> On Tuesday, 27 July 2021 at 21:37:57 UTC+10, Nicola wrote:

> I am back, too,

Good to have you back.

> Derek, I have read your Transaction Sanity document. When academics
> discuss "transactions", they do so in the context of what eventually
> becomes a start transaction… commit block in implementations. For
> instance, they ask themselves: under what conditions are lost updates
> (and other anomalies)

I don’t accept that. They don’t even know about Lost Updates. There is nothing in the literature about it.

Further, /By their fruits ye shall know them/. For more than twenty years, the CS grads we hire have no clue about these issues, meaning that they are taught nothing about real world Transactions. BIg corporations pay big money to get their developers educated precisely because they are not educated in the “education” system, they are ignorant of CS even though their degree is CS.

> be avoided when:
>
> -- Transaction 1
> start transaction;
> -- Perform some operations, including updates
> commit;
>
> -- Transaction 2
> start transaction;
> -- Perform some operations, including updates
> commit;
>
> are executed concurrently?

There is not enough detail in that to say anything, one way or the other. It is just two generic code segments. Give it more detail and ask a specific question.

Starting from scratch is insanity.

> Admittedly, the examples that are usually
> found in textbooks are quite misleading: the classical fund transfer is
> used because the context is readily understood, but it should be taken
> with a grain of salt, and certainly not as a way to suggest how a real
> banking transaction should be implemented.

I don’t have a problem with that particular example. I have a problem that the textbooks give no further examples, that they do not get into the issues (contention; reduction; concurrency; etc). The textbooks have a consistent single-user mindset, promoting schizophrenia.

> It's naive to think that the
> "naive solution" in your document is what a (good) teacher would suggest
> as a practical way to solve the problem stated in that document.

I did not say that a teacher taught them that, no idea where you got that. I can’t defend what I did not say.

I said (you can take implications of what I said), that naïve developers do that. Precisely because they are clueless re the consequences, precisely because they are NOT taught anything about OLTP (technology in the real world since 1965, on SQL Platforms since 1984). The sequence is:
- they code “transactions” badly, with no template
- they deliver the system with minimal testing
- everything looks fine
- as more and more users start using the system, Transactions disappear; money is given to people who are barred from receiving that money; deadlocks start to happen; etc
- a vague notion that the system had concurrency problems is formed
- the developers scramble to fix the worst few “transactions” and the most obvious; the easiest, thing to do is BEGIN TRAN in the client
- it immediately fixes the lost data; lost “transaction” problems, so they have confidence in that particular band-aid
- but it introduces massive lock contention

> To make
> the point clearer (and simplifying things a bit): the transaction theory
> that we teach, and which you can find in (good) textbooks, is the theory
> you would need to write a transaction manager for a DBMS (i.e., the
> theory needed by Sybase developers),

1. For Sybase Engineers (your “developers”)
You have said some silly things in the ten years that you and I have conversed, but that is definitely the silliest. As per the evidence (not a claim; not an opinion), the academics (
a) maintain a steadfast denial of the real world; of the industry; of the facilities in commercial SQL Platforms,
(b) theorise about a contrived notion of the real world; a contrived notion of the industry; a contrived notion of the facilities in commercial SQL Platforms, aka FANTASY
(c) as evidenced in hundreds of papers that produce sweet fanny adams, that have zero effect on the real world; on the industry; on commercial SQL Platforms
(d) and they are going to tell Engineers of commercial SQL Platforms, who wrote the THEORY, and who still today write progressions to the theory (most are PhDs)
(e) about the commercial SQL Platforms that were established FORTY YEARS ago, and have FORTY YEARS of maturity, that are secured with scores of PATENTS (as well as proprietary methods)
(f) FORTY YEARS after the fact.

I love it. You can’t make this stuff up. If it was not in writing, no one would believe it.

The is the same cultivated insanity that academics who allege to be in this field have, and have had since 1970. Inventing the wheel from scratch requires the denial that the wheel was invented FIFTY YEARS ago. Theorising about “semantic models” FORTY YEARS after IDEF1X, in hysterical ignorance that the /RM? is based on FOPC; Predicates; that any suggested model has to be Logic; Mathematics; Semantic.

This is the same hilarious inferiority complex (presenting as “superiority complex”) that academics have. This is the same idiocy that Date; Darwen; Fagin; etc have, trying to say that they can teach anything, ANY THING, about the /RM/, or progress the /RM/.

Remember, Codd was not an academic, he was a scientist who worked for a platform supplier. The academics never accepted him, they have not produced a single progression of the /RM/, but they have produced hundreds of papers promoting 1960’s Record Filing Systems fraudulently labelled as “relational”.

2. Example
Ok, take it another way, provide one single example of a thing that academics have produced that “Sybase Engineers” need.

3. For wannabe ArrDeeBeeEmEss developers
Even that is stupid. Why on earth would anyone theorise about something that has been implemented and proved for FIFTY YEARS ?

>>>>
Here is my take on science, as it has been 350BC to 1911. (Which is under attack since 1911, by Modernism.) It is a body of knowledge (certainty: the Latin word means knowledge not speculation). Of course, one can speculate BASED ON that body of knowledge, and that is HYPOTHESIS, which when confirmed by tests is elevated to THEORY. If and when the theory is proved (proper proof, proper method, not merely a mathematical definition of a theory), it progresses to a truth, and is added to that body of knowledge.

When that theory, which is proved in one or the other vertical gets proved in many verticals, it is elevated to LAW, such as the LAW of Thermodynamics, and it applies to all other sciences. Eg. evolutionism fails the second LAW of Thermodynamics.
<<<

OLTP is LAW. The proof in the mainframe rendition was fixed in 1965. The proof for the minicomputer rendition and SQL Platforms was fixed in 1984 (Britton-Lee in the late 1970’s). I was benchmarking CINCOM/TOTAL NDBMS against Britton-Lee proprietary DBMS in 1979. (Britton-Lee became Sybase in the SQL rendition.) Nothing has changed.

Anyone theorising about OLTP after 1984 is in schizophrenic denial of the real world. The hallmark; the proudly declared badge, of academics in this field. They openly declare that implementation concerns are divorced from theory. Thus their theory is fantasy, about a problem that has been solved FIFTY YEARS ago.

Oh wait, now that they are pushing PissGress, which is an “implementation”, suddenly they have started theorising about implementations, contradicting their own declaration. But it is still divorced from reality, a contrived fantasy. Instead of finding out what Commercial SQL Platforms have doen for FORTY YEARS, they deny its existence, and contrive to invent the wheel, from scratch.

What is demanded, what is NOT taught, is education about the real world, education about OLTP LAW. Instead, you guys teach a MV-non-CC or “MVCC” mindset, and no OLTP at all.

Separate point. If you want to theorise about something that relates to Transaction Processing, that might illuminate the subject, study chip technology, particularly the advances in [software, live objects] THREADS. It is not a coincidence that Sybase ASE uses a single unix process for the server, and multiple software THREADS internally (fully configurable) ... that, wait for it ... handle multiple user THREADS, some of which are competing Transactions.

It is with the same level of stupid, and the same level of schizophrenia, that academics teach ERD for “relational” modelling, or theorise about “semantic models”, FIFTY YEARS after IDEF1X, FORTY YEARS after it was established as a Standard.

Put another way, just look at the stupefying “implementation” known as PoopDePoopGress, that evidences the level of theorising that the academics actually theorise about, that they are aware of.

--------------------------------
WAKE THE FROG UP
---------------------------------------------------------------
IT IS A BACKWARD PIECE OF PIG POOP
------------------------------------------------------------------------------------------------------------------------------------------------------------
IT DOES NOT HAVE EVEN THE BASICS THAT COMMERCIAL RDBMS HAVE HAD, FOR FIFTY YEARS
------------------------------------------------------------------------------------------------------------------------------------------------------------

You can’t make this stuff up. About two hundred programs competing like the asylum dwellers that wrote it, pretending to be a server, hosting multiple stale offline versions or records, with a determination to deny reality of Online; of Transactions; of Processing. That is the evidenced level of your theorising. People who have not lost their brains identify that as contrived fantasy.

And when it doesn’t work according to their fantasy, they redefine the established terms. That crosses the line from insanity into criminal insanity. Pure evil.

> rather than the theory needed to
> correctly implement the transactions for a banking system or any other
> application, which builds on top of the former and is seldom, if ever,
> discussed in textbooks.

Ok, so you have confirmed what I said.

> Hence, I am eager to hear what you have to say
> about it.

Be my guest, ask a specific question. Either follow and progress this thread, or any of the other that you started but have not completed.

As I have stated, I teach science; truth; knowledge, that does not change. I have no opinions or that subject matter. If you are a scientist, please stop demeaning the subject to mere opinion; personal claim. It is science; knowledge; certainty; permanent (truth is permanent).

Even got SQL Server and Sybase installed to verify your claims

On the one hand, great, cool, knock yourself out.

I won’t suggest that you can use it properly, because you have that evidenced academic backward mindset. I have seen quite a few academics make hopeless databases in Sybase, because they are clueless about the /RM/; SQL; and database design, and then cry that Sybase is broken. This is why I have declared, both on c_d_t and at TTM:

----------------------------------------------------------------------------------------------------------------------
IF THERE IS ANYTHING THAT YOU THINK THE /RM/ OR SQL CANNOT DO,
GIVE IT TO ME, AND I WILL GIVE YOU THE SOLUTION
----------------------------------------------------------------------------------------------------------------------

After about five sickening embarrassments for Darwen, the freaks at TTM stopped asking me.

----

On the other hand, why ? Any declaration that I make is supported by science. If you think something is merely a claim, just ask for confirmation. All the product manuals are online. There is no need to install a platform and generate a database for the purpose of checking a claim or opinion.

But again, back to the one hand, it is great that you have finally started to obtain knowledge about the real world, about what a commercial SQL Platform is. You are breaking through one aspect of academic denial of reality, that is to be commended.

No suggestion that you can learn about the project (Relational database design; contention reduction; OLTP; etc) from product manuals, but in order to use the product properly, I do suggest you download the full set of manuals (PDFs). Re this thread, read the P&T Locking manual.

You may find these docs useful:

____ https://www.softwaregems.com.au/Documents/Sybase%20GEM%20Documents/

Cheers
Derek