Hotsos Symposium 2009 Day 1

Blackboard is in the house for the third consecutive year…”Aw Yeah!” I am pretty sure that’s what was in my head Monday morning when I stepped into the keynote by Chris Date. He gave a 3 hour keynote and session on Database Foundations.

Mr. Date is the author of one of the most famous database texts available on the free market, Introduction to Database Systems. He also has a few other texts that are going to possibly make my “Geek List” for reading.

Presentation #1: Foundation Matters

I won’t even attempt at saying this guy doesn’t know his stuff. He’s like the father of relational databases. Just look at his bio and his publications. I will say I think the guy might be a little off-balance. I’ll explain a little more below…

The presentation started off somewhat interesting. He talked about how amazed he was that so many database vendors and even database professionals are “misguided” about relational theory. His main argument is that bad teaching has caused bad learning amongst database practitioners. Most importantly that database practitioners and software developers lack a fundamental understanding of relational database foundations. Read slides 2 through 6 for his examples of lack of foundation.

Date and a co-author, Hugh Darwen have put what they call The Third Manifesto which Date calls a detailed study of the impact of type theory on the relational model of data, including a comprehensive model of type inheritance. This came out of three reasons:

  1. Equating relations and object classes (“The First Great Blunder”)
  2. Mixing pointers and relations (“The Second Great Blunder”)
  3. More generally, failing to understand relational theory

As Date says, “The Third Manifesto is evolutionary, not revolutionary: Builds on (and tidies up) the relational model – not an attempt to replace that model…” It apparently came as a response to The Object-Oriented Database System Manifesto and Third-generation database system manifesto by Michael Stonebreaker.

Key Takeaways…

  • Speaker was very anti-SQL. He equated SQL to Cobal in that there will be a migration to a “better” language for more appropriate and correct “relational” communication. Yet he failed to explain what that language would be. He just reiterated that SQL was not the answer.
  • In the relational world, VALUES are immutable and cannot change.
    • eg: Table has 3 rows, then insert a 4th
    • Therefore table is mutable therefore is a variable not a value.
  • Circle is really an Elipse (Who didn’t know that?)

Lost Credibility

He lost credibility with me about two things. The first was his denial of SQL as an appropriate language to interact with relational database systems. He could have made a good point if he provided an alternative language that could and should replace SQL. The second thing was referencing this company called Required Technologies, Inc. in which he said this company built what’s called a Trans Relational Model. As Date said, this company built a relational system that claimed to have no DML or DDL, not to mention a CBO or joins. For some reason I felt like the only person in the room saying this guy is either senile or foolish. Well it turns out there are others who think the same as me.

Advertisements

5 thoughts on “Hotsos Symposium 2009 Day 1

  1. Hugh Darwen

    “He could have made a good point if he provided an alternative language that could and should replace SQL.”

    Didn’t he mention Tutorial D, the language he and I devised and describe in our book on The Third Manifesto? Because we devised it for teaching purposes, it is missing certain features that would be needed for commercial use, and certain matters of detail are left undefined in the grammar, but there is no doubt it could easily be adapted. And there are various other prototype implementations of The Third Manifesto–details at http://www.thethirdmanifesto.com.

    Reply
  2. babyshark13 Post author

    I believe Tutorial D was called out in his slides, but Mr. Date skipped past it during the speaking portion of his session. I did my best to pay close attention, hanging on every word, but remember asking myself why he would be so critical about SQL and not offer an alternative at the same time.

    I do appreciate the response to my posting. Glad to see someone of such high regards in the software community is reading my blog. Way to keep my honest 😉

    Reply
    1. Hugh Darwen

      Thanks for the kind reply. I should perhaps have commented on this too: “Circle is really an Elipse (Who didn’t know that?)” (That should be “Ellipse”, of course.)

      I think you might have missed the point. Of course “everybody knows” that every circle is an ellipse and he certainly wasn’t trying to teach that to anybody. The point is that the concept of subtyping that embraces such notions–specialisation by constraint–is not used, and cannot be used, in object-oriented computer languages such as C++, Java, and the so-called “object/relational” parts of SQL. That struck us as a bit of a shame.

      The work that Date and I did on this subject showed that specialisation by constraint (S-by-C) can indeed be supported in a truly relational language, thanks to the fact that such a language has no notion of pointers (including oids acting as pointers). Our proposed Inheritance Model (IM) consists of 26 prescriptions for a computer language to follow in order to embrace S-by-C. Our book devotes 110 pages to the topic, including concrete syntax that we added to Tutorial D to support that model. It’s not at all the trivial matter that your comment suggests it is!

      Reply
  3. babyshark13 Post author

    Consider this…every opportunity you and/or Mr. Date have to present your ideas is a teachable moment. I somehow walked away from the presentation unconvinced about his objections to SQL. In essence the opportunity to convince another software practitioner like myself went to the wayside.

    I’ve been thinking as to why the message wasn’t particularly clear to me. The most constant language in my programming vocabulary outside of C has been SQL for the past 12 years. Mr. Date’s comments were so extreme that I subconsciously tuned out for the rest of his lecture. He lost me within a few minutes of his SQL comment. I waited for a few minutes to see where he was going in his conversation thread. He failed to counter-argue why Tutorial D was a suitable replacement to SQL. That’s why I wrote what I wrote.

    Your interest in my blog, specifically in this conversation thread has encouraged me to review your combined work. I do encourage you to continue reading my blog and commenting when you feel appropriate.

    I do have one question for you…how exactly did you stumble upon my blog?

    Reply
  4. Hugh Darwen

    Yes, we’re still learning about how to air our criticisms of SQL without offending those who’ve grown to like it. My 2nd-year students at Warwick University are generally appreciative but some students do say in their feedback comments that they find “constant SQL bashing” a bit tiresome. As a result, I try to teach the theory without mentioning SQL too often, but I have to mention it when I teach a point of the theory that SQL deviates from, simply because many of the students are already familiar with SQL to some extent.

    Tutorial D itself isn’t a suitable replacement for SQL–the language is for teaching purposes only and is missing several features needed for an industrial strength DBMS. But a suitable industrial strength language could certainly be derived from it.

    I was pointed at your blog by one of the people involved in SOUG.

    Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s