Re: Storage sizes for dates/times (documentation bug?)

Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
> Which means that storing date + timetz in two separate columns is not
> quite the same as storing a timestamptz. Oops.

Quite so. Our docs already point out that timetz is really a completely
brain-damaged concept, anyway.

There's been some talk of adding an explicit zone representation to
timestamptz, but so far I haven't been convinced that it's worth
doubling the storage requirement (which is what it would take,
considering alignment...). ISTM that we have defined timestamptz
in such a way that it solves many real-world problems, and timestamp
also solves real-world problems, but the use-case for a timestamp plus
an explicit time zone is much less clear.

regards, tom lane