Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!cs.utexas.edu!uunet!odi!dlw From: dlw@odi.com (Dan Weinreb) Newsgroups: comp.databases Subject: Re: More on RDBMS's in CASE tools Message-ID: <1989Aug17.000608.27431@odi.com> Date: 17 Aug 89 00:06:08 GMT References: <3360@rtech.rtech.com> Reply-To: dlw@odi.com Organization: Object Design, Inc. Lines: 74 In-reply-to: dennism@menace.rtech.COM's message of 15 Aug 89 01:04:58 GMT In article <3360@rtech.rtech.com> dennism@menace.rtech.COM (Dennis Moore (x2435, 1080-276) INGRES/teamwork) writes: In the July 24th 1989 issue of "Digital Review," page 38, a product matrix indicates that INDEX contains an internal *R*DBMS (note: RDBMS, not DBMS). I spoke with the author of the article, who claimed that the matrix was derived from product literature and conversations with the companies. Unfortunately, my direct knowledge of the feelings of people at Index Technology come from discussions made under non-disclosure agreements. However, Index has recently announced publicly their intention to get involved with Ontologic. I don't remember the precise contents of their announcement, and should not risk misquoting them by guessing. However, I think they specifically stated that they intend to use Ontologic's forthcoming OODB product in their next generation of CASE tools. Anyone who is interested should find a copy of their press release and check it more carefully. Here's one interesting public statement that I can report. Burt Rubenstein, VP and co-founder of Index Technologies, was speaking at the Second International Workshop on CASE, on July 13, 1988, at about 10:00 a.m. (It's important to be very specific with facts, you know. I'm not precisely sure of Rubenstein's precise title, but it is something similar to "VP of Technology" or "VP of R&D". If you find that his title is something else, I'm sure you'll correct me. In this particular session of the workshop, Rubenstein was summarizing general conclusions. He recalled that one of the conclusions from the First International Workshop was: "Object-oriented database systems will be replacing relational database systems in CASE software over the next five years or so." (Just in case you happened to be there with a tape recorder, I supposed I had better point out that this may not be an exact, word-for-word quote, but it is very close to one, and does not distort the meaning in any way.) You can decide for yourself the extent to which he is speaking for himself, for speaking for his company, or saying what he thinks many other people think. He did not elaborate on the point at that particular time. I will go out on a limb and repeat a statement that I was told by a senior Index Technology technical manager, while I was in a non-disclosure meeting at Index's offices. I don't think they'll mind, and if they do I apologize. We asked what database technology they use in their present products. He named a particular B-tree package; it's one of the ones that's often advertised in magazines, and you see it mentioned in comp.databases from time to time. He said "That's all we use. There is very little database technology in our existing product." I don't know how to reconcile this with the article you saw in Digital Review. I can suggest one possibility: there is no completely universally applied litmus test as to what constitutes a "relational database managegment system" and what does not. It may simply be a matter of interpretation. You'll have to decide for yourself whether the Digital Review article should be read to imply, for example, that there is a non-procedural query languages such as SQL or QUEL happening somewhere in the existing Index Technology product. I don't think anyone was lying at any point at all; rather, I think that the accepted standards for what the question "do you use a relational database" means are generally very broad, and in this case there is a valid sense in which the answer is "yes" and also a valid sense in which the answer is "no". And perhaps there is some other explanation of the discrepancy. Rather than worry about it, however, it would be more worthwhile to explore the real interesting technical heart of the matter: what are these so-called OODB's, and why does anyone think they're good, and for what specific purposes do they claim to be well-suited? I expect to address these points further in future postings, since it seems as if there's at least a noticable number of people reading comp.databases who are interested in these questions. Dan Weinreb Object Design, Inc. dlw@odi.com