Re: Decline of Science: Computer Science and Databases

From: Leandro Guimarăes Faria Corsetti Dutra <lgcdutra_at_terra.com.br>
Date: Thu, 31 Oct 2002 14:25:58 +0100
Message-ID: <aprb17$46dpn$1_at_ID-148886.news.dfncis.de>


Carl Rosenberger wrote:

> To me the documentation looks as if it could take some time to get
> started:
> http://docs.alphora.com/Alphora.Dataphor.DAE.Language.D4.html

        Yes, it does. The documentation is still a job in process, and the fact that they still rely on a separate (usually SQL) storage layer instead of being a full RDBMS adds some complication.

        Not to mention the mindset shift from SQL to relational…

> People are usually not willing to work their way into a proprietary
> API that needs a year of studying to understand the concepts.

        Actually one needs "a year of studying" to became a relational expert.   That will be it for the concepts, since Dataphor adds nothing conceptually to the relational model. Other than that, it's just a D implementation, so you don't need one year "to understand the concepts".

        Also, no need to learn an API just to get the feel of it. It's a data language that can be used interactively, so the API is only needed if you really want to program with it.

        About being proprietary: it is not free software, but the language is well defined and therefore open to reimplementations. So in 20 years time there could be an ISO D4 committee.

> Have you been working with Dataphor, Leandro?

        Just toying around…

> Can you give us a short sum-up of the benefits with maybe a few nice
> examples?

        Well, it's the first implementation of the relational model since UCB Ingres QUEL and IBM BS12, and the field has progressed a lot since that time.

        For me the biggest benefit is that its "tables" are true relations, not like SQL bags. This gives you "view" (derived relations) updateability, and that in turns gives data independence.

        Also, there is support for type hierarchies with inheritance thru S by C, and these types are user-defined. That gives great support for both OOP and "rich data types".

        For the purist, this is a faithful (but honest about current limitations) implementation of D&D's TTM D.

-- 
  _
/ \ Leandro Guimarăes Faria Corsetti Dutra        +41 (21) 216 15 93
\ / http://homepage.mac.com./leandrod/        fax +41 (21) 216 19 04
  X  http://tutoriald.sourceforge.net./      Orange Communications CH
/ \ Campanha fita ASCII, contra correio HTML      +41 (21) 644 23 01
Received on Thu Oct 31 2002 - 14:25:58 CET

Original text of this message