Our class library currently gets its default TimeZone from the system
property "user.timezone". This isn't documented in the Javadocs for 1.2,
though it exists in the JDK (I can print it out) and it is listed in a few
bugs in the BugParade.
Japhar isn't setting user.timezone, though I wrote a trivial patch to do
so. My real question is whether we should be relying on the VM to set this
and even using this property at all. If Japhar sets it, they will likely
want to get it to the value used by the JDK, which I don't think is what we
want. For me, the JDK says my user.timezone is "CST" but this is not enough
information to completely specify the timezone. I'd prefer something like
"CST6CDT" for example.
Anyhow, I was thinking that maybe we should have a static initializer that
looks up the
default timezone ourselves. This would require native code of course. I
think this gives us the best control. The only problem it would create is
that the user couldn't override the default timezone on the command like
with a -D option.
Thoughts?
--
Aaron M. Renn ([EMAIL PROTECTED]) http://www.urbanophile.com/arenn/
