Debbie Saves Christmas – Database Development in a Devops Wonderland : Re-runnable DDL

Debbie felt a shiver run down her spine. To be fair, that wasn’t much of a surprise since Lapland at this time of the year does tend to be a little chilly.
However, it wasn’t the weather that was the cause of her discomfort. Someone high up in the IT Department of her employer, The National Elf ( aka Santa’s Grotto) had decided that Continuous Integration was the way to go and had decreed that it should be used forthwith across all projects and technologies in the Company.
This included the application that Debbie was responsible for.
Written around 15 years ago, this Stock Control Application had already survived one major database upgrade but was now resolutely “stuck” on Oracle 11g.
The thing about so many modern software development techniques is that they were based on the premise that code was file based. Of course, this was also true ( or at least, true enough) for some database objects, but tables were a little different.
You couldn’t simply “replace” a table like you could any other program as doing so would destroy any data in that table. For this reason, any changes required to tables for a mature application such as this would be applied by means of DDL ALTER statements.
Of course, there are tools around for this sort of thing. Liquibase, FlexDeploy – these were just two of the tools that Debbie had no chance of getting approval to use in the face of a bureaucracy that made the Vogon Civil Service look like it was following Extreme Programming.
If she was going to get her changes through by her Christmas Eve deadline, she would have to get creative…

Advertisements

utPLSQL 3.0 – How to have your cake and eat it

“You can’t have your cake and eat it !” This seems to be a regular refrain from the EU in the ongoing Brexit negotiations.
They also seem to be a bit intolerant of “cherry picking”.
I’ve never really understood the saying, “You can’t have your cake and eat it”.
What’s the point in having the cake unless you are going to eat it ?
Fortunately, I’m not alone in my perplexity – just ask any Brexiteer member of the British Cabinet.
For those who want to make sense of it ( the saying, not Brexit), there is a handy Wikepedia page that explains all.

When it comes to Unit Testing frameworks for PL/SQL, compromise between cake ownership and consumption is usually required.
Both utPLSQL 2.0 and ruby-plsql-spec have their good points, as well as some shortcomings.
Of course, if you want a more declarative approach to writing Unit Tests, you can always use TOAD or SQLDeveloper’s built-in tools.

Recently, a new player has arrived on the PL/SQL testing scene.
Despite it’s name, utPLSQL 3.0 appears to be less an evolution of utPLSQL 2.0 as a new framework all of it’s own.
What I’m going to do here, is put utPLSQL 3.0 through it’s paces and see how it measures up to the other solutions I’ve looked at previously.
Be warned, there may be crumbs…

ORA-06592 and the Case of the Happy Australians

Another Ashes Tour to Australia has come and gone and the home team once again hold The Urn.
For any non-cricket fans, I should probably explain.
Every four years, England sends their Men’s and Women’s Cricket Teams to Australia on a goodwill mission.
The object of the exercise is to make Australians feel good about their country as their teams inevitably triumph.

These recently concluded contests provide the theme for the illustration of the less-than-straightforward circumstance surrounding the ORA-06592 error which follows.
When encountering this error, you’ll probably see something like

ORA-06592: CASE not found while executing CASE statement

06592. 00000 -  "CASE not found while executing CASE statement"
*Cause:    A CASE statement must either list all possible cases or have an
           else clause.
*Action:   Add all missing cases or an else clause.

Despite this apparently definitive advice, you don’t always need to cover any possible case, or include an ELSE clause…

Private Functions and ACCESSIBLE BY Packages in 12c

My recent post about PLS-00231 prompted an entirely reasonable question from Andrew :

“OK so the obvious question why [can’t you reference a private function in SQL] and doesn’t that defeat the objective of having it as a private function, and if so what about other ways of achieving the same goal ?”

I’ll be honest – that particular post was really just a note to self. I tend to write package members as public initially so that I can test them by calling them directly.
Once I’ve finished coding the package, I’ll then go through and make all of the helper package members private. My note was simply to remind myself that the PLS-00231 error when compiling a package usually means that I’ve referenced a function in a SQL statement and then made it private.

So, we know that a PL/SQL function can only be called in a SQL statement if it’s a schema level object or it’s definied in the package header because that’s the definition of a Public function in PL/SQL. Or at least it was…

In formulating an answer to Andrew’s question, it became apparent that the nature of Private functions have evolved a bit in 12c.

So, what I’m going to look at here is :

• What are Private and Public package members in PL/SQL and why you might want to keep a package member private
• How 12c language features change our definition of private and public in terms of PL/SQL objects
• Hopefully provide some up-to-date answers for Andrew

PLS-00231 – The best laid (cunning) plans and private package functions

There are times when I feel like Baldrick.
One moment I’m all boundless optimism and cunning plans and the next, I’m so confused I don’t know what my name is or where I live.
One such recent bout of existential uncertainty was caused by the error mentioned in the title of this post, or to give it it’s full name :

PLS-00231 : Function <function name> may not be used in SQL

So, from the beginning…

Using Edition Based Redefinition for Rolling Back Stored Program Unit Changes

We had a few days of warm, sunny weather in Milton Keynes recently and this induced Deb and I to purchase a Garden Umberella to provide some shade.
After a lifetime of Great British Summers we should have known better. The sun hasn’t been seen since.
As for the umbrella ? Well that does still serve a purpose – it keeps the rain off.

Rather like an umbrella Oracle’s Edition Based Redefinition feature can be utilized for purposes other than those for which it was designed.
Introducted in Oracle Database 11gR2, Edition Based Redefinition (EBR to it’s friends) is a mechanism for facilitating zero-downtime releases of application code.
It achieves this by separating the deployment of code to the database and that code being made visible in the application.

To fully retro-fit EBR to an application, you would need to create special views – Editioning Views – for each application table and then ensure that any application code referenced those views and not the underlying tables.
Even if you do have a full automated test suite to perform your regression tests, this is likely to be a major undertaking.
The other aspect of EBR, one which is of interest here, is the way it allows you to have multiple versions of the same stored program unit in the database concurrently.

Generally speaking, as a database application matures, the changes made to it tend to be in the code rather more than in the table structure.
So, rather than diving feet-first into a full EBR deployment, what I’m going to look at here is how we could use EBR to:

• decouple the deployment and release of stored program units
• speed up the process of rolling back the release of multiple stored program unit changes
• create a simple mechanism to roll back individual stored program unit changes

There’s a very good introductory article to EBR on OracleBase.
Whilst you’re here though, forget any Cross-Edition Trigger or Editioning View complexity and let’s dive into…

Dude, Where’s My File ? Finding External Table Files in the midst of (another) General Election

It’s early summer in the UK, which means it must be time for an epoch defining vote of some kind. No, I’m not talking about Britain’s Got Talent.
Having promised that there wouldn’t be another General Election until 2020, our political classes have now decided that they can’t go any longer without asking us what we think. Again.
Try as I might, it may not be possible to prevent the ear-worm phrases from the current campaign slipping into this post.
What I want to look at is how you can persuade Oracle to tell you the location on disk of any files associated with a given external table.
Specifically, I’ll be covering :

• getting the name of the Database Server
• finding the fully qualified path of the datafile the external table is pointing to
• finding other files associated with the table, such as logfiles

In the course of this, we’ll be challenging the orthodoxy of Western Capitalism “If You Can Do It In SQL…” with the principle of DRY ( Don’t Repeat Yourself).
Hopefully I’ll be able to come up with a solution that is “Strong and Stable” and yet at the same time “Works For The Many, Not the Few”…