Category: Oracle

So OCS 10g has been released and it’s “a member of the Oracle Fusion Middleware family of products”.

Let’s hope that during the upgrade to DB10g and AS10g they actually thought about the stability of the product, not just packing in more functionality that most people won’t use.

The next hurdle for me is getting the people here to buy in to an upgrade. My feeling is that the current verison is so bad, what have we got to lose! I don’t think they’ll see it that way 🙂

Cheers

Tim…

I don’t know if you know this, but I hate Oracle Collaboration Suite (OCS). It is the most useless pile of camel dung I’ve ever cast my eyes upon. It’s really is not worth the price of the CDs it’s delivered on.

Yesterday it crashed in the afternoon so I restarted it all (DB, infrastructure and 3 middle tier servers). By the time I got home it had gone again, so guess what I had to do… Restart the lot again. About an hour later it had gone again, but I wasn’t at home so a collegue had to do the honours. My last text message on the subject was a little after 22:00. This morning I get to work and (can you guess what’s coming next?) it has all crashed again. At least it’s consistently rubbish!

My only hope is that my company give up on it, rather than wait for Oracle to fix it, because I know Oracle will focus on adding more gimics to it, rather than getting it to work properly.

My advice to anyone contemplating using this product is DON’T!!!!

If you are forced to use it, like me, consider hiring people to cover a 24 hour reboot service!

Cheers

Tim…

I got an email yesterday saying that an article of mine has been accepted for the IOUG SELECT magazine. The article is a small section from the book I’m currently writing, which should be finished by the end of this month. I’m not sure when the article is going to be published so watch this space…

I’m looking forward to finishing the book and getting my teeth stuck into Oracle 10g Release 2. I’ve been so busy recently that I’ve not had much time to check it out, and as a result I’ve not been adding many new articles to ORACLE-BASE.com. A situation that will be rectified soon 🙂

Cheers

Tim…

Dodgy SQL
Developer: What’s wrong with this SQL statement?
Me: I’ve done an SQL trace on it and it churns through 1.6 Gig of data, 0.3 Gig from disk taking 439 seconds, the rest coming from memory, but incuring a 140 second delay due to assorted cluster waits.
Conclusion: It’s a quality statement, not! 🙂

Dodgy Products
My hatred for Oracle Collaboration Suite (OCS) is growing by the day. It’s completely…

I’ve decided to take leaf out of David Aldridge’s book and create a list of my own:

Alternative meanings for the OCS acronym:

• Overly Complicated Sh*t
• Obtuse Craptastic System
• It doesn’t work, it’s complete crap, just ditch it and write something decent that works and doesn’t need loads of harware to do a really simple job!

OK, the last one doesn’t quite work, but the sentiment is spot on! Please feel free to contribute your own ideas 🙂

By the way, did I mention that OCS is a steaming pile of…

Cheers

Tim…

PS. Oracle applications are a total mess as well.

PPS. The application servers are a nighmare. Hopefully Larry will get his head out of his ass and do something about them!

PPPS. Why are there so many bugs in the DB these days?

Yesterday I sounded like Barry White, today I can’t speak, much to the delight of my co-workers. I feel fine, but when I talk nothing comes out. Perhaps I should write all this text with a white font on a white background to simulate my current situation.

This reminds me of a summer holiday during University, when I worked on a production line in a factory for 3 months. All but one of the guys I worked with were deaf, and the one guy that could hear was at the other end of the line so I never got to speak to him. Fortunately the guys were really good at lip reading, so for 3 months I talked at work mostly without making any sound. I remember having a “chat” in the canteen one lunchtime and I started to speak out loud and got some funny looks from some of the guys off another line. It was only then I realized they had assumed I was deaf to as they had never heard me speak out loud 🙂

I had to teach a class last night using the “do as I do” method. Thought I would save my voice. Fat lot of good that did me!

I’ve just noticed that a Developers Release of Oracle Database 10g Release 2 (10.2.0.1.0) for Microsoft Windows is available for download. Not sure what’s missing from this, but it’s not a fully supported production release like the Linux version.

Cheers

Tim…

I’ve said it before and I’ll say it again, Oracle products don’t run properly on Tru64 anymore!

The latest issue came from an observation by a developer and involves problems with using NVL inside an EXISTS subquery. The following code is the test case I sent in my TAR to Oracle support.

CREATE TABLE tab1 (
code         VARCHAR2(10),
description  VARCHAR2(50)
);

ALTER TABLE tab1 ADD (
CONSTRAINT tab1_pk PRIMARY KEY (code)
);

INSERT INTO tab1 (code, description) VALUES ('1', 'ONE');
INSERT INTO tab1 (code, description) VALUES ('2', 'TWO');
INSERT INTO tab1 (code, description) VALUES ('3', 'THREE');

CREATE TABLE tab2 (
code         VARCHAR2(10),
tab1_code    VARCHAR2(10),
description  VARCHAR2(50)
);

ALTER TABLE tab2 ADD (
CONSTRAINT tab2_pk PRIMARY KEY (code)
);

ALTER TABLE tab2 ADD (
CONSTRAINT tab2_tab1_fk FOREIGN KEY (tab1_code)
REFERENCES tab1(code)
);

CREATE INDEX tab2_tab1_fk_i ON tab2(tab1_code);

INSERT INTO tab2 (code, tab1_code, description) VALUES ('1', '1', 'ONE');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('2', '1', 'ONE');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('3', '1', 'ONE');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('4', '2', 'TWO');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('5', '2', 'TWO');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('6', '2', 'TWO');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('7', '3', 'THREE');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('8', '3', 'THREE');
INSERT INTO tab2 (code, tab1_code, description) VALUES ('9', '3', 'THREE');

COMMIT;

VARIABLE p1 VARCHAR2(1)
BEGIN
:p1 := '%';
END;
/

SELECT t1.code
FROM   tab1 t1
WHERE  EXISTS(SELECT rowid
FROM   tab2 t2
WHERE  t2.tab1_code = t1.code
AND    t2.code like :p1)
ORDER BY t1.code;

CODE
----------
1
2
3

3 rows selected.

SELECT t1.code
FROM   tab1 t1
WHERE  EXISTS(SELECT rowid
FROM   tab2 t2
WHERE  t2.tab1_code = t1.code
AND    t2.code like NVL(:p1,'%'))
ORDER BY t1.code;

CODE
----------
1
1
1
2
2
2
3
3
3

9 rows selected.

SELECT t1.code
FROM   tab1 t1
WHERE  EXISTS(SELECT rowid
FROM   tab2 t2
WHERE  t2.tab1_code = t1.code
AND    t2.code like COALESCE(:p1,'%'))
ORDER BY t1.code;

CODE
----------
1
2
3

3 rows selected.

From this you can see that then the bind variable is set to ‘%’ the query without the NVL works as expected, but the query using the NVL displays a row in the master table for each row found in the subquery. In comparison, the query using the COALESCE function works fine.

Now I realise that this is not a great query anyway, but you would still expect it to work!

Happy days in Tru64 land again 🙂

Once again I’m drawn to the inevitable conclusion that running Oracle products on anything but a tier-one platform is a bad idea. What’s more, judging by the recent arrival of Oracle 10g Release 2, there is now only a single tier-one platform. I don’t see a version for Solaris, HP-UX or Windows yet.

As for me, I’m still waiting for 10.1.0.4.0 to be released on Tru64. Release 2 is a long way off 🙂

Cheers

Tim…