Re: S.O.D.A. database Query API - call for comments

On Thu, 03 May 2001 01:35:07 -0700,
"Lee" == Lee Fesperman <firstsql_at_ix.netcom.com> wrote:

Lee> Tobias Brox wrote:
>>
>>
>> I think it should be possible to make an object oriented database by
>> building "object-orientedness" at the top of an RDBMS. And I think it
>> should be possible to make it just as good as building a truely object
>> oriented database from scratch. But I'm not going to prove that - at
>> least not right now ;-)
>>
>> There are powers both within the relational model and the object model,
>> and as far as I see, they don't contradict each other that much. One
>> class (without inheritance) is a table, one object (without inheritance)
>> is a row in the table. Inheritance is just a special case of a link
>> between two objects - one object without inheritance, and another object
>> just containing what's added in the child-class.
>>
>> Methods should be stored in the database somehow, and should replace the
>> old triggers and stored procedures.

Lee> Storing a table row as an object messes up the query language. How about
Lee> storing objects as column values in table rows. Then you can keep a
Lee> declarative query language and be able to call methods on columns. Our
Lee> ORDBMS (in beta) does this - http://www.firstsql.com/firstsqlj/

Darwen & Date declare that this is the only correct way to work with objects and relations -- you're in good company:

  [darwen95]
  HUGH DARWEN AND C.J. DATE,
  The Third Manifesto,
  SIGMOD RECORD 24(1):39-49, March 1995.   

  The third manifesto aims to supercede the first two, [atkinson89] and   [cadf90]. A rather technically oriented document, it formally defines the   main properties of a hypotetical database language D. This language has a   strong foundation in the relational model and has other characteristics   that enrich its type system by providing user-defined types for   domains. The belief of the authors is that the relational model can   accommodate the required object oriented extensions without being corrected   or perverted. The paper firmly states that SQL is "a perversion" of the   relational model and makes it clear that D should be based on more formal   and abstract grounds. D accesses the database only through declarative   queries. The paper is more formal and precise than the other two manifestos   and it describes an appealing, robust system. Unfortunately it doesn't   exist yet!

I found a copy here:

On this page -- which seems to have a quite few good links on the subject --

  http://citeseer.nj.nec.com/darwen95third.html

I found a link to the original manifesto:

  http://citeseer.nj.nec.com/rd/86668813%2C107209%2C1%2C0.25%2CDownload/http%253A%252F%252Fwww.uk.research.att.com/%257Efms/db/darwen95.ps.gz

And stop press, there's a discussion on this going on now on: slashdot: http://slashdot.org/article.pl?sid=01/05/03/1434242

Cheers,

                                                                      Philip

-- 
If you have a procedure with 10 parameters, you probably missed some. (Kraulis)
-----------------------------------------------------------------------------
Philip Lijnzaad, lijnzaad_at_ebi.ac.uk \ European Bioinformatics Institute,rm A2-08
+44 (0)1223 49 4639                 / Wellcome Trust Genome Campus, Hinxton
+44 (0)1223 49 4468 (fax)           \ Cambridgeshire CB10 1SD,  GREAT BRITAIN
PGP fingerprint: E1 03 BF 80 94 61 B6 FC  50 3D 1F 64 40 75 FB 53