INF: Handling of NULLs, Null Strings, and Blanks in SQL Server

ID Number: Q45917

1.00 1.10 1.11 4.20

OS/2

Summary:

The following information discusses how NULLS, null strings, and

blank strings are handled in SQL Server.

More Information:

SQL Server automatically strips trailing blanks from character strings

before storing the string if the data type is VARCHAR, or if the data

type is CHAR and nulls are allowed. A trailing blank that is also the

first character in the string is not stripped. Trailing blanks are

preserved if the data type is CHAR and nulls are not allowed.

If the string contains only blanks (0x20), a single blank is stored.

If the string has zero length and contains nothing at all (that is,

""), SQL Server converts it to a single blank before storing it.

The only way to get SQL Server to store a true null is to use the

"NULL" keyword in the INSERT or UPDATE statement.

For example, suppose table T1 has a column of type CHAR(10) with nulls

allowed, as follows:

INSERT T1 VALUES (NULL) <--- Inserts a true null

INSERT T1 VALUES ("") <--- Inserts a single blank (0x20)

INSERT T1 VALUES (" ") '' '' '' ''

INSERT T1 VALUES (" ") '' '' '' ''

INSERT T1 VALUES (" ") '' '' '' ''

If the data type is CHAR (fixed length) with nulls allowed, the

trailing blanks that were stripped are not restored by DB-LIBRARY

(db-lib). For example, if a column is defined as CHAR(10) NULL and if

('rec1 ') is inserted, the dbdatlen() function will return 4. If

the column is defined as CHAR(10) without nulls allowed and the same

data loaded, the dbdatlen() function will return 10.

Additional reference words: 1.00 1.10 1.11 4.20 Transact-SQL dblib