Oracle SQL and PL/SQL Coding Standards – Cat Herding for Dummies

Whilst in Montreal recently, Deb and I made a pilgrimage to the Circuit Giles Villeneuve, home of the Canadian Grand Prix. When not in use, the track is open to the public. It’s divided into two lanes – one for people to walk and cycle down down, and a one for people to drive down.
You can just imagine flying round in an F1 car. You come out of the excruciatingly slow L’epingle hairpin and build up to top speed as you tear down the Casino Straight. Ahead lies the final chicane before the start/finish line. A tricky right left combination with the treacherous curb on the inside of the last turn ready to spit the unwary into the Wall of Champions on the opposite side of the track.
At over 300 kph you start to think about spotting your braking point. Suddenly, this comes into view….

What do you think this is, a race track ?

… and now you know what it’s like to be a programmer, who has channeled raw inspiration through his or her dancing fingers to produce a thing of beauty and elegance…only to run into the QA person pointing out that the commas are in the wrong place according to page 823, paragraph 2 sub-section e of The Coding Standards.

Often measured in weight rather than the number of pages, Coding Standards documents are often outdated, arbitrary and just plain wrong.
On the other hand, their absence can cause much heartache, not least to those poor souls in support who are trying to maintain code where the Agilista philosophy of Code over Documentation has been taken to the ultimate extreme.

What follows is an attempt to make sense of the Coding Standards conundrum.
I’ll look at what I think a Coding Standards document should contain, and what it shouldn’t.
Then I’ll give some suggestions as to standards for Oracle SQL and PL/SQL which you can either embrace or throw rocks at, depending on your preference.
Before all of that however, I feel the need for some serious catharsis…

Anchored Declarations and the Brownie Point Economy

This week’s hot conversational topic in the Nut and Squirrel was the Global Economic Crisis, with particular reference to a little-reported side-effect that has huge ramifications. I am, of course, referring to the devaluation of Brownie Points.

Unless you’re English, I guess some explanation may be called for at this point. So, at the risk of getting all anthropological…
Brownie Points are awarded by females to their mate for certain actions. When enough brownie points have been accrued, the male of the species can have these converted into a Pass.

The Pass can be used for a night out with the lads, at the footie, or whatever other pursuit is of interest.

This system is instinctively understood by females, although, due to the sudden and unexpected fluctuations in value, less so by the males.

You will often hear conversations such as :

“We’re going to Luton Saturday, their playing Mansfield, you reckon you can get a pass ?”
“Sorry, we’re at the Garden Centre on Saturday, need to earn some brownie points.”

This system goes under many different guises but is essentially the same the world over.
Lately however, it has become apparent that males are having to work harder for their brownie points than previously. The suspicion is that this is related to the retail price of women’s shoes.

What a mess. Where can you put your hard earned savings to ensure an index-linked return ?
Well, if you really want to know, you’ll have to read the Financial Times. Variables in PL/SQL, however, can be indexed-linked to database columns by the simple expedient of an anchored declaration.

ROWNUM, Random updates and the Alert Log – A trip down Memory Lane

I met up with an old friend recently. Deb and I were in Toronto, hometown of a certain Simon Jennings.
Apart from being a top bloke, Simon was also my first mentor in the mysterious ways of SQL, a sort of Obi-Wan Kenobi to my Anakin Skywalker, but without the light-sabers.

Ah those heady days when the world was young…and I first discovered ROWNUM.
That pseudo column was one I used often when investigating the structure and data within tables. It was also, the source of some confusion.

Oracle External Tables or What I did on my Holidays

This week’s missive is coming to you from the netbook. Deb and I have pushed the boat out this year and we’re currently in Canada for our holiday.
This has nothing at all to do with Oracle External Tables, but does explain the flavour of the examples that follow.

If you can’t do it in PL/SQL, do it in SQL

The tension was palpable on the bridge of The Enterprise. The hulk of the giant Teredation Cruiser filled the View Screen.

With baited breath they awaited the response of the Teredation Captain to the message they had just transmitted.

Suddenly the image on the screen changed from that of the menacing warship to Simon, Lord High Hatter of the Teredations. In a voice which hinted at an unaccustomed uncertainty, he exclaimed,
“What is this PL/SQL of which you speak ?”

OK, so maybe it wasn’t the bridge of the Enterprise so much as in the beer garden at the Nut and Squirrel. The question, however, is pretty much accurate.

In an attempt to distract himself from the sad news that Claude Gnapka had finally left Luton for Walsall, Simon posed the following programming problem :

He needed a SQL query to return the first working day on or after the 23rd of the month, together with the first working day of the following month. He wanted both dates to be returned in the same row.
The catch ? Simon works on Teradata which, owing to a glitch in the Universal Translator ( or something), doesn’t have anything like PL/SQL or T-SQL built in.

DBMS_APPLICATION_INFO – Are we nearly there yet ?

Deb has come to the conclusion that, when on a long car journey, I’m not a great passenger.
“Are we nearly there yet ?” I enquire politely…usually around five minutes into a four-hour journey.
“No, not yet”, comes the patient reply.
“Are we almost nearly there yet ?”
“No, I’ll let you know when we are”, she responds with iron patience.
A few minutes pass…
“Are we almost nearly almost there ?”
At this point, I’m usually offered the option of walking the rest of the way.
It’s the same with long-running programs on the database. I want to know how far it’s gone and how far it’s got to go.
Oh, what Deb would do for an in-car equivalent of DBMS_APPLICATION_INFO.

Deb would probably have some sympathy with the DBA who gets a call from a user who has a process running and is, essentially, asking “are we nearly there yet ?”
Fortunately, with a bit of foresight and the judicious application of a little DBMS_APPLICATION_INFO magic, the answer to this question is right there in V$SESSION, or even V$SESSION_LONGOPS.

The Oracle Database – Where’s the “Upgrade Now” button ?

Regular readers of this blog ( hello Mum) will have noticed that I’ve been a bit quiet lately.
Having tired of my life of leisure, I now have a proper job and have consequently been a little bit busy.

I have though, had time to reflect on the question of Oracle database upgrades and why they are never ever standard.

“We’ve decided to upgrade the database”. Those words are bound to strike equal amounts of anticipation and abject terror into the heart of any DBA…or at least, any DBA who has gone through an upgrade before.

To be fair, it’s not the upgrade itself that is going to cause the problem, it’s all of the misconceptions that come with it.

The CIO/IT Manager/Project Manager is almost bound at some point to come out with one of the following gems :

“We don’t need to test, the application code will still work in exactly the same way”
“We need to upgrade/move our servers/Application/Operating System anyway so we can do it all at the same time to minimize downtime.”
“According to the salesman Oracle can now make the tea, so it must be true.”

My personal favourite is from a person who shall remain nameless :

“We want to be fast-followers. X team is upgrading their database in July so we should be done by June”.

PL/SQL Injection – The Doctor Will See You Now

Following on from my recent post about SQL Injection I had an excellent comment from Gary.

Whilst we’ve been able to establish that the PL/SQL solution we implemented does not suffer the same vulnerability to injection as the concatenated SQL statement, this does lead us to a further question – does using PL/SQL automatically render us immune from injection attacks ?

For the purposes of this post, I’m going to leave PHP to one-side and concentrate on the PL/SQL side of the matter.

Oracle, bind variables and SQL Injection – Keeping out unwanted guests

My son, Michael ( yes, it is the only name I can spell), is currently following in his father’s footsteps and studying Computer Science.
As is only natural, he does occasionally have the urge to rebel against all that his parents hold dear. In his case he’s rejected the path of light and Linux and has become … a Microsoft Certified Professional. Oh the shame. Where did I go wrong ?
All of which links, if somewhat tenuously, to the subject at hand. When he took his first steps into the world of programming, we had a look at PHP ( as part of a LAMP set-up, naturally).
In one of the introductory manuals, we came across an example of how to authenticate web users against a database.
The author was clearly trying to introduce various language concepts and would certainly not claim that his example was intended for production use. However, with a bit of tweaking for use against an Oracle database, it does offer a very clear illustration one area of the potential vulnerabilities of web applications to SQL Injection attacks. It also offers the opportunity to illustrate a major benefit of using bind variables in queries against Oracle – i.e. protection against SQL Injection.

I know that a fair few people who stumble across this site are new to Oracle and want to play around with Oracle XE. These people are also usually pretty experienced in other technologies (hi Wayne, hope you’re still enjoying all that sunshine).

So, the purpose of this post is to :

• Illustrate the way in-line SQL statements can be injected
• Show how this can be countered in an Oracle database by use of bind variables
• Have a look at letting Oracle handle user authentication
• Celebrate the visionary genius of Messrs Young, Young and Johnson. “For Those About to Rock” was not merely an album of raucous Blues-based Heavy Metal, but a prophecy about the potential pitfalls of Web Application development.
• Oh, and give you the chance to laugh at my PHP prowess ( or lack thereof)

Using Flashback for Automated Testing – a stepping-stone to World Domination

As I’m between jobs at the moment, I’ve had time to sit and reflect on possible future career directions.
With Deb’s help and inspiration (“if that washing isn’t done by the time I get back, there’ll be trouble!”) I think I may have come up with something.
Yep, I’m looking into the possibility of becoming an Evil Genius and Megalomaniac. After all, I’ve already got the alter-ego – I mean, “The Anti-Kyte”, that’s never going to be a super hero name, is it ?
Apart from that, there’s also the costume to consider. Let’s face facts here, mine is just not a figure that would be flattered by brightly coloured spandex.

If I’m going to run an efficient evil empire, I’ll need a properly designed and efficient data-driven application. Everything must be thoroughly tested. The temperature control on the piranha tank failing or some meddling super-spy escaping the cunningly planned trap at a crucial moment could do untold damage to the share price.
A suite of automated tests that I can run repeatedly to verify both functionality and performance under load is a must-have.
In that case, I’m going to need a way of running my test suite, then quickly rolling back the resultant database changes, fixing any problems and running it all over again.
It’s time to take a fresh look at the Flashback functionality available in the Oracle Database.