From: Dirk Schreib (Dirk.Schreib_at_[hidden])
Date: 2002-04-24 01:42:56
Based on the discussion about this topic here I vote for NO.
I don't care for the internal implementation but the interface
should be relatively stable, which is not the case.
First of all I would like to see a dependency diagram
like this (dependency != inheritance)
simple_date (Gregorian?, no DST, no LS, ...)
The most effective way (and sufficient for many (most?) applications) is
to have a "floating" date base. That means
simple_date( "now" ) == local_date( "now" ). But for
dates 4 years old (before 1/1/99) you will miss the leap seconds.
If you try it the other way around with TAI as a base you always
have to pay a prize in adding the leap seconds (32? in the moment).
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk