On Mon, 27 Sep 2004 17:27:30 -0400, lex at cc.gatech.edu <lex at cc.gatech.edu> wrote:
> Tim Rowledge <tim at sumeru.stanford.edu> wrote:
> > Why return munged up seconds when quite often the image code
> > immediately works backwards to derive the day/hour/year etc? Let's pass
> > the prim a sensible date/time object aand fill in the
> > year/day/hour/minute/second/mSec/flavour-of-the-day-icecream etc. No
> > need to worry about epoch stuff. Leave the OS to care about timezone,
> > DST.
> >
> 

I like Tim's proposal.  I'd actually like two primitives that fill in
a date/time object, one for the OS and one for NTP.   It seems like it
would be helpful to have "VMTime now" and "NTPTime now" available...

> 
> To clarify one subthread of the discussion, I did not mean that the
> primitive should return a true mungification algorithm, but rather
> should return the UTC time plus the local time -- or equivalently, local
> time plus a delta to get to UTC time.  If we really go crazy, optionally
> let the VM stick a string in there to describe the time zone being used,
> e.g. "EST".

Sure, sounds good.