Re: A Logical Model for Lists as Relations

dawn wrote:
> Marshall Spight wrote:
> >
> > What about imperative operations? This is a bit more complicated.
> > We have to consider whether we want to enforce the usual
> > list semantics, where for example if we delete an element, the
> > higher indicies are accordingly reduced.
>
> You likely want both operations - a delete and a ripple delete. A
> deleted list value without the ripple delete would make that value a
> "2-valued-logic null".

An empty set.

> You wouldn't want to have only the ripple
> delete in case there is an application need to remove a value in one
> transaction before "replacing" it in another. No need to renumber,
> only to do it again with an insert.

What's wrong with "update" in that circumstance?

> By the way, what are your goals/requirements for reinventing this wheel
> -- what are MS, Intersystems, IBM... not doing that you want to do, or
> are you just trying to roll your own? Just curious. One guess is that
> you want declarative constraint handling, perhaps? I don't know if
> anyone has that in combination with list-valued attributes.

Well, yeah I do want to roll my own. From the beginning I've wanted a more algebraic algebra, more general declarative constraint handling, more embedding of proofs in code, higher order functions, stuff like that.

Marshall Received on Wed May 10 2006 - 16:10:45 CEST