Re: MERGE as the imperative form of aggregation

From: Marshall <marshall.spight_at_gmail.com>
Date: 16 Apr 2007 07:36:19 -0700
Message-ID: <1176734179.142622.193360@p77g2000hsh.googlegroups.com>

On Apr 16, 6:57 am, paul c <toledobythe..._at_oohay.ac> wrote:

> By the way, I don't think assignment is necessary, therefore it's not
> primitive. But I have no idea whether it's primative!

As a quick aside, what is primitive or not is a matter of design. No operation is or is not fundamentally primitive; it depends on the system that operator is embedded in.

Relational assignment could be the only imperative operation supported, and insert and delete could be defined in terms of it. In which case, assignment is primitive, and insert and delete aren't. It is equally the case that insert and delete could be the only imperative operations supported, and assignment could be defined in terms of them, in which case assignment is not primitive and insert/delete are.

We could have a boolean algebra system which supplied only NAND, in which case it would be primitive and everything else would be derived. We could equally well substitute NOR, or the pair (AND, NOT), or we could put in more than we need and use (AND, OR, NOT) etc. etc.

Marshall Received on Mon Apr 16 2007 - 09:36:19 CDT