Newbie: Concurrent query and update

Don't bother wasting your time with Derek. As you are
finding out, when pressed on one of his faulty technical
posts he will attack the poster or ramble on about anything
other than the technical issue at hand.

Someone invited this bloke to our offices last year. Even
the junior members of our team could tell that he didn't
know what he was talking about. When it came time to take a
break for lunch management told him not to bother coming
back after lunch.

Some people are ignorant of their ignorance.

> > Generally people who have standards
>
> your standard looks like post bad information and when
> someone points to your mistakes you spend lots time tapp
> dancing and saying nothing. you just posted 3 posts and
> said nothing useful

> On 2008-08-27 01:31:34 +1000, jtotally_bogus-at-sbcglobal.net (J) said:

> I think there are cases where one needs multiple pieces of information
> from more than one select statement to draw conclusions. By holding
> exclusive locks on the select statements within a transaction one can
> assure that the information is consistent at that point time and act
> accordingly.

Ah yes, the Phantom Read problem. I agree that the prescribed method
is to open a tran and use HOLDLOCK. However, that acquires Shared
locks, not Exclusive locks.
--
Cheers
Derek
Senior Sybase DBA / Information Architect
Copyright © 2008 Software Gems Pty Ltd
Quality Standards = Zero Maintenance + Zero Surprises
Performance Standards = Predictability + Scaleability

> On 2008-08-27 05:01:15 +1000, ozdba said:

Do you care to identify your "self" or the "customer" ?

Must feel real safe and strong being an unidentified flying object who
snipes from the dark shadows.

"ozdba". More likely junior developer with dbo access.
--
Cheers
Derek
Senior Sybase DBA / Information Architect
Copyright © 2008 Software Gems Pty Ltd
Quality Standards = Zero Maintenance + Zero Surprises
Performance Standards = Predictability + Scaleability

Two days and this abuse remains on the Sybase website. the evidence is
the "moderators" do not exist. Or else they are Mark or Jason or
Kevin, just accidentally failing to notice and delete abusive posts.
When you collude with the abusers, you prove yourself to be one.
--
Cheers
Derek
Senior Sybase DBA / Information Architect
Copyright © 2008 Software Gems Pty Ltd
Quality Standards = Zero Maintenance + Zero Surprises
Performance Standards = Predictability + Scaleability

On 27 Aug 2008 12:16:58 -0700,
in sybase.public.ase.general
Derek Asirvadem wrote:
>Two days and this abuse remains on the Sybase website. the evidence is
>the "moderators" do not exist. Or else they are Mark or Jason or
>Kevin, just accidentally failing to notice and delete abusive posts.
>When you collude with the abusers, you prove yourself to be one.
>--
>Cheers
>Derek

I don't "can" posts any more unless they are obviously spam - I haven't done so
for months. I have sent it up the pipe to have the posts canned.
--
Jason L. Froebe
TeamSybase
http://www.froebe.net/blog
MyDatabases Free Magazine http://froebe.net/blog/mydatabases-magazine/

> On 2008-08-28 05:58:14 +1000, "Jason L. Froebe [TeamSybase]"
> said:
>
> I don't "can" posts any more unless they are obviously spam

Why not ? Nothing personal, but what is the use of Sybase having
moderators that do not (or cannot ot will not) moderate ? It is silly
to have able bodied humans acting as spam filters.

> - I haven't done so for months. I have sent it up the pipe to have the
> posts canned.

Ok, thanks for letting me know.
Therefore it is the pipe that is not working.
Reporting a Post is not working as well.

Actually the problem is lack of registering, which allows people to
impersonate others, and also allows known people to pretend they are
someone else. It may not be a concidence that Mark Parsons has his
annual argument with me about the exact same subject, every year, and
immediately following we have an American pretending to be an Aussie
posting abuse.

Since moderation is de facto non existent, it is high time we put
registration in place,
--
Cheers
Derek
Senior Sybase DBA / Information Architect
Copyright © 2008 Software Gems Pty Ltd
Quality Standards = Zero Maintenance + Zero Surprises
Performance Standards = Predictability + Scaleability

Derek Asirvadem wrote:
> Actually the problem is lack of registering, which allows people to
> impersonate others, and also allows known people to pretend they are
> someone else. It may not be a concidence that Mark Parsons has his
> annual argument with me about the exact same subject, every year, and
> immediately following we have an American pretending to be an Aussie
> posting abuse.

Actually, I point out your technical mistakes whenever you make them and you just happen to make the same mistake every
year. You should consider validating your recommendations for technical correctness if you have a problem with being
corrected.

The fact that others notice your technical shortcomings would tend to imply that said shortcomings are easily
noticed/picked up by others.

As for Aussies posting about your technical shortcomings ... I would expect your neighbors to be more in the know than
non-Aussies.

> Since moderation is de facto non existent, it is high time we put
> registration in place,

I second that motion, then we can ban individuals just like ISUG does.

I don't can postings. I could care less if certain individuals wish to make an arse of themselves. If those
individuals later want their postings canned then they should consider not submitting cannable posts in the first place.

Derek Asirvadem wrote:
> Two days and this abuse remains on the Sybase website. the evidence is
> the "moderators" do not exist. Or else they are Mark or Jason or Kevin,
> just accidentally failing to notice and delete abusive posts. When you
> collude with the abusers, you prove yourself to be one.

> On 2008-08-26 09:43:54 +1000, Derek Asirvadem
> said:
>
> It is not I who is asking for Optimistic Locking. Good transaction
> standards (ACID roperties and their SQL/OLTP/Client-Server
> implementation), good design standards (avoid holding locks for longer
> than absolutely required), demand Optimistic Locking.

It appears some coders have not heard of Optimistic Locking, or more
important, they do not understand the problems that Optimistic Locking
overcomes. Let me vent my anger at the abusive ignorami by providing
honest seekers with a bit of useful information.

This has nothing to do with Sybase or Oracle or MS or databases; it has
everything to do with writing a system for concurrent access by more
than one user. There are many problems that need to be understood and
avoided; a few concepts to be understood and coded by design; too many,
that's why it takes formal education (3 days). I will deal with just
the one earmarked Solution here.

Scenario I
----------
Order Entry system. One design goal of the system is high concurrency.
No backorders allowed. No orders taken unless we have quantity in
stock. There is a book D1 "Concurrency for Dummies" in stock.
10:00 - Customer C1 phones order clerk O1 and interacts regarding Sales
Order S1 for D1 x 100. Gets told we only have 80 in stock; order
quantity changed to 80. Continues interaction re price of fish and the
quantity of oil in Uzbekistan.
10:05 - Customer C2 phones order clerk O2 and interacts regarding Sales
Order S2 for D1 x 50. No time for pleasantries. Order S2 confirmed
and booked.
10:15 - Order Clerk O1 tells Customer C1 that he has to get on with the
next cust, and closes the order.

Now there are three possible outcomes, depending on the education level
of the app coder. There is a fourth possibility if we have an
application DBA but nowadays coders code data "bases" as well, so we
will focus on the coder's perspective:
a. Computer Science degree and can read books: Upon "Save" Order S2 is
rejected by the stored proc/transaction with error "Order quantity 80
exceeds stock quantity 30"
b. No tertiary but can read: Order saved, QuantityInStock is
_negative_ 30 units. Order Clerk is blamed for coder's stupidity,
loses job, wife takes hammer to coder.
c. No tertiary, cannot read: Order saved, QuantityInStock is zero.
Warehouse staff blamed for coder's stupidity, they make personal
deliveries to coder in the parking lot, late at night.

The coder who can state the problem exactly, in one sentence, gets an
ice cream cone at recess. After that, we will worry about the solution.

Scenario II
----------
Credit Control system, could be part of Order Entry system in Scenario
I. No new Orders allowed above Customer's Credit Limit minus
sum(outstanding invoices).
11:00 - The law has caught up with Customer C1; he phones in to change
his address and interacts with Order Clerk O3. Address is 123 Sunset
Blvd, Hollywood, CA. Credit Limit is $5,000
11:05 - Credit Officer reads his reports, notices customer C1 is way
beyond his payment capability, is 90 days overdue; decides to downgrade
his Credit Limit. Updates Customer row, sets Credit Limit to $0.
11:10 - Order Clerk saves new Address, Cell 666 Cell Block D, Alcatraz, CA.

Two possible outcomes, depending on the :
a. Computer Science degree and can read books: Upon "Save" the update
is rejected by the stored proc/transaction with an error
b Has access to Internet and can type ...
13:15 - Customer C1 sets up new business with new partners. Phones in
to place an Order. Interacts with Order Clerk O4. Clerk tells him he
has $5,000 credit Limit minus $4,000 outstanding. Places order for
book D2 "Hot Air Balloons for Dummies" x 100 for $1,000.
16:55 - Credit Officer fires Order Clerk.
21:35 - Husband takes pitchfork to credit officer.
23:45 - Husband buys book from new friend and places order for balloon.

The coder who can state the problem exactly, in one sentence, gets an
ice cream cone with chocolate topping at recess. After that, we will
worry about the solution.

Scenario II is one form of the "Lost Update" problem. For those of you
who are thinking ahead, yes, there are two solutions to this form of
Lost Update. However, one of the solutions is Optimistic Locking: if
it is implemented it addresses both Scenario I and Scenario II
problems; and the second solution (which requires a separate onerous
coding standard) can be eliminated. IOW, Optimistic Locking is the
more elegant and less onerous solution; if it is there (to overcome
other problems such as Scenario I) the second code code construct is
not necessary.
--
Cheers
Derek
Senior Sybase DBA / Information Architect
Copyright © 2008 Software Gems Pty Ltd
Quality Standards = Zero Maintenance + Zero Surprises
Performance Standards = Predictability + Scaleability

> On 2008-08-28 07:56:57 +1000, "Mark A. Parsons"
> said:
>
> Actually, I point out your technical mistakes whenever you make them
> and you just happen to make the same mistake every year. You should
> consider validating your recommendations for technical correctness if
> you have a problem with being corrected.

I suppose it is beyond your imagination that
(a) there could be known problems
(b) there are known solutions
(c) that can be eliminated by the implemenation of standards
(d) it is you who is making the same mistake every year

I have created my current education (Sybase-specific) from paid
education (not Sybase- or RDBMS-specific) which I received from
respected education providers and mentors, and modulated it over long
and hard experience. I seek and take validation from my peers and
mentors. I receive substantial validation and acknowledgement from my
customers and their developers who implement my standards (or parts
thereof) and obtain tangible relief from the problems they have. It is
idiotic to ask for validation from people who by definition are
uneducated about the subject, so I have not attempted that.

Casting pearls before swine (that is a saying over here). The people
who have the problems and the budget pay for the solution. Waste of
time selling it to anyone else.

> The fact that others notice your technical shortcomings would tend to
> imply that said shortcomings are easily noticed/picked up by others.

It could also be that the level of knowledge of posters these days is
so low, that they do not understand standard terms and standard
solutions (as proved in this post). If you are willing to be civil,
then continue the thread (where I have posted a question, the
continuation) and I will respond, maybe you and certain posters could
learn something.

It could also be that certain known people with fixed ideas, are
pretending to be someone else, and driving their established position
home.

One man's "technical shortcoming" is another man's standard with proven
technical and commercial value. Again, I would not try to sell it to
people who are ignorant that they have a problem that was well known
forty years ago, or that there is a solution to the problem from that
vintage, rendered in a template for Sybase use.

> As for Aussies posting about your technical shortcomings ... I would
> expect your neighbors to be more in the know than non-Aussies.

Hilarious. First it is an American or Canadian doing a poor job of
pretending to be Aussie. Second it is an unidentified person stating
obvious lies (hence I cannot take it personally or respond). Third
they have been challenged to identify the alleged customer, and they
have not. (I have never given "talks" to customers, invitation or not;
I do fixed price work to fix their problems and to educate their coders
in high concurrency, high performance systems). As it stands it is
challenged and unsubstantiated allegation from an unidentified person
or a known person pretending to be someone else.

Just because someone posts, does not mean it is fact. LOL. They are
evidently as ignorant (no offence) of the problem, the solution, and
the standard, as you.

Your collusion/impersonation is demonstrated.

>> Since moderation is de facto non existent, it is high time we put
>> registration in place,
>
> I second that motion, then we can ban individuals just like ISUG does.

Cute. If ISUG operated the way they declare, they would have banned
certain posters long ago, and would not have had the years of problems
they did; which ended up with banning certain posters now. No use
crying about the responses, just deal with the causative abuse, the
first abuser, and you can avoid the response. Registration would
reduce these unidentified creatures, whose very ability to post without
identifying themselves allows them to post abuse

Don't forget, your posts and attacks on me here, have been summarily
deleted as well.
--
Cheers
Derek
Senior Sybase DBA / Information Architect
Copyright © 2008 Software Gems Pty Ltd
Quality Standards = Zero Maintenance + Zero Surprises
Performance Standards = Predictability + Scaleability