Re: Date/Time Classes

From: fkoh_at_[hidden]
Date: 2001-03-22 20:43:43

--- In boost_at_y..., Toon Knapen <toon_at_s...> wrote:
> "George A. Heintzelman" wrote:
> > 
> > > Speaking of implementation independent timing classes reminded 
me that I
> > > will very shortly be needing some way of representing dates and 
times in a
> > > project I am working on. Does anyone have any recomendations? 
This is the
> > > sort of library I would have thought would be ideal for boost.
> > 
> > I think this would be great to have standardized, even at the 
standard
> > time() granularity of one second. I've seen many tries at such 
classes,
> > none of them wholly satisfactory. A few things that I have 
observed
> > which I think should be taken into account in any such classes:
> 
> It's my impression there are two camps with an interest in 
date/time.
> 
> The first group needs it for 'business' applications to do global
> calendering operations.
> A second group wants it for timing routines/algorithms.
> 
> Where the first group needs a class that represents global time,
> the second group needs all the accuracy they can get. 
> A lightweight class using only one datamember
> (e.g. clock_t) can not satisfy both groups (datarange is
> too limited). Increasing the size of the class on the other
> could increase the overhead (which is of utmost importance
> to the second group).
> Maybe there is an approach which works for both but I just
> wanted to raise the point and look at the scope of the
> problem before thinking about a solution.
> 
> t

Looks like a perfect opportunity to apply the
ideas of the Czarnecki&Eisenecker book.