emergency.shower@gmail.com wrote:
> Everything seems to be OK, except that read operations from the
> America/St_Johns server lead to a deviation of minus half an hour,
> probably due to a rounding error in the driver's getTimestamp
> implementation.
Ok, it looks like there is another longstanding bug in TimestampUtils
that I didn't find: it just doesn't parse timezones with non-zero minute
offsets at all (there's code there that tries to do it, but it's incorrect)
I'll commit a fix to CVS shortly, once I've sorted out a test to go with it.
-O