The Joel on Software Discussion Group (CLOSED)

A place to discuss Joel on Software. Now closed.

This community works best when people use their real names. Please register for a free account.

Other Groups:
Joel on Software
Business of Software
Design of Software (CLOSED)
.NET Questions (CLOSED)
TechInterview.org
CityDesk
FogBugz
Fog Creek Copilot


The Old Forum


Your hosts:
Albert D. Kallal
Li-Fan Chen
Stephen Jones

'The Third Manifesto' and Fabian Pascal/Chris Date.

As near as I can figure, these guys are complaining that the impedance mismatch between databases and OO languages is due mostly to SQL and it's poor implementation of "Relational Algebra." They claim that if we had a new database query language that tapped the full power of relational algebra, then everything would be fine.

Date & Pascal are classic "Architecture Astronauts." They bitch and moan about how databases aren't utilizing their precious relational algebra, yet they haven't stepped up to the plate and actually implemented usable software that does. I guess they expect to just shout down wisdom from their ivory tower and then we, the grunts, go and do the rest.

What really burns me about these two is that they aren't software developers, yet they have the gall to cast aspersions on the entire field just because we don't use their bullshit math properly. When these two assholes actually put their money where their mouth is, I'll pay them some notice.
Marc Fleury's short, short temper
Monday, November 22, 2004
 
 
You are stupid! You don't know what you are talking about! You must be an American. Only an American could be as stupid as you are.

If you all just did what I said, everything would be fine. You are solving problems with object-oriented paradigms that have been solved years ago by relational algebra. Everything has been solved by relational algebra! I can't name those problems off the top of my head... but trust me, they were solved.

You are total failures in life unless you master relational algebra.
Fabian Pascal
Monday, November 22, 2004
 
 
How dare you make fun of me with that name, "Marc Fleury's short, short temper." I do not have a short temper you douche bag!

You pricks are just jealous of me and my success. JBoss is by far the number one J2EE app server out there. Yes, it's the slowest and hardest to set up, but it has logging and AO (not AOP, we do AO). That's right you envious pussies. Every time you fire up JBoss, it vomits out thousands of lines of logging messages, logging everything in sight!!!

Geronimo is a knockoff, JRun is a toy, CDBullshit is crap... Is there even a comparison? No, there is not. JBoss rules hands down.
Marc Fleury
Monday, November 22, 2004
 
 
Peronsally I'm convinced that the 'Third Manifesto' represents the future of software development.  The principles it outlines represent a foundation upon which the next generation of object-relational languages will be built upon.

The most incredible thing is that they have made it a gift to whoever willing to do the grunt work.

Not so long ago Codd made such a gift to those willing to take it.  Some chap did the grunt work and founded Oracle.  He really should have realised that people were creating perfectly acceptable databases from scratch using stacks, linked lists and hash tables.

Then there were the chaps at Xerox's Parc who wasted there time trying to apply the theoriaes of biology to programming.  Good job Steve Jobs and Bill Gates ignored what they were doing.
Ged Byrne Send private email
Monday, November 22, 2004
 
 
At the risk of spoiling what is a stunning flame war in progress, there are reasons to at least aim for improving SQL. It's entrenched, yes, but that doesn't mean you can't consider improvements, even if they will never get widely used. For an interesting slant on this, going back to some of the early relational algebra with a new implementation, and fuller set with extensions, check out Tutorial D, which from memory is at sourceforge somewhere, though I could be wrong.
Andrew Cherry Send private email
Monday, November 22, 2004
 
 
Woops sorry Ged, didn't mean to imply that you were part of the flame war - was typing while you posted!
Andrew Cherry Send private email
Monday, November 22, 2004
 
 
Andrew,

No problem.

The implementation on SourceForge is Dave Voorhis's REL.

http://dbappbuilder.sourceforge.net/Rel.html

It looks very promising.
Ged Byrne Send private email
Monday, November 22, 2004
 
 
Ah that's right - I remember him being a UK based chap! It does look very interesting, it's on my list of things to investigate much more deeply as soon as I find a way of doing without sleep, and therefore free up another seven hours a day :)
Andrew Cherry Send private email
Monday, November 22, 2004
 
 
If you separate the character of Fabian from what he and Chris Date are saying then its far more acceptable.  Fabian has been screaming about the Emperor's new clothes for so long and how crap everything is that a lot of mainstream people tune him out.
Simon Lucy Send private email
Monday, November 22, 2004
 
 
I agree with Simon.  I think Fabian Pascal has a good point: that virtually all SQL implementations stray from proven relational database theory in ways that they shouldn't.  But he shoots himself in the foot because he is a hateful, pompous, bellowing ass who has no idea how to communicate with people who disagree with him, and who seeks out and focuses on disagreement even where there is none.  This ruins his credibility.  I'm not sure why Chris Date maintains such an apparently close association with F. Pascal; Date is much more even-headed and has top-notch academic credentials.
Herbert Sitz Send private email
Monday, November 22, 2004
 
 
The other big problem for Fabian Pascal, in my opinion, is that he virtually always ignores comments by developers who say that the SQL databases we have (Oracle, SQL Server, etc., etc.) just plain work, and what's more, for our purposes they work pretty damn well. 

Pascal would do better if he focused on these comments and showed exactly how things would work better with a database made like he wants.  Instead, he spews invective at developers who are satisfied with status quo and spouts endlessly about how existing db's ignore relational database theory.  He'd do much better if he was a nicer guy and just calmly explained to developers exactly how things could be improved by sticking closer to relational db theory.  But he won't, because he is at bottom just a nasty, nasty little man, mad at the world because nobody recognizes his self-proclaimed genius.
Herbert Sitz Send private email
Monday, November 22, 2004
 
 
Yes, that's generally the reaction Fabian generates.

Occsionally its entertaining, and every couple of years or so I'll stumble across some group or other that he's in the process of falling out with and its all usually much the same kind of stuff.

I wouldn't mind all the carping and retching if someone would just come up with a commercial product but I've been hearing most of these same arguments since around 1985.

It is not easy.  SQL works because it approximates.  Tutorial D might solve some of the array engine problems but its a god awful language.
Simon Lucy Send private email
Monday, November 22, 2004
 
 
“The other big problem for Fabian Pascal, in my opinion, is that he virtually always ignores comments by developers who say that the SQL databases we have (Oracle, SQL Server, etc., etc.) just plain work, and what's more, for our purposes they work pretty damn well.”

Herbert, being a disciple of Date or a pupil of Pascal, I would say that neither Date nor Pascal suggest that you should throw away SQL DBMS products – at least not until some sort of truly relational DBMS comes about. SQL products are significantly better than the hierarchical DBMS they replaced (IBM IMS) and worlds better than rolling your own filesystem, XML, etc.

But, compared to a TRDBMS, the SQL products fall flat on their faces in ways that actually impact application development: Pascal *does* indicate why it is important that SQL products fail relationally – his entire book “Practical Issues in Database Management” (including the series of Foundation Papers at DBDebunk.com although for they do require a fee to view) indicate not only the failings but how these failings cause problems to you, the developer.


“Instead, he spews invective at developers who are satisfied with status quo and spouts endlessly about how existing db's ignore relational database theory.”

Have you ever talked with Pascal? One-on-one about this topic? 99% of the anti-Pascal crowd have not, I guarantee. I have, and the impression I get from him is not of anger at end-users but of disappointment; the common misconception with Pascal is that he is ‘angry’ at users for their ignorance. Not so. Having talked with him on several occasions he has made it quite clear to me that he does not make their ignorance a personal failing – it's not their fault. They don't go through proper education and they are not required to have it. So they are neither aware that they are ignorant, nor do they have any incentive to do anything about it.

My feeling, though, is that he is less understanding with practitioners, when confronted with their ignorance, retreat to name-calling and basically dodge the issue of their ignorance. As he said (in something recent, I can’t recall) “the only thing worse than ignorance is arrogant ignorance”.

Of course, I agree that you’ll get more flies with honey than vinegar, but when you see how much trash is truly out there (view any Slashdot story’s posts, for example) and are forced to deal with it on a daily basis, you quickly lose the will to ‘be nice’.
Dr. Mario Send private email
Monday, November 22, 2004
 
 
Dr. Mario -- Yes, I have Pascal's "Practical Issues" book and it's a good piece of work, displaying little of the short temper or animosity that is characteristic of his many articles and writings on the web.  And I don't doubt that he might be fine to talk with one-on-one.

But the sad fact is that his personality on the web and on his "Database Debunking" site is probably all the average developer will ever know.  And his demeanor in those places is disgraceful.  What's more, it works against his mission of educating those in the tech world about the failings of current SQL databases.  People either tune him out or are repelled by him.  The repulsion often happens to people who have gone to him to express some sort of agreement, when he chases them away by attacking what they've said to show how stupid they are and that what they've actually said is contrary to what he's saying.

Every once in a while I go back and take a look at "Database Debunkings".  And every time I'm quickly repelled by the nasty tone of virtually everything there.  His favorite method of education is to start by  trying to embarrass and humiliate people who have attempted to express their agreement with him.  He then attacks what they've said and distances the very people whom he could recruit for his cause.  And in many cases he offers up quotes of what these people have said as evidence for how stupid they are, and he doesn't even explain how what they said is wrong, just leaves it as (apparently) self-apparent for the reader to see. 

It's harsh, but my opinion is that the man has deep personal problems that greatly hinder his effectiveness at spreading his gospel, good though his gospel may be.
Herbert Sitz Send private email
Monday, November 22, 2004
 
 
In the past I've seen conversations begin reasonably enough and then for some reason or other he goes into attack dog mode.

The fundamental reason that there isn't a truly relational engine is that commercially there is no market for it.  None that could afford it, not even the military.

And even so, the fault is not with SQL syntax necessarily, execution order doesn't have to be contingent with statement order and query paths can be reordered by the underlying engine.

But imagine re-engineering all the major database client applications out there to move from a statment oriented DBMS language to a partially procedural, partially parallel array processing language?

So, I'd be considerably more convinced if someone that was that exercised about it developed a relational engine that hid that complexity and built on the basis of what we already have.
Simon Lucy Send private email
Monday, November 22, 2004
 
 
It's rare for F. Pascal to say anything positive about any commercial database products.  He has said, though, that Dataphor (http://www.dataphor.com ) does a good job of implementing some of the proposals in _Third Manifesto_:

http://www.wilshireconferences.com/interviews/pascal.htm

I'm curious whether anyone on JoS has any experience with Dataphor.
Herbert Sitz Send private email
Monday, November 22, 2004
 
 
SQL databases are great. I hope my bank keeps on using them. I don't know about a real RDBMS. I think they would be more usefeul for providing application persistence services. For now I manage pretty well without SQL databases.

The trick is to avoid "Enterprise" development. You know, Black Hole and all.
fool for python
Monday, November 22, 2004
 
 
This quote of Pascal's, from Jan2004, has me puzzled.

"A true RDBMS (TRDBMS) e.g. one based on the TransRelational™ implementation model, will have a considerably wider range of optimization options than SQL products, and will perform several folds better with fully normalized databases."
http://www.tdan.com/sms_issue27.htm

Not the Third Manifesto, but the TransRrelational implementation model.  He links to the site http://www.requiredtech.com/, which appears to have been put together by a schoolboy back in 1997.

There's a seminar on Database debunk: http://www.dbdebunk.com/page/page/619867.htm

This TransRelational thing appears to he The Third Manifesto: Trademarked.

Whatever it is, it hasn't even got a discussion on C2 yet.  Pure vapour. 

This type of thing really doesn't help there case.
Ged Byrne Send private email
Tuesday, November 23, 2004
 
 
In the end the argument often boils down to:

It will be better because it will be better.
Simon Lucy Send private email
Wednesday, November 24, 2004
 
 
Simon,

You are quite misinformed on what a RDBMS is and what it is not.

“The fundamental reason that there isn't a truly relational engine is that commercially there is no market for it.  None that could afford it, not even the military.”

No, the fundamental reason is that most people don’t know what a RDBMS is, so what motivation is there for anyone to actually make one?

“But imagine re-engineering all the major database client applications out there to move from a statment oriented DBMS language to a partially procedural, partially parallel array processing language?”

This makes no sense. Codd’s original language was wholly declarative (is that what you mean by ‘statement oriented’?) but the language used to query/talk to a Relational DBMS, as long as it adheres to Relational principles, can 'look' like anything the implementer desires. No where are Date and Pascal advocating some sort of “partially procedural, partially parallel array processing” language. They advocate declarative languages which operate on relations (arrays make no sense in a Relational context).

“Pure vapour.”

The current implementation is the Alphora Dataphor product, which someone mentioned earlier. However, Date and Pascal are simply laying the roadmap for a TRDBMS; how does the lack of them actually implementing it debunk the Relational Model?

They are not an army of skilled software engineers who work for a multi-billion dollar company: they are two mathematicians/logicians who understand data management fundamentals and what a RDBMS should be.

“So, I'd be considerably more convinced if someone that was that exercised about it developed a relational engine that hid that complexity and built on the basis of what we already have.”

The Alphora Dataphor product actually sits on top of a various number of existing DBMS products in order to utilize their storage techniques, etc. Basically, your program talks to Dataphor which then transforms that into requisite SQL statements (unless you use the built-in storage engine).

“In the end the argument often boils down to: It will be better because it will be better.”

Pascal and Date have written entire books on how it will be better. Only if you choose to ignore those materials can you make a statement such as the above.
Dr. Mario Send private email
Wednesday, November 24, 2004
 
 
Dr Mario,

Rel and Alphora Dataphor both implement T H E  T H I R D  M A N I F E S T O.

The third manifesto is a solid, well documented description of how things could be done.

Date and Darwen go to great lengths to stress that TTM describes how things should work on the C O N C E P T U A L level, because 'Logical Differences and Big Differences'

Now Pascal is talking about the Transreational I M P L E M E N T A T I O N model.  He is talking about it in Hyperbole as the greatest thing to every happen to relational technology.

"So what is the TransRelational model?  In essence, it is an implementation technology; it represents among other things a radically new and elegant approach to the question of implementing the relational model.  When the relational model first appeared, skeptics accused it of being impossible to implement efficiently.  Over the next few decades, therefore, vast sums of money were spent on attempts to prove the skeptics wrong.  And those efforts did not prove in vain; today's products do perform fairly well, if not always spectacularly.  But at what cost?  The products have become quite notoriously unwieldy and difficult to manage, with their huge array of storage structures, access methods, optimization techniques, tuning knobs, analysis utilities, performance hints, installation options, and so on, and the job of the database administrator has become virtually impossible.

 

The TransRelational model has the potential to change all that.  It provides a totally new approach to implementation, one that is dramatically different from those that have been tried in the past and found wanting (including all of the approaches encountered in today's SQL products).    Such an implementation would be orders of magnitude faster than, and would deliver a far greater degree of data independence than, today's SQL products.  It would also greatly simplify the job of the DBA.  In a nutshell, the TransRelational model allows us to build DBMSs that--at last!--truly deliver on the full promise of the relational model."

So where is it?  He links to a poorly put together homepage with no information.  Look at it, it's embarrassing:  http://www.requiredtech.com/

The poor format would be excusable if the site contained every a scrap of a whiff of information about the TRM.

CJ Date is giving seminars with rather puzzling topics:
http://www.dbdebunk.com/page/page/619867.htm

The only other thing that I can find is this rather intriguing topic in a discussion forum: http://www.webservertalk.com/message486606.html  I wonder who this James Hewitt chap is.

I am a stauch supporter of The Third Manfiesto, and I'm not finding this poorly put together hype very constructive.  Then again, maybe I'm totally wrong.

If there is a good page explaining what the Transrelation Implementation Model then tell me, I shall enjoy eating my hat if it means witnessing such an incredible leap forward in Database technology.  Humble pie will never have tasted so good.

I made my position clear in my previous posts.  You should have  read them before making a statement such as the above.
Ged Byrne Send private email
Thursday, November 25, 2004
 
 
Hmmm.  More digging shows that all is explained in Appendix A for Introduction to Database Systems 8th Edition.

I only brought the 7th Edition last year, but it looks like I'm going to have to buy the whole book if I want to find out what this whole TRM think is all about.
Ged Byrne Send private email
Thursday, November 25, 2004
 
 

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics
 
Powered by FogBugz