Re: Informal Survey #1 -- joins on foreign keys

Dana utorak, 10. travnja 2012. 00:49:44 UTC+2, korisnik com..._at_hotmail.com napisao je:
> On Wednesday, 4 April 2012 16:06:47 UTC-7, vldm10 wrote:
>
> > We do not understand one another regarding some important issues.
>
> Unfortunately you don't understand my references to many that are basic, simple and common in logic and mathematics.
>
> > I also think that you pay attention only to the model. However db also works with real word objects.
>
> Predicate logic wffs and relational expressions correspond. Each connects the >database to the world by its predicate extensions holding the tuples of values >giving true propositions.

It seems to me that you claim that the Relational model solves some problems in mathematics and logic.

As I already wrote, “This whole story about "predicates-relations-statements”, has already been done by Gottlob Frege, 120 years ago.” See my message from March 6th 2012. Frege, in fact, did a lot more. He is the father of "Predicate Calculus" and modern logic. His Predicate logic is not limited by wffs. It is on the level of natural language. The RM is not on the level of natural languages.

>Thus every predicate and relation expression connects the database and the >world. Other connection is unnecessary. (Though perhaps useful.)

Sorry, but I disagree with this claim. One important part of this “connects the database to the world” is semantics. E. Codd hasn’t done anything significant within the field of Semantics. G. Frege was the founder of Semantics, and many other people have given significant contribution to it.

Another important part related to this “connection” is the logic. Let me mention A. Tarski and his definition of truth. Tarski worked on this definition for 10 years. Bela von Juhos also gave an important contribution to this definition of truth (where is the truth – in an object or in a meta language?)

E. Codd used truth tables because they work with compound propositions. Note that truth tables are on the intuitive level. In contrast to RM, my data model uses atomic propositions. Therefore the extensions correspond to atomic predicates. Note that G. Frege used trees for compound propositions, instead of the truth tables.

I also presented definition of the predicate, which is based on G. Frege work: A predicate is any incomplete phrase with specified gaps such that when the gaps are filled with names of things the phrase becomes a proposition.

Regarding to this definition, it is clear that G. Frege introduces the relation with named attributes, long before E. Codd.

This link between the database and the world, is one of the fundamental questions, which is only partially resolved. Famous mathematician in the last 100 years, have dealt with this subject. Some other sciences as philosophy, cognitive science and linguistics are also dealt with this topic.

So, if your intention is to show that E. Codd's relational model is able to fit into an existing mathematical theory, then I agree with you. But if your intention is to prove that the results that I mentioned above are Codd's contribution, then I'm sorry, I disagree. There is no doubt that E. Codd results from db theory have lasting value.

>Sorry but your papers are mostly incomprehensible and they misunderstand the >RM.

Here you are much generalized, so that no one sees what you mean. I publish my papers on the user group so that they are public and subject to criticism. Of course I do not think that I am infallible. But the criticism must be PRECISE and CONCRETE.

I'll try to explain my paper with a few extras. Section 3 defines the truth conditions. It also provides a link between meaning and truth. The truth conditions are often referred too by Frege, Godel, Tarski and Montague. Yet they do not define these terms. Here I have tried to define the truth conditions in a language that works with entities, relationships, attributes, states and events. Section 5 is about identification. For example OO approach extensively used identification, which is not defined in OO theory. RM also makes extensive use of identifiers, but there is no theory for them. Section 7 can be understood if you look at the theory of Abstract State Machine (ASM). As far as I know the two groups at Microsoft Research are working on this. Unlike them, my paper provides a concrete solution.

Vladimir Odrljin Received on Sun Apr 29 2012 - 20:46:04 CEST