Abstract:
In 2005, I wrote an article in which I discussed the most important features of the standards ODMG 3.0 (ODMG object model) and the SQL:2003 (SQL data model) and convincingly (as it seemed to me) argued that the similarity between the object model and object extensions to SQL is purely external, that close in mind syntactic constructions hide deep differences at the model level. Examples of such differences include von Neumann-style dereference of ObjectIDs in the ODMG model vs join-style dereference of reference values in the SQL model, separate and independent store of objects of one and the same object type in the ODMG model vs store of all raws of a typed table (SQL analogy of object) within this table, store of ObjectISs within extents in the ODMG model vs store within analogy of extent of objects their self in the SQL model, etc. Since then, it took many years for which I understood many things that were wrongly or insufficient correctly understood by me then, and gradually came to the conclusions that:
differences that seemed to me deep, are not such, and indeed are not differences at the model level;
the object extensions of SQL provide no less (and rather large) capabilities than the ODMG object model;
reasonably (from the standpoint of the database community) used DBMSes based on the ODMG data model, one will create databases and tools to manipulate them similar to those prescribed by the SQL data model.
Fulltext:
PDF file (271 kB)