David Fetter <david@fetter.org> writes:
> On Tue, Dec 15, 2009 at 11:31:05AM -0800, Scott Bailey wrote:
>> As for the extra bits, would it be better to just require continuous
>> ranges to be either [] or [)? But I don't know which would be
>> preferred. My inclination would be toward [), but Tom seemed to
>> indicate that perhaps [] was the norm.
> [] makes certain operations--namely the important ones in
> calendaring--impossible, or at least incredibly kludgy, to do. I
> think we ought to leave openness at each end up to the user,
> independent of the underlying implementation details.
Yes. A range implementation that couldn't support all four cases
of [], [), (], () would be seriously crippled IMO.
regards, tom lane