Re: BUG #1518: Conversions to (undocumented) SQL year-month and

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> ... I don't think we are
> going to have to change any internal representations.

In order to support the spec *exactly*, we would have to. For instance
we cannot presently tell the difference between '13 months' and '1 year
1 month' ... they both end up stored as '13 months'. I can't say that
I find this very important, but it does put limits on how exactly we can
emulate the spec.

Note that I don't find "emulate the spec exactly" to be a desirable goal
anyway; its failure to consider daylight-savings issues is an
unacceptable shortcoming. The one internal representation change I
would really like to see is to store days separately from months and
seconds, so that we can distinguish '24 hours' from '1 day' (the latter
should sometimes be equivalent to 23 or 25 hours).

regards, tom lane