>> Anyway, I have no idea who's DateAndTime is better.
>
> Actually, I've tried both, though Brents has been the second, and by pure 
> coincidence it's the one I tested more.
>
> I've had no problems with either, though I'm not sure wether what they do 
> is "the Right Thing" (tm), as it's all I know ...
>

My DateAndTime and Duration implementation are both 100% ANSI compatible. I 
confirmed this by using the ANSI SUnit tests on SqueakMap.

My implementation of DateAndTime has nanosecond precison (although 
millisecond precison is all the vm will currently provide).
My implementation is however a superset of this functionality, in that it 
is integrated with the Smtalltalk-80 Date and Time classes.

It is also integrated with the Squeak extensions: Week, Month, year etc.

My motivation was

1. to prevent ever having to write another timestamp class.
2. to never ever make another mistake at the iteration boundaries by 
exploiting do: [] semantics of Smalltalk collections.

Chronology 1.3 is under development, which it intended to support Magma.

Cheers

Brent