SQL & PL/SQL
If I had to recommend THE one most exploitable skill to have for an Oracle Developer or DBA, it would have to be HOW TO PLAY GOLF. I have observed few other talents give regular Mary and Joe, greater access to people from all levels of the corporate world. You will meet everyone from the janitor, to management elite without the fear or intimidation that we sometimes feel in these circles. For some reason, on the fairway all are equal; the only things that matter are your swing and your handicap. Alas, I never learned to play golf, so I must settle for imparting an actual job skill. And for that, my best first choice would be SQL FROM SQL.
Accessing the database from the outside world basically comes down to two options - direct querying or executing stored procedures. Procedural access is often chosen for the wrong reasons - making maintenance significantly harder.
One handy new feature of 10gR2 is its vastly expanded ability to do case-insensitive sorts and compares (a subset of its expanded ability with other special sorts and compares, such as special linguistic sorts).
A common practice for handling errors in PL/SQL procedures is to catch all errors in the top-most database layer and convert them into error codes and human readable messages for client applications. This technique is a relict from the past and, in fact, a very bad practice from today's perspective, since it can lead to data corruption.
Recently a friend asked me a simple question; "How do I display the results of a REFCURSOR through SQL*Plus?". The answer is of course is just as simple as the question, you "SELECT" it like anything else. He just had not seen the syntax before. Being me though, I wouldn't let him get away with such an easy offing, so I sent him instead a quick write-up on REFCURSORS and, the just as interesting, CURSOR EXPRESSIONS. This write-up shows various coding samples of how to use these to great effect. He liked it. He said it should be written up somewhere, and I knew just where. Hope you like it too. If you do, please add something in a reply, especially if you have a better or unique or interesting way of exploiting these features. I'd like to read about it, maybe even steal your code...
Recently, I have been helping a company in redesigning their schema, and one issue was that they stored hieracal trees in varchar2(4000) column, one full branch per row.
Data was if this kind:
/Oracle/8/1/5/Microsoft /Oracle/8/1/5/VMS /Oracle/8/1/5/0/1/VMS
Their main problem was that searching in this was really ugly, as they had to full table/index scan a lot, when searching using like '%%'
I came up with this solution to convert this into a small table that can be queried using connect by instead:
CREATE OR REPLACE TYPE node_parent_type AS OBJECT ( lvl NUMBER, node VARCHAR2(200), PARENT VARCHAR2(200) )
PL/SQL is an excellent language for Oracle. Integrated with the database, highly useful extensions to SQL, extremely powerful when exploiting Oracle Objects, and in spite of all this, it is still way easy to learn. But like any procedural programming language, one can get lazy with it, tending to accept the first solution arrived at. Looping constructs in particular seem to be used as crutches rather than necessary components of a solution. Bad looping causing performance issues is a problem that liters the PL/SQL landscape. But it is easy to spot and fix.
OUTER-JOIN is a very handy feature of SQL. But its value at solving certain classes of SQL query problems aside, it is also one of the most error prone Oracle SQL extensions we can put to use. Even advanced developers can make these mistakes. So let us discuss the error prone nature of this feature, and how to fix it.
PL/SQL is a great language. It's relatively simple to learn, is well integrated with the Oracle database, and can often be the most efficient way to perform complex or large scale database operations. In fact PL/SQL is so useful, it's difficult to believe that its origin is SQL*Forms -- and that PL/SQL was once was an optional cost add-on to the database. Working with Oracle over the past few decades has come a long way, and PL/SQL had evolved into a mature, robust and highly functional database language.
Top-n Query is one of the more advanced SQL problems. How does one retrieve N first (or least) rows from a record set? For example, how does one find the top five highest-paid employees in a given department?