0
$\begingroup$

In 1995, two DB theory researchers, C.J. Date and H. Darwen, published a text with the rather impressive-sounding name "The Third Manifesto" with their ideas regarding future DBMSes, which would better harmonize a relational model with object-oriented programming, including user-defined types.

It is my understanding that object-relational DBMSes has been mostly/entirely abandoned as a subject of research. So, has this manifesto had a significant impact on the DBMS world community over the past 23 years? If so, what are the salient aspects of this impact?

$\endgroup$

1 Answer 1

2
$\begingroup$

Hmm, before some moderator closes this q as being a matter of opinion, not substance ...

I think you need to consider whether the first two manifestos (to which TTM was a response) had any impact? I'd say little or no: there were a few OODB ideas that made it into the SQL standard (but were little implemented). There are a few niche commercial products that claim to be OODBMSs. But mostly the industry has stuck with SQL tables; and applications or development tools provide an ORM layer (Object-Relational Mapping).

A comparable question: have 'Codd's 12 Rules' had a significant impact on the DBMS world? I'd say not much, even though most industry practitioners would at least be aware of them.

Hugh Darwen, by the way, is not principally a "DB theory researcher". He was part of the UK development team for an early IBM relational system Business System 12 -- which arguably was more relational than SQL has ever achieved [**]; and he was IBM's representative on the SQL standard committee for many years. He only lately moved into academe. His first draft of TTM was a visceral reaction against the fulminations of the SQL committee, written while he was at one of their plenaries.

Chris Date, IMO, is not much of a researcher either: he's more of an educator and writer of textbooks. I'm unconvinced he's added much to the research rigour of the TTM document.

Apart from that, your q seems to me to show many confusions.

If you read TTM (and the supporting explanations), I see no intent to "harmonise" with OOP. Note that an OODBMS is not aimed purely at OOP's: it's aiming to support notions of encapsulation of data. TTM's claim is that a Truly Relational DBMS already includes sufficient means to encapsulate data -- including under the Liskov/Wing Substitution Principle; and that SQL has bastardised the applicability of the Relational Model. Separation of concerns is principally to be achieved through schema design; we don't need perversion or subsumption of the Relational Model.

What has support for user-defined types got to do with it? Any modern language in any paradigm has those. The usual catchcry is 'types are orthogonal to the model'.

A large part of the TTM Prescriptions and Proscriptions are taken up not with opposing OODBMSs, but opposing SQL. Has TTM had a "significant impact" on SQL? I'd say no: SQL has continued to get worse and worse. Each revision of the standard has retained all the old awfulness (due to the shackle of backwards compatibility); has added new awfulness; and has multiplied up the effective awfulness in trying to support multiple 'features' that are in tension.

There are a few products that have attempted to implement TTM's ideas -- see "Projects" linked from the TTM website. I'd describe them as 'hobby products' developed and supported by individuals -- individuals who are rather cranky, strongly opinionated and difficult to deal with IMO. I would never recommend anybody use these products. (And for that reason, I'm not going to suggest you go ask your question at the TTM forum. You have been warned.)

[**] The more I discover about BS12, the more impressed I am by how much it got right even when the Relational Model was very far from mature. How come it was killed off but SQL continued? IBM internal politics, plus commercial pressure: Larry Ellison stole some significant brainpower from the IBM System/R research team (who arguably never understood what Codd was talking about); they re-implemented at Oracle what had been only a prototype at IBM; then commercial pressure resulted in a race to get 'something' to market, and never mind the quality. Mike Stonebraker/Ingres did originally try harder to follow the model, but eventually had to cave to commercial pressure.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.