Forums

Forum Home

Object-Role Modeling

ORM2

proposal: to un...

proposal: to unambiguate ORM terms from the p.v. of O., R. and M.

Guennadi V. Vanine

170 posts

Joined
5/28/2005

proposal: to unambiguate ORM terms from the p.v. of O., R. and M.
Posted: 25 Jun 05 8:04 AM Modified By Guennadi V. Vanine on 6/25/2005 8:14:06 AM)

I reprint here some ORM definitions:

Relation - a named table with a set of named columns and populated by a set of unnamed rows

Role - a part played ny an object type (again by a population, i.e. many objects?!) within a prediicate

Uniqueness - (sorry I just Copy&Pasted this "definition") - prevents duplication of role instances

These are really definitions how they are used (and not just proclaimed) in ORM. Meanwhile there are quite a bunch of other "definitions" with the use of underdefined, ambiguous and contradictory to each other terms and implications.

I understand that this is very comfortable to have a named table with unnamed rows for persistence and storage,

but if ORM is about Object , Role and Modeling than the notions/terms should not be about tables, rows and populations but about roles (and relations) played by each instance of an entity type:
for ex., Each Person(or ssn number) was born in how many Countries (i.e. CountryCodes)? In each Country (i.e. CountryCode) how many different Persons(i.e. ssn numbers) were born?

Really, such understandig and even such types of questions appear in ORM (and VEA, VisioModeler, InfoModeler), though without any real utilization and being immediately completely inverted into the terms of RDBMS persistence, i.e. instance populations like column table uniqueness but not object (of the same instance) role uniqueness
While the RDBMS conveniences shouldn't even appear at conceptual level modeling.

This is very nice to have uniquenes in column of persisted table. But in modeling I am interested, first of all, in uniqueness of roles (role instances) PER (one, each of the) instance/object. The same role instance among population (many different instances of object entity) is very helpful for storage but not for modeling.

Meanwhile there is a big confusion and even absence of unique unambiguous definitions of terms in ORM. This is also in odds with UML while it is written that ORM notation and terms are equivalent (ueaah, they are just proclaimed so).

Then appear obvious problems in tools mapping, interoperability which are attributed to tools implementation bugs and underdevelopment while they are theoretical ambiguities impossibilitating any simple and obvious practitioning .

thanks god, i.e. Scot, saving the ORM cause in all EMEA by patiently answering ALL questions, Геннадий Владимирович Ванин (aka Guennadi Vladimirovitch Vanine or Vanin Gennady )

Page 1 of 1

Object-Role Modeling

ORM2

proposal: to un...