Re: O'Reilly interview with Date

From: Kenneth Downs <knode.wants.this_at_see.sigblock>
Date: Sun, 14 Aug 2005 10:27:32 -0400
Message-Id: <ft04t2-eoe.ln1_at_pluto.downsfam.net>


Paul wrote:

> Kenneth Downs wrote:

>>>b) a single interval column? Then you have a problem joining to the
>>>start date of the interval. The interval is atomic, but you could have a
>>>start() function that picks out the start of the interval. Even so, this
>>>ceases to be a standard join.
>> 
>> Not really, it is standard for the interval type.  Joining single to
>> interval says single must be within the interval.  At least this is what
>> you come up with if you are thinking of expressing intervals as uniques
>> and references.

>
> What I'm saying is if you want separate joins both to the interval *and*
> to the start or end dates.

OK.

>
> For example, thinking of booking holidays or something: there might be a
> supplement to the price if the duration spans a bank holiday, say, so
> you'd need a "join" to the interval.

Hmmmm.

>
> But also there might be a reduction to the price if the holiday starts
> on some dates specified in another table, so you'd want a join to the
> start date as well.
>
> Then you could have either one of the joins as standard in your setup,
> but not both.
>

These examples alas only underscore my own lack of clear understanding of what the complete treatment of intervals is. The examples I've offered so far come from my own experience, but that is of course too small a sample set.

-- 
Kenneth Downs
Secure Data Software, Inc.
(Ken)nneth_at_(Sec)ure(Dat)a(.com)
Received on Sun Aug 14 2005 - 16:27:32 CEST

Original text of this message