[kaffe] Bug Report in GregorianCalendar.set or getMinimum

Dalibor Topic robilad@yahoo.com
Mon, 15 Jul 2002 01:59:08 -0700 (PDT)


Hi Christian,

--- Christian Hammers <ch@lathspell.de> wrote:
> GregorianCalendar is implemented unixtime based so
> limited to 1970 - 2037.
> But the actual bug is that getMinimum(Calendar.YEAR)
> reports 0 as minimum,
> set() accepts 1800 and println prints it, but as
> soon as I do an add() the
> date jumps to 1970. 

Thanks for your bug report. As I'm not too familliar
with kaffe's implementation of
java.util.GregorianCalender, I need some help to
understand the issue.

It seems that the error is introduced in
computeFields, where the 1970 comes into play. Could
you come up with a simple test case that triggers the
bug? I'd like to put it in our regression test suite,
if you don't mind.

> So either 
>  - make it non unixtime based
>  - let getMinimum return 1970 (dito for getMaximum
> etc)
>  - throw exceptions

I think the best option would be to make it
non-unixtime based, as otherwise the use of dates like
GREGORIAN_CHANGE is limited. Would you be interested
in helping me fix those bugs, as you probably know
more about Calendars than I do?

best regards,

dalibor topic

__________________________________________________
Do You Yahoo!?
Yahoo! Autos - Get free new car price quotes
http://autos.yahoo.com