.getTimeZone("Europe/Berlin"));
  long time = cal.getTime().getTime();
  System.out.println(time);

JDK version utc time in ms
1.4.0_01 1017539999000 (Sun Mar 31, 2002 3:59:59 Europe/Berlin)
1.3.1 1017536399000 (Sun Mar 31, 2002 1:59:59 Europe/Berlin)