michael.daly@canrem.uucp (michael daly) (01/29/91)
Some may complain that this is not an OO topic, but I feel it is. I have not found any information on data modeling in any object oriented design/analysis/programming textbook. I have recently finished a contract on the construction of an application for a mainframe using object oriented design techniques. It was coded in PL/I, with programmers creating objects and (most of them) enforcing OO concepts like encapsulation etc. There were two concepts used by the architect of this system that seem uncommon in OO environments - relational objects and data modeling. I don't want to discuss the relational objects (a solution to the dilemma of OODB vs RDBMS?), but I am having a problem with data modeling. I joined the project after the data modeling was done and had no direct exposure to the techniques. I have a generic understanding of the end result of data normalization, but don't know how to do it. I have been told that there is a "cookbook" approach that will result in a reasonable data model (at least up to third normal form, I think). This approach can be specified in a few written pages. Can anyone offer some advice on this topic, especially in an object oriented environment? +------------------------------------------------------------------+ | Michael Daly | Usenet: michael.daly@canremote.uucp | | | Internet: michael.daly%canrem#lsuc.uucp | +------------------------------------------------------------------+ --- ~ DeLuxe}ab #289 ~ -- Canada Remote Systems. Toronto, Ontario NorthAmeriNet Host
wlp@calmasd.Prime.COM (Walter Peterson) (02/01/91)
For an OO approach to data modeling take a look at: Runbaugh, Blaha, Premerlani, Eddy and Lorensen "Object-Oriented Modeling and Design" Prentice Hall, Englewood Cliffs, NJ, 1990 ISBN 0-13-629841-9 -- Walter L. Peterson | ______ Internet : wlp@calmasd.Prime.COM | \ / CompuServe : 70441,3177 | \ / Silence = Death "The opinions expressed here are my own." | \/
haim@taichi.uucp (24122-Haim Kilov(L028)m000) (02/06/91)
The idea of object information modeling is great and has in fact been
used for some time. The generic object information modeling concepts
superficially look like extended ER ones, but differ in at least the
following manner:
- they are precisely defined
- they are implementation-independent (i.e., don't use attributes,
normalization, and other relational concepts)
- they are not "CASE tool"-oriented
- they are extensible.
The definitions of these concepts are based on primitive (CRUD) behavior,
i.e., on operations jointly owned by an object and associated objects.
A paper about this approach will be presented at TOOLS'91 in Paris in March
("Generic information modeling concepts: a reusable component library"
by myself). We at Bellcore use this approach.
Hope this helps.
-Haim Kilov
haim@bcr.cc.bellcore.com