Re: Object-oriented thinking in SQL context?

Bob Badour wrote:

> Marshall wrote:
> 

> 
> I suggest a more appropriate venue might be comp.graphics.research
> ...

I don't know what this particular OT is, perhaps important-sounding newspeak for what I think of as optimization, a topic that usually seems to have surprisingly narrow scope as far as most people are concerned. I don't claim that recursive types are necessary, however TTM whiich I admire and consider as comparable to rdbms as Principia Mathematica was to the field of logic and numbers (and both are over-ambitious IMHO, also TTM possibly misleads people into thinking that a programming language is part of the RM) has some gems, such as Appendix A. My reading of Appendix A is that it does not preclude recursive relation definitions in any way, which I find very intriguing, knowing how careful D&D are. I think it's important to try to figure out the implications of Appendix A, especially the "formal definitions" when a recursive definition is involved. Not to say TCLOSE isn't useful, rather to know that a TCLOSE implememtation is complete in a logical sense. I don't like the treatments I've seen which always involve graphs which in turn probably makes people think they must implement graphs. Also, I do sometimes wonder whether such an exploration might not expand our understanding view of other relational ops such as projection. If I were running a research lab, I would certainly assign somebody to this.

> ...
...
>>> In another post he implied incorrectly that distributed database
>>> systems must support distributed transactions. Again this stems from
>>> ignorance.
>>>
>>> His only interesting contribution was a reference to a paper by Jim
>>> Gray that claimed update anywhere-anytime-anyway transactional
>>> replication was inherently unable to scale because reconciliations
>>> grow with cubic order. ,,,

The great intriguing idea that Jim Gray originated in the late 1970's or early 80's, I forget now, was predicate locks. They held the promise of a truly logical concurrency theory but I don't know that he developed them very far and the industry certainly lagged behind, not implementing even the most trival of logical locks until the 1990's. I once hired a university assistant prof to implement a fairly simply version but he had a nervous breakdown, convinced that his house and car were bugged and that the secret police were tailing him . Gray presaged this when he described lock managers as some kind of bottomless pit out of which no developer could ever climb. . Received on Thu Jun 18 2009 - 04:52:56 CEST