Re: Signal to Noise Ration on DBDebunk Declining

>> Has anyone else noticed the signal to noise ration on dbdebunk.com
declining over time? I used to think those guys were the coolest, but I'm disenchanted lately. <<

I tend to agree with you on that. But at some point, you have to start repeating yourself or you have slow days and dry spells.

>> They don't have as much substance as they seem to. Ex.: they're
always going on and on about being formal and precise, but they don't in fact use any formal methods whatsoever. The closest they get is the BNF in TTM, but apparently that's never even been run through a parser generator, as it has bugs in it. <<

Neither date nor Pascal have spent *any* time on a standards committee where someone would run your BNF thru a checker. Steve Feltz (sp) on the original X3H2 committee used to make sure that we were an LALR(1) grammar and Bruce Horowitz made sure the logic was right (Bell labs and studied logic under Raymond Smullyan).

I think that Fabian is trying to make a living from selling the "good stuff" on the website. Chris has a nice income from his textbook sales and does not need to work. Chris also gets invited back to places he has spoken at before :)

>> Maybe my next book should be SQL for Smarties, 3rd ed. :-) <<

Yes. Definitely yes. Two copies, home and office. Maybe a third for the glove compartment of your automobile. Does your mother have a copy? Christmas is just around the corner.

I am also doing a book on SQL Programming Style alng the lines of the old Henry Ledgard PROGRAMMING PROVERBS since SQL seems as strange to newbies as structured programming did to us in the 1970's.

>> At least I never hear Mr. Celko slamming Mr. Pascal, although he
certainly has just cause. <<

I have been attacked by Fabian for decades now. Big, fat, hairy deal.  A few things bothered me for a day or so. The rest of it is just Fabian being Fabian.

1. I was misquoted about 1NF; I made a remark that people try to fake repeated groups and arrays in SQL with a list of common programming tricks. The part of about these tricks being fakes was left out.
2. Currently, someone wrote in about an obvious typo in one of my old stock "cut & paste" files on the nested sets model. It has been corrected for years, but nothing dies on the internet :) The reply that I gave to the writer back then was "Shit, you're right." and that is still a good answer.
3. Someone sent them one of SQL puzzles from over a decade ago with an Oracle solution. Date did the problem in his personal DB language, but he changed the specs! The original problem was to show the current and previous payrate with the current and previous effective dates in one row per employee.

If the employee was a new hire, the previous data was to be shown as NULLs. Since Date does not have NULLs in his model, he used a dummy date of '1900-01-01' or something (I don't have time before a dinner date to look it up).

I sent Fabian a new version of the solution, using SQL-92 and CASE expression to replace the old SQL-86 solution an dpointed out the change in the specs. My email was never acknowledged, much less published.

In the early days when Date and I were in DBMS and DATABASE PROGRAMMING & DESIGN, I would deliberately take the opposite side of an issue that appeared in his column. Both magazines were owned by the same company (now CMP), so conflict boosted sales. I cooked his version of relational division (it did not meet his own criteria of division by an empty set) and computing a Median in SQL. He cooked an example of various 3NF schemas of mine. Great fun!

Date usually comes back with some content and a clear model; Pascal rants about the conspiracy against him and comes back with ad hominem. Received on Mon Nov 08 2004 - 17:47:46 CST