How I Learned to Stop Worrying and Love NULL in SQL

You should not try to prevent NULL values - instead, write your query in a way to overcome its limitations.

24 days ago   •   6 min read

By Rajendra gupta
Table of contents

The NULL value is a data type that represents an unknown value. It is not equivalent to empty string or zero. Suppose you have an employee table containing columns such as EmployeeId, Name, ContactNumber and an alternate contact number. This table has a few mandatory value columns like EmployeeId, Name, and ContactNumber. However, an alternate contact number is not required and therefore has an unknown value. Therefore a NULL value in this table represents missing or inadequate information. Here are other meanings NULL can have:

  • Value Unknown
  • Value not available
  • Attribute not applicable

In this post we will consider how NULL is used in creating tables, querying, string operations, and functions. Screenshots in this post come from the Arctype SQL Client.

Allowing NULL in CREATE TABLE

To a table structure, we need to define whether the respective column allows NULL or not. For example, look at the following customer's table. The columns such as CustomerID, FirstName, LastName do not allow NULL values, whereas the Suffix, CompanyName, and SalesPerson columns can store NULL values.

CREATE  TABLE Customers(
	CustomerID SERIAL  PRIMARY  KEY,
	FirstName varchar(50) NOT  NULL,
	MiddleName varchar(50) NULL,
	LastName varchar(50) NOT  NULL,
	Suffix varchar(10) NULL,
	CompanyName varchar(128) NULL,
	SalesPerson varchar(256) NULL,
	EmailAddress varchar(50) NULL
)

Let’s insert a few records into this table using the following script.

INSERT INTO Customers 
	(FirstName, MiddleName, LastName, Suffix, CompanyName, SalesPerson, EmailAddress)
VALUES
	('John',NULL,'Peter',NULL,NULL,NULL,NULL),
	('Raj','M','Mohan','Mr','ABC','KRS','raj.mohan@abc.com'),
	('Krishna',NULL,'Kumar','MS','XYZ',NULL,'Krishna.kumar@xyz.com')
The fast and easy-to-use SQL client for developers and teams

 

Using NULL in the WHERE Clause

Now, suppose you want to fetch records for those customers who do not have an email address. The following query works fine, but it will not give us a row:

Select * FROM Customers WHERE Emailaddress=NULL
Values that are NULL cannot be queried using =
Values that are NULL cannot be queried using =

In the above select statement expression defines “Where the email address equals an UNKNOWN value”. In the SQL standard, we cannot compare a value to NULL. Instead, you refer to the value as IS NULL for this purpose. Note: There is a space between IS and NULL. If you remove space, it becomes a function ISNULL().

By using IS NULL instead of equals you can query for NULL values.
By using IS NULL instead of equals you can query for NULL values.

Integer, Decimal, and String Operations with NULL

Similarly, suppose you declared a variable but did not initialize its value. If you try to perform an arithmetic operation, it also returns NULL because SQL cannot determine the correct value for the variable, and it considers an UNKNOWN value.

SELECT 10 * NULL
Multiplying an Integer by NULL returns NULL
Multiplying an integer by NULL returns NULL
SELECT 10.0 * NULL
Multiplying a decimal by NULL returns NULL
Multiplying a decimal by NULL returns NULL

NULL also plays an important role in string concatenation.  Suppose you required the customer's full name in a single column, and you concatenate them using the pipe sign(||) .

SELECT Suffix,  FirstName, MiddleName, LastName, Suffix, 
(Suffix || ' ' || FirstName || ' ' || MiddleName || LastName ) AS CustomerFullName  FROM Customers
Setting a string to NULL and then concatenating it returns NULL
Setting a string to NULL and then concatenating it returns NULL

Look at the result set - the query returns NULL in the concatenated string if any part of the string has NULL. For example,  the person in Row 1 does not have a middle name. Its concatenated string is NULL as well, because SQL cannot validate the string value contains NULL.

There are many SQL functions available to overcome these NULL value issues in string concatenations. We’ll look at them later in this article.

The NULL value in SQL Aggregates

Suppose you use aggregate functions such as SUM, AVG, or MIN, MAX for NULL values. What do you think the expected outcome would be?

SELECT Sum(values) AS sum
    ,avg(values) as Avg
    ,Min(Values) as MinValue
    ,Max(Values) as MaxValue
  FROM (VALUES (1), (2), (3),(4), (NULL)) AS a (values);
In aggregate functions NULL is ignored.
In aggregate functions NULL is ignored.

Look at the above figure: it calculated values for all aggregated functions. SQL ignores the NULLs in aggregate functions except for COUNT() and GROUP BY(). You get an error message if we try to use the aggregate function on all NULL values.

SELECT 
    Sum(values) AS sum
    ,avg(values) as Avg
    ,Min(Values) as MinValue
    ,Max(Values) as MaxValue
           FROM (VALUES (NULL), (NULL), (NULL),(NULL), (NULL)) AS a (values);
Aggregating over all NULL values results in an error.
Aggregating over all NULL values results in an error.

ORDER BY and GROUP BY with NULL

SQL considers the NULL values as the UNKNOWN values. Therefore, if we use ORDER By and GROUP by clause with NULL value columns, it treats them equally and sorts, group them. For example, in our customer table, we have NULLs in the MilddleName column. If we sort data using this column, it lists the NULL values at the end, as shown below.

SELECT Suffix,  FirstName, MiddleName, LastName, Suffix, 
(Suffix || ' ' || FirstName || ' ' || MiddleName || LastName )
 AS CustomerFullName
 FROM Customers
 Order BY MiddleName
NULL values appear last in ORDER BY
NULL values appear last in ORDER BY

Before we use GROUP BY, let's insert one more record in the table. It has NULL values in most of the columns, as shown below.

INSERT INTO Customers (FirstName,MiddleName,LastName,Suffix,CompanyName,
SalesPerson,EmailAddress)
 values('Sant',NULL,'Joseph',NULL,NULL,NULL,NULL);

Now, use the GROUP BY clause to group records based on their suffix.

SELECT count(*) as Customercount , suffix
    FROM Customers
    Group BY Suffix
GROUP BY does treat all NULL values equally.
GROUP BY does treat all NULL values equally.

As shown above, SQL treats these NULL values equally and groups them. You get two customer counts for records that do not have any suffix specified in the customers table.

Useful Functions for Working with NULL

We explored how SQL treats NULL values in different operations. In this section, we will explore a few valuable functions to avoid getting undesirable values due to NULL.

Using NULLIF in Postgres and MySQL

The NULLIF() function compares two input values.
● If both values are equal, it returns NULL.
● In case of mismatch, it returns the first value as an output.
For example, look at the output of the following NULLIF() functions.

SELECT   NULLIF (1, 1); 
NULLIF returns NULL if two values are equal
NULLIF returns NULL if two values are equal
SELECT   NULLIF (100,0); 
NULLIF returns the first value if the values are not equal.
NULLIF returns the first value if the values are not equal.
SELECT   NULLIF ('A', 'Z'); 
NULLIF returns the first string in a string compare.
NULLIF returns the first string in a string compare.

COALESCE function

The COALESCE() function accepts multiple input values and returns the first non-NULL value. We can specify the various data types in a single COALESCE() function and return the high precedence data type.

SELECT COALESCE (NULL, 2, 5) AS NULLRESPONSE;
COALESCE returns the first non NULL data type in a list.
COALESCE returns the first non NULL data type in a list.
SELECT coalesce(null, null, 8, 2, 3, null, 4);
alt text

Summary

The NULL value type is required in a relational database to represent an unknown or missing value. You need to use the appropriate SQL function to avoid getting undesired output for operations such as data concatenation, comparison, ORDER BY, or GROUP BY. You should not try to prevent NULL values - instead, write your query in a way to overcome its limitations. This way you will learn to love NULL.

JOIN the Arctype Newsletter
Programming stories, tutorials, and database tips every 2 weeks

Spread the word

Keep reading