Re: "Structured" Entity-Relationship Model?

> On Monday, 22 April 2013 01:45:11 UTC+10, Roy Hann wrote:
>
> I suppose if the DBMS insists of testing the
> referential integrity contstraints before you assert that you have
> completed a transaction that has left the database consistent (i.e. you
> COMMIT) that would be a problem, but that's just a broken DBMS. Indeed
> most SQL DBMSs seem to want to test constraints after every blessed
> statement, which is silly.

It might be good if you get out and get some air.

All SQL-Compliant platforms perform statement-level checking, yes. Thousands of developers, have succeeded in producing millions of databases, which maintain database [C]onsistency, without that being an "impediment". That means they do not have Deferred Constraint Checking.

Of course, that excludes PostGreNonSQL, MyNonSQL, etc, because they are not SQL compliant. The people who use them write code in isolation, and have numerous weird and wondrous beliefs, firm ones, about the real world, but they never get out to experience it.

> SQL applies a series of
> tiny, local, sequential changes to the database. It is absurd to
> expect every such incremental change must leave the database
> consistent.

Ditto. Those millions of databases that never lose [C]onsistency or integrity, have that quality. It is possible that the one who has never coded in the real world, is the absurd one, and that the real world is not. Mental patients usually think they are sane, and that the entire world, is insane. They never get out. (Until recently.)

All you need is formal education on how to write those "tiny, local, sequential changes" in such a way that:

- conform to a natural order
- they never collide
- there is no repetition of code segments
- each and "every such incremental change" is compliant with the declarations
- leaving the database [C]onsistent

> It is absurd that one could not do "bulk loading" or "bulk
> deleting" or any other series of things, in any random order, deferring
> the question of consistency until you, the programmer, want to make that
> claim.

That is not how the real world works.

We make that claim when we define the database, in the declarative phase, before any data is loaded into it. At least those thousands of us who use SQL-Compliant platforms.

For the thousands who use freeware and non-compliant platforms, there is no end to the series of "features" that they need, in order to implement their "databases". Second last was the insane MVCC. Last was DCC. Next will be whatever the next mental limitation is exposed to be. There is no end to it. We do not suffer the impediments, and we do not need the "feature" to overcome the impediments. One really has to consider the kind of "databases" that these people with such strongly established impediments will "produce".

Again, all you need is formal education. Things like "any random order" have to be exorcised from your mind. Then you can, like the rest of us, make the claim first and load second. Tell us what you are selling first, then make the sale. The reverse is the hallmark of a fraud or nutcase. You would be sitting there, all alone in the market, after everyone has gone home, separating the chaff from the wheat, by hand.

> Erwin wrote:
>
> Is the statement-level constraint checking still "silly" if multiple assignment
> as per TTM is available ?

Statement-level constraint checking is standard. It existed in platform since 1984, before the standard arrived.

Multiple assignment is a joke. A bigger joke the DCC, but from the same asylum.

TTM is an orgy of jokes. After 16 years (or whatever) of zero delivery, the jokes are sad, and tired. All they have are definitions and demands, like Roy's, which are irrelevant to the database community. The only relevance they do have is celebral, in the smoke-filled chambers of those completely isolated from the world, full of their own self-importance. The entire TTM proposition is erected on a Straw Man. Academics love their Straw Men.

By Drawne's own declaration, when faced with a serious question, hey, it is a "toy language". Subsequent to my posts there, three years ago, he was going to come up with an "industrial strength toy language".

he has no formal education in Computer Science. Of course, that doesn't stop him inflicting his insanity on university students at Warwick.

Best to increase his meds.

Cheers
Derek Received on Tue Feb 25 2014 - 14:24:34 CET