Re: ID field as logical address
Date: Fri, 29 May 2009 19:41:11 GMT
Bob Badour wrote:
> If it is unnamed, how would you refer to it? Names are important things.
That was sort of the point, I wouldn't need to refer to a set-valued attribute. One motive had to do with finding a recursive structure that might implement TCLOSE more efficiently at least in some cases. Another had to do with missing information. Of course the admittedly vauge idea I had in mind would mean that the names in the tuple triples would need to be unique within the relation as a whole and I would certainly need to come up with a definition for this that would be of similar quality to TTM's Appendix A, which I never succeeded at.
There is very little implementation theory, for example about optimization, most of what I've seen is adhoc. Received on Fri May 29 2009 - 21:41:11 CEST