Skip to main content

PL/SQL 101: Nulls in PL/SQL


The Oracle Database supports a concept of a null value, which means, essentially, that it has no value. The idea of having nulls in a relational database is controversial, but Oracle Database supports them and you need to know how they can impact your work in PL/SQL.

First, and most important, remember that:
Null is never equal to anything else, including null. And certainly 0.
Null is never not equal to anything else, including null.
DECLARE
   var INTEGER;
BEGIN
   IF var = NULL THEN ...
   IF NULL = NULL THEN ...
   IF var <> NULL THEN ...
   IF NULL != NULL THEN ...
END;

you can rest assured that the code represented by "..." will never be executed.

Note: NULL in the above code is a literal value with a non-value of null.

The same holds true for where clause predicates in a SQL statement. The following queries will never return any rows, regardless of the contents of the employees table.

SELECT *
  FROM employees
 WHERE employee_id = NULL
/

SELECT *
  FROM employees
 WHERE employee_id != NULL
/

SELECT *
  FROM employees
 WHERE NULL = NULL
/

SELECT *
  FROM employees
 WHERE NULL != NULL
/

Of course, you will rarely write code that includes explicit references to the NULL literal. You are more likely to see and write code like this:

PROCEDURE my_proc (value_in IN VARCHAR2)
IS
BEGIN
   IF value_in != 'abc' THEN ...
END;

If a NULL value could be passed to my_proc for the value_in parameter, that inequality will never evaluate to true. That might be fine, but you need to give consideration very deliberately to these questions: "Could a null value be passed in here? Should I allow a null value to be passed here?"

Then take the necessary steps to bulletproof your code from nulls (see last section in this blog post for details).

With that, let's dive into the details.

NULLs in Oracle Database

Nulls work the same in SQL and PL/SQL and, as is the case with so much else in PL/SQL, you should apply the documented descriptions from SQL to PL/SQL unless otherwise noted, so here's the doc on Nulls in SQL.

A column's value can be null unless you defined that column with a NOT NULL constraint, as in:

CREATE TABLE my_users
(
   id                 NUMBER GENERATED ALWAYS AS IDENTITY,
   user_id            INTEGER NOT NULL,
   user_name          VARCHAR2(100) NOT NULL,
   twitter_account    VARCHAR2(100)
)
/

When a row is inserted into and updated in this table, the following columns cannot be set to NULL:
  • ID
  • USER_ID
  • USER_NAME
I don't have to provide a Twitter account name.

Wait, you might be saying: Why do you have to provide a value for ID? You didn't specify NOT NULL." 

That's because identity columns can never be set to NULL. Identity columns are new to 12.1 and replace the use of sequences to generate unique values for primary keys. 

Moving on to PL/SQL, a variable's value can be null unless you declared that variable with a NOT NULL constraint or use the CONSTANT keyword to define your variable as a constant. That behavior is shown in the code below.

DECLARE
   my_name   VARCHAR2 (100);
BEGIN
   IF my_name IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Name not set');
   END IF;
END;
/

Name not set

DECLARE
   my_name   VARCHAR2 (100) NOT NULL;
BEGIN
   IF my_name IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Name not set');
   END IF;
END;
/

PLS-00218: a variable declared NOT NULL must have an initialization assignment

DECLARE
   my_name   VARCHAR2 (100) NOT NULL := 'Grandpa';
BEGIN
   IF my_name IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Name not set');
   ELSE
      DBMS_OUTPUT.put_line (my_name);
   END IF;
END;
/

Grandpa

DECLARE
   my_name   CONSTANT VARCHAR2 (100);
BEGIN
   IF my_name IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Name not set');
   END IF;
END;
/

PLS-00322: declaration of a constant 'MY_NAME' must contain an initialization assignment

DECLARE
   my_name   CONSTANT VARCHAR2 (100) := 'Grandpa';
BEGIN
   IF my_name IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Name not set');
   ELSE
      DBMS_OUTPUT.put_line (my_name);
   END IF;
END;
/

Grandpa

Note that you cannot specify that a parameter in a procedure or function must be NOT NULL. Instead, you must use a datatype that cannot be null, or add code to your subprogram to "protect" it from null values. See "Bulletproofing For Nulls" below.

NULLs and Empty Strings

Here's something to keep in mind:
Oracle Database currently treats a character value with a length of zero as null. However, this may not continue to be true in future releases, and Oracle recommends that you do not treat empty strings the same as nulls.
In other words, the following block displays "Null" twice:

BEGIN
   IF '' IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Null');
   END IF;

   IF '' IS NOT NULL
   THEN
      DBMS_OUTPUT.put_line ('Not');
   END IF;

   IF NULL IS NULL
   THEN
      DBMS_OUTPUT.put_line ('Null');
   END IF;

   IF NULL IS NOT NULL
   THEN
      DBMS_OUTPUT.put_line ('Not');
   END IF;
END;

But we encourage you to not write code that assumes that this will always be the case.

Hey, having said that, the Oracle Database team has a very strong history of not changing behavior in new releases that causes problems in the 1,000,000s of lines of existing SQL and PL/SQL code out there "in the wild."

Declarations - NULL's the Default!

Whenever you declare a variable, it's value is set to NULL - unless you assign a different value to it. So, yes, you could do the following....

DECLARE
   var2 INTEGER := NULL;
BEGIN
   ...
END;

But I suggest that you do not. You take the risk of other people reading your code and thinking "Gee, I guess Steven doesn't understand PL/SQL too well."

Instead, simply declare the variable and let the PL/SQL engine take care of the default initialization to NULL, as in:

DECLARE
   var2 INTEGER;
BEGIN
   ...
END;

Don't worry: Oracle is never going to change this behavior, so you don't have to "take out insurance."

Boolean Expressions with NULL

Study this "Truth Table" and make sure you are comfortable with everything you see in it. You can even verify it to yourself with this LiveSQL script.

Notice that if x is null or y is null, then x AND y is always null. But x OR y evaluates to TRUE if either x or y is TRUE, even if the other value is NULL. [Removed 2018-03-06. See comments by Jan below]


The most important thing to remember is that in almost every case, if any part of your expression evaluates to NULL, the entire expression will evaluate to NULL.

In addition, pay close attention to how you construct conditional statements and expressions, when an expression might evaluate to NULL.

Compare the following two blocks.

Block 1

BEGIN
   IF expr
   THEN
      do_this;
   ELSE
      do_that;
   END IF;
END;
/

Block 2

BEGIN
   IF expr
   THEN
      do_this;
   ELSIF NOT expr
   THEN
      do_that;
   END IF;
END;
/

In Block 1, if expr evaluates to NULL, then do_that will be executed.

In Block 2, if expr evaluates to NULL, then neither do_this nor do_that will be executed.

But what if you want to treat two NULLs as being equal when you compare two values? Marcus in the comments below suggests creating an "isEqual" function that handles this for you. He was inspired by an AskTOM thread from years past (2012), in which Tom suggested using DECODE (available in SQL only), since it treats two NULLs as equal:

where decode( col1, col2, 1, 0 ) = 0  -- finds differences
where decode( col1, col2, 1, 0 ) = 1  -- finds sameness - even if both NULL

Here's the implementation of Marcus's isequal function for dates:
PACKAGE BODY helper
IS
   /**
   * @desc Check whether both values are equal or NULL
   * @param p_value1 date 1
   * @param p_value2 date 2
   * @return TRUE, if both values are equal or both values are NULL, else FALSE
   */
   FUNCTION isequal (p_value1 IN DATE, p_value2 IN DATE)
      RETURN BOOLEAN
   IS
   BEGIN
      -- Check NULL, otherwise NULL is returned!
      RETURN    (p_value1 IS NOT NULL AND p_value2 IS NOT NULL AND p_value1 = p_value2)
             OR (p_value1 IS NULL AND p_value2 IS NULL);
   END isequal;
END;

The NULL Statement

NULL is not just a (not) value in PL/SQL. It is also a statement, albeit a ''no-op" (no operation)—it only passes control to the next statement. Here are two examples.

I use the NULL statement inside an ELSE clause. I don't need to do this, but sometimes it's worth being explicit and letting future programmers know that you thought about the ELSE part of this IF statement and you really really don't want to do anything if expr is not TRUE (FALSE or NULL0.

BEGIN
   IF expr
   THEN
      do_this;
   ELSE
      /* Critical to not do anything if expr is false */
      NULL;
   END IF;
END;
/

In this next block, I use NULL; as the "target" for a GOTO statement. Yes, that's right, PL/SQL has a GOTO statement and you should rarely if ever need to use it. If/when you do, however, you need at least one executable statement after your label. If the point of the GOTO is simply to go to the end of the subprogram, then NULL; is the perfect statement with which to finish up.

BEGIN
   IF expr
   THEN
      GOTO do_nothing;
   END IF;

   more_stuff;

   <<do_nothing>>
   NULL;
END;
/

Bulletproofing For Nulls

Null values can be a real pain. They can cause unexpected failures and exceptions in your algorithms. They can mess up queries, either returning rows you didn't want or excluding rows when you want them.

So the first thing to do is decide very careful where and when you want to allow nulls. If a column's value should never be null, then define it that way. If a variable should never be set to NULL, then add the NOT NULL constraint or define it as a constant (as appropriate).

If you are writing a subprogram (procedure or function) and a parameter's value should never be null, you cannot add "NOT NULL" to the parameter definition. Instead, you will need to do one of the follwogin:

1. Use a datatype for the parameter that cannot be null. These can be base datatypes, such as NATURALN, or subtypes. See the examples below.

2. Add code to your subprogram to check or assert that the actual argument value is not null.

So yes there are some base datatypes which are inherently NOT NULL-able, such as NATURALN.

DECLARE
   n   NATURALN;
BEGIN
   NULL;
END;
/

PLS-00218: a variable declared NOT NULL must have an initialization assignment

And when I use that datatype for my parameter, it stops NULLs cold in their tracks.

CREATE OR REPLACE PACKAGE pkg AUTHID DEFINER
AS
   PROCEDURE no_nulls (n_in IN NATURALN);
END;
/

CREATE OR REPLACE PACKAGE BODY pkg
AS
   PROCEDURE no_nulls (n_in IN NATURALN)
   IS
   BEGIN
      DBMS_OUTPUT.put_line (n_in);
   END;
END;
/

BEGIN
   pkg.no_nulls (NULL);
END;
/

PLS-00567: cannot pass NULL to a NOT NULL constrained formal parameter

What if you are passing a string to your procedure and there is no base VARCHAR2 datatype which excludes NULLs? Then you declare your own subtype and make that new datatype not-nullable. Let's take a look:

CREATE OR REPLACE PACKAGE pkg AUTHID DEFINER
AS
   SUBTYPE nn_varchar2 IS VARCHAR2(100) NOT NULL;
   PROCEDURE no_nulls (n_in IN nn_varchar2);
END;
/

CREATE OR REPLACE PACKAGE BODY pkg
AS
   PROCEDURE no_nulls (n_in IN nn_varchar2)
   IS
   BEGIN
      DBMS_OUTPUT.put_line (n_in);
   EXCEPTION
      WHEN OTHERS
      THEN
         DBMS_OUTPUT.put_line ('Error!');
         RAISE;
   END;
END;
/

BEGIN
   pkg.no_nulls (NULL);
END;
/

PLS-00567: cannot pass NULL to a NOT NULL constrained formal parameter

Note that the exceptions raised above occur before the procedure is even executed, so the exception handler of the procedure cannot trap the exception (you do not see "Error!" after trying to run pkg.no_nulls). That may be just what you want. But:

a. If you cannot use a NOT NULL datatype for your parameter, or....
b. You want to be able to trap the exception inside the subprogram....

Then you should write your own assertion code that runs right at the top of the subprogram. Here's an example:

CREATE OR REPLACE PACKAGE pkg AUTHID DEFINER
AS
   PROCEDURE no_nulls (n_in IN VARCHAR2);
END;
/

CREATE OR REPLACE PACKAGE BODY pkg
AS
   PROCEDURE no_nulls (n_in IN VARCHAR2)
   IS
   BEGIN
      IF n_in IS NULL
      THEN
         RAISE_APPLICATION_ERROR (-20000,
           'NULLs not allowed in pkg.no_nulls');
      END IF;

      DBMS_OUTPUT.put_line (n_in);
   EXCEPTION
      WHEN OTHERS
      THEN
         DBMS_OUTPUT.put_line ('Error! ' || SQLERRM);
         RAISE;
   END;
END;
/

BEGIN
   pkg.no_nulls (NULL);
END;
/

Error! ORA-20000: NULLs not allowed in pkg.no_nulls

Better yet is to use a pre-defined and application-standard assertion package. That way, your code could look more like this:

CREATE OR REPLACE PACKAGE pkg AUTHID DEFINER
AS
   PROCEDURE no_nulls (n_in IN VARCHAR2);
END;
/

CREATE OR REPLACE PACKAGE BODY pkg
AS
   PROCEDURE no_nulls (n_in IN VARCHAR2)
   IS
   BEGIN
      assert.is_not_null (n_in, 'NULLs not allowed in pkg.no_nulls');

      DBMS_OUTPUT.put_line (n_in);
   EXCEPTION
      WHEN OTHERS
      THEN
         DBMS_OUTPUT.put_line ('Error! ' || SQLERRM);
         RAISE;
   END;
END;
/

Don't have an assertion package? No problem! Grab mine from LiveSQL.

Avoid NULL Headaches

It really doesn't matter if you believe ever-so-strongly that Oracle (and other RDBMS vendors) should not have allowed NULLs in its relational database. If you'd like to go down that rabbit hole, start here.

NULLs are in Oracle Database and they are here to stay.

So you should understand nulls, the NULL value, the NULL statement, and how nulls can cause confusion and errors.

Then take steps in your code to avoid that confusion and those errors.

1. Do everything that you can declaratively (such as defining a column as NOT NULL).

2. Bullet-proof your subprograms with assertion logic to ensure that your program is free of nulls (when that is how it should be).

Comments

  1. Hello Steven,
    In the section "NULLs in Oracle Database" you stated that the column USER_NAME in the table cannot be set to NULL but you haven't specified a "NOT NULL" constraint on that column.
    Kind regards, Niels

    ReplyDelete
  2. You might add how to "compare" NULLs. Sometimes it is necessary to treat NULL like an existing value.
    We have an overloaded isEqual function and we use DECODE in SQL as suggested by Tom:
    http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:4410543900346652511

    ReplyDelete
    Replies
    1. Thanks, I'm adding this to the Boolean Experssions with NULL section. Would you care to share your isEqual function(s)?

      Delete
    2. PACKAGE helper ...
      /**
      * @desc Check whether both values are equal or NULL
      * @param p_value1 date 1
      * @param p_value2 date 2
      * @return TRUE, if both values are equal or both values are NULL, else FALSE
      */
      FUNCTION isEqual
      (
      p_value1 IN DATE
      ,p_value2 IN DATE
      )
      RETURN BOOLEAN
      IS
      BEGIN
      RETURN ( p_value1 IS NOT NULL-- Check NULL, otherwise NULL is returned!
      AND p_value2 IS NOT NULL
      AND p_value1 = p_value2
      )
      OR ( p_value1 IS NULL
      AND p_value2 IS NULL
      );

      END isEqual;

      overloaded for VARCHAR and NUMBER

      Delete
  3. Hi Steven below is the my requirement could you please help on the Same

    Requirement:
    Modify the ca_custodian_iso_map_rule table using following
    1.). Manage all mappings from event bridge table to source table.

    ReplyDelete
    Replies
    1. bb, I am sorry but this requirement means nothing to me.

      Delete
  4. Hi Steven,
    in the section "Boolean Expressions with NULL" you state the following:

    "Notice that if x is null or y is null, then x AND y is always null."

    This is not the case and you show it just below in the Logical Truth Table (and even with your live sql script).

    TRUE AND NULL is NULL, but
    FALSE AND NULL is FALSE

    And I really can understand that behavior. In "FALSE AND NULL" we know that the final result will be "false" as one of the two boolean expressions is "false". There is no need to check the second expression, which has no defined value in this case (and so might result in a NULL result).

    Best regards. Jan

    ReplyDelete
    Replies
    1. Jan, could you please point me to a LiveSQL script that proves the above assertion? The truth table above says x or y is NULL if x is FALSE and y is NULL. "FALSE OR NULL" is FALSE. But AND? I don't think so. Please, prove it! Thanks!

      Delete
    2. Hi Steven,
      the truth table says in row 6 column 3 that "FALSE AND NULL" is FALSE.
      And it says in the same row 6 column 4 that "FALSE OR NULL" is NULL.
      And your LiveSQL script already proved that.
      Here ist "my" prove: I just took your script, added a translation to text for the boolean values and reduced it to the cases in question: https://livesql.oracle.com/apex/livesql/s/gc9p0rsziru2vzwxf3sstmo21

      Did I get something wrong?

      Best regards, Jan

      Delete
    3. Ha! That's what I get for answering commentary on my blog at 4:45 AM. Sorry bout that, Jan, you are right and I am wrong.

      Delete
    4. No worries!
      I had an advantage - I'm 9 hours ahead and already had my sleep. ;)

      Delete

Post a Comment

Popular posts from this blog

Quick Guide to User-Defined Types in Oracle PL/SQL

A Twitter follower recently asked for more information on user-defined types in the PL/SQL language, and I figured the best way to answer is to offer up this blog post. PL/SQL is a strongly-typed language . Before you can work with a variable or constant, it must be declared with a type (yes, PL/SQL also supports lots of implicit conversions from one type to another, but still, everything must be declared with a type). PL/SQL offers a wide array of pre-defined data types , both in the language natively (such as VARCHAR2, PLS_INTEGER, BOOLEAN, etc.) and in a variety of supplied packages (e.g., the NUMBER_TABLE collection type in the DBMS_SQL package). Data types in PL/SQL can be scalars, such as strings and numbers, or composite (consisting of one or more scalars), such as record types, collection types and object types. You can't really declare your own "user-defined" scalars, though you can define subtypes  from those scalars, which can be very helpful from the p

The differences between deterministic and result cache features

 EVERY once in a while, a developer gets in touch with a question like this: I am confused about the exact difference between deterministic and result_cache. Do they have different application use cases? I have used deterministic feature in many functions which retrieve data from some lookup tables. Is it essential to replace these 'deterministic' key words with 'result_cache'?  So I thought I'd write a post about the differences between these two features. But first, let's make sure we all understand what it means for a function to be  deterministic. From Wikipedia : In computer science, a deterministic algorithm is an algorithm which, given a particular input, will always produce the same output, with the underlying machine always passing through the same sequence of states.  Another way of putting this is that a deterministic subprogram (procedure or function) has no side-effects. If you pass a certain set of arguments for the parameters, you will always get

My two favorite APEX 5 features: Regional Display Selector and Cards

We (the over-sized development team for the PL/SQL Challenge - myself and my son, Eli) have been busy creating a new website on top of the PLCH platform (tables and packages): The Oracle Dev Gym! In a few short months (and just a part time involvement by yours truly), we have leveraged Oracle Application Express 5 to create what I think is an elegant, easy-to-use site that our users will absolutely love.  We plan to initially make the Dev Gym available only for current users of PL/SQL Challenge, so we can get feedback from our loyal user base. We will make the necessary adjustments and then offer it for general availability later this year. Anyway, more on that as the date approaches (the date being June 27, the APEX Open Mic Night at Kscope16 , where I will present it to a packed room of APEX experts). What I want to talk about today are two features of APEX that are making me so happy these days: Regional Display Selector and Cards. Regional Display Sel