RE: [boost] Date/Time formal review

From: Jeff Garland (jeff_at_[hidden])
Date: 2002-04-16 21:46:22

Glen Knowles wrote:
>Would it be reasonable to require that all date systems to provide:
>1. conversion to a basic counted system with resolution and base of its
>choice

No, I don't think so.  This isn't a current requirement.  The internal
representation used with the gdtl::time template can be an integer, struct, or whatever you want.
That said, I expect most systems will use a
counted representation perhaps augmented with some extra data (say a timezone
spec, dst flag, etc).  Posix_time uses a counted representation.

>2. the resolution of its basic counted system

Resolution of time durations is already a requirement.  The static function 'resolution' specifies
it.

>3. its base in a universal time format

As we add additional time systems I would typically expect conversions to UTC.  But why does this
need to be a 'requirement'?  Seems to me, if it is useful and reasonable to build it will be built.
If there is some reasonable infrastructure that can help multiple time systems perform this
conversion then we will build it.  I believe all of these things are pure extensions of the current
library.

Jeff