Re: thinking about UPDATE
From: Jan Hidders <jan.hidders_at_REMOVETHIS.pandora.be>
Date: Wed, 21 Jul 2004 21:56:33 GMT
Message-ID: <pan.2004.07.21.21.57.18.40106_at_REMOVETHIS.pandora.be>
Date: Wed, 21 Jul 2004 21:56:33 GMT
Message-ID: <pan.2004.07.21.21.57.18.40106_at_REMOVETHIS.pandora.be>
On Wed, 21 Jul 2004 16:46:44 +0000, Marshall Spight wrote:
>
> I note F' has type a1 ... an -> a1 ... an, or more generally S -> S
> where T : {S}. I guess C would be called the characteristic function of
> the update set, yes?
Not exactly. The update set (the set of updated tuples) is in fact the intersection of the relation and the (possibly infinite) set that is characterized by this function.
> Now I'll try to extrapolate from there to think about INSERT OR UPDATE.
> I suspect the key lies in generalizing type of the map function F' from
> S -> S to S -> {S}, or something similar.
Interestingly enough the (list) calculus on which XQuery is based more or less does it that way for selections/deletions.
- Jan Hidders