Re: The OverRelational Manifesto ("TheORM"). Needs to be debunked.

From: U-gene <grigoriev-e_at_yandex.ru>
Date: 4 May 2006 04:48:38 -0700
Message-ID: <1146743318.244091.126850_at_j73g2000cwa.googlegroups.com>


But where I wrote, that "TheORM" is about new data model? Show me my these words and I will say "I'm sorry". Of course it is about special kind of relation system, which has special kind of relational language. OK, somebody doesn't like term "object" - I will use term "moleculare semantic unit" (MSU), so it can be named as relation data base MSU-oriented managment system for (may be better is MSU-oriented managment system for relation data base) and same language.

But what doesn't you like when I say about two orthogonal projections?

I'll try to explain it once again using old analogy I met long years ago. Some guy (I don't remember who he was) sayd, that using RDBMS with OO system looks like the disassembling of car during its garaging. When we use a car (object) it is integral, whole thing, but when it is in garage (in RDB) - it is a set of details.

I'm not sure if this comparition is correct. It seems like this guy've forgotten that both RDBMS and OO-system are just a programs, which executed on computers, where all data are stored in set of very simple variables (= Phisically car is just set of simple atoms). OO-systems and RDBMS offer two different ways to combine these very simple physical variables into complex logical variables, and these variables are really different. So we have one set of atoms (formed as objects) on OO-side and second set of atoms (formed as relvars)

The fact is that the atoms themselves are very-very similar on both sides. So it is possible to create system where only one set of atom will be formed at the same time as set of objects (I'm sorry - MSU) from one point of view and as set or relvars from other, ortogonal one. Both these points of view is described with only one structure definition. It means that both MSU-variables and relvars (and they structures) are named with the same names.

The last is one of key points of "TheORM". It means that developer, which works with data organized in this way, doesn't really have to think what projection he use to access with data. He just has to use correct operation with correct variable names. If the operation is group one based on RM operation (For example SELECT) it means that manipulations on data occures through R-projection. if the operation is group one based on RM operation it means that manipulations on data occures through R-projection. The operation, which looks like opertion using in OO-system (for example creating some MSU item), means manipulation through MSU-Oriented projection. But from developer's point of view it means, that he can use both relational and MSU-Oriented operations right after some comlex structurewas described. Received on Thu May 04 2006 - 13:48:38 CEST

Original text of this message