The NULL value

Several considerations need to be taken regarding SQL columns allowing NULL values.

SQL Basics about NULL

The SQL concept of NULL means "no value" or "undefined value". This is used in SQL to insert or update table rows and omit values for some fields that are not mandatory.

The NULL value is not zero: Zero is a real, numeric and non-null value.

Depending on the database engine, an empty string is null, or non-null. For more details, see Empty strings and NULLs.

The [NOT] NULL column constraint

SQL table columns can be defined with the constraint NULL or NOT NULL (to deny NULL values):
CREATE TABLE customer (
   cust_id INTEGER NOT NULL PRIMARY KEY,
   cust_name VARCHAR(50) NOT NULL,
   cust_creation DATETIME YEAR TO SECOND NOT NULL,
   cust_address VARCHAR(100) NULL,
   ...
)

When omitting the NULL or NOT NULL constraint in CREATE TABLE, most database engines define the column as nullable. However, when no constraint is specified, some database brands such as Microsoft™ SQL Server deny NULL values in their default configuration. There is an SQL Server database option for that: See SQL table definition in the SQL Server adaptation guide.

Mandatory information should always be defined with a NOT NULL constraint. For example, in an SQL table storing customer information, columns such as the customer name, the row creation date/timestamp, columns referencing other tables rows with foreign keys, and obviously the primary key column, should be defined with the NOT NULL constraint. It is also a good practice to deny NULL values for flag/option columns like a Yes/No status, or gender information.

On the other hand, secondary data such as a comment field, or fields that can only be assigned after the first row creation (such as update timestamps), should accept NULL values.

Comparing NULL values

When NULL values enter in the game, it is important to understand that in means "no value".

When used in comparison, arithmetic expressions, the SQL engine can't do anything with NULL. In such case, the expression usually evaluates to NULL as well.
Important:

Do not use the = or != operator to check for NULL values!

The standard SQL operator to check for NULL values is the IS [NOT] NULL operator.
SELECT COUNT(*) FROM customers WHERE cust_comment IS NULL
Furthermore, when manipulating non-null values, keep in mind what happens with an operator, when NULL values can show up. In the next SQL example, without knowing the semantics of the != comparison operator, it's not obvious to guess if the last SELECT statement is supposed to return only the row with value 200:
CREATE TABLE tab1 ( col1 INTEGER );
INSERT INTO tab1 VALUES (NULL);
INSERT INTO tab1 VALUES (100);
INSERT INTO tab1 VALUES (300);
SELECT SUM(col1) FROM tab1;
SELECT AVG(col1) FROM tab1;
SELECT * FROM tab1 WHERE col1 != 100;

To simplify the usage of SQL columns that can be NULL, the most database engines provide SQL functions to replace a potential NULL value by a default, non-null value:

Table 1. SQL functions for NULL by database brand
Database Server Type NVL() ISNULL() IFNULL() COALESCE()
IBM® Informix® Yes No No Yes
IBM Netezza Yes Yes No Yes
Microsoft SQL Server No Yes No Yes
Oracle® MySQL No No Yes Yes
MariadDB Yes No Yes Yes
Oracle Database Server Yes No Yes Yes
PostgreSQL No No No Yes
SAP HANA® No No Yes Yes
SQLite No No Yes Yes
Dameng® Yes Yes No Yes

As you can see, the COALESCE() function, which is standard ANSI, is supported by all database engine brands supported by Genero, and since it follows simple SQL function syntax, it can be used in static SQL.

Sorting NULL values

When sorting SQL rows with the ORDER BY clause on a column that accepts NULL values, pay attention to the way the database engine considers nulls versus non-null values when sorting.

For example, with Informix, when using ORDER BY colname [ASC], rows with NULL appear at the top of the result set. This is different with PostgreSQL, where null values by default sort as if larger than other non-null value.

Table 3. Position of rows with NULL when sorted
Database Server Type NULL rows position (ASC)
IBM Informix First
Microsoft SQL Server First
Oracle MySQL Last
MariadDB First
Oracle Database Server Last
PostgreSQL Last
SAP HANA First
SQLite First

Consider using NOT NULL constraints for ORDER BY sort columns, when the position of the rows with NULL values matters.