I was writing up the code samples for how to handle parsing with VVVV, and was getting some strange results. Parsing of VVVV has gotten completely and utterly hosed (in ICU4J, maybe also in ICU4C).
The problem is in the code after:
we have the line:
1744 switch (patternCharIndex)
It is supposed to switch based on the character. "V..." produces patternCharIndex == 29
But 29 is missing from the switch statement: all we have is:
1879 case 17: // 'z' - ZONE_OFFSET
1880 case 23: // 'Z' - TIMEZONE_RFC
1881 case 24: // 'v' - TIMEZONE_GENERIC
The addition of one extra line fixes the problem.
case 29: // 'V' - TIMEZONE_SPECIAL
I checked in a new test under trunk, in DateTimeGeneratorTest.java
Some additional updates in #5990.