Collations in MariaDB Xpand

Checking the Server Collation

You can check the server collation by executing the SHOW SESSION VARIABLES statement:

SHOW SESSION VARIABLES
   LIKE 'collation_server';

+------------------+-----------------+
| Variable_name    | Value           |
+------------------+-----------------+
| collation_server | utf8_general_ci |
+------------------+-----------------+

Effect of the Server Collation

The server collation is used to determine the default collation of new databases and tables, unless a different collation is specified for a given database or table. To demonstrate this effect:

1. Confirm the current server collationThe server collation is utf8_general_ci on the Distributed Transactions topology. You can confirm that with SHOW SESSION VARIABLES:

   SHOW SESSION VARIABLES
      LIKE 'collation_server';
   

   +------------------+-----------------+
   | Variable_name    | Value           |
   +------------------+-----------------+
   | collation_server | utf8_general_ci |
   +------------------+-----------------+
   

2. Create a database:

   CREATE DATABASE db1;
   

3. You can confirm this database's default collation by querying the information_schema.SCHEMATA table:

   SELECT DEFAULT_COLLATION_NAME
   FROM information_schema.SCHEMATA
   WHERE SCHEMA_NAME='db1';
   

   +------------------------+
   | DEFAULT_COLLATION_NAME |
   +------------------------+
   | utf8_general_ci        |
   +------------------------+
   

4. Also create a table in this database, which has at least one column which contains characters. This column could be a data type such as VARCHAR, CHAR, TEXT, or any of the aliases for those types:

   CREATE TABLE db1.tab (
      id int PRIMARY KEY,
      str varchar(50)
   );
   

5. You can confirm that the table's default collation is the same as the database's default collation by querying the information_schema.TABLES table:

   SELECT TABLE_COLLATION
   FROM information_schema.TABLES
   WHERE TABLE_SCHEMA='db1'
      AND TABLE_NAME='tab';
   

   +-----------------+
   | TABLE_COLLATION |
   +-----------------+
   | utf8_general_ci |
   +-----------------+
   

6. You can also confirm the str column's collation by querying the information_schema.COLUMNS table:

   SELECT COLLATION_NAME
   FROM information_schema.COLUMNS
   WHERE TABLE_SCHEMA='db1'
      AND TABLE_NAME='tab'
      AND COLUMN_NAME='str';
   

   +-----------------+
   | COLLATION_NAME  |
   +-----------------+
   | utf8_general_ci |
   +-----------------+
   

Checking a Database's Default Collation

You can check a database's default collation by querying the information_schema.SCHEMATA table:

SELECT DEFAULT_COLLATION_NAME
FROM information_schema.SCHEMATA
WHERE SCHEMA_NAME='db1';

+------------------------+
| DEFAULT_COLLATION_NAME |
+------------------------+
| utf8_general_ci        |
+------------------------+

Setting a Database's Default Collation

A database's default collation can be set at creation by executing the CREATE DATABASE statement and specifying the DEFAULT COLLATE database option:

CREATE DATABASE db1
   DEFAULT COLLATE 'utf8mb4_bin';

The new database's default character set will be the character set associated with the specified collation.

Changing a Database's Default Collation

A database's default collation can be set by executing the ALTER DATABASE statement and specifying the DEFAULT COLLATE database option:

ALTER DATABASE db1
   DEFAULT COLLATE 'utf8mb4_bin';

The database's default character set will be the character set associated with the specified collation.

Effect of a Database's Default Collation

A database's default collation is used to determine the default collation of new tables within the database, unless a different collation is specified for a given table. To demonstrate this effect:

1. Check the database's default collation by querying the information_schema.SCHEMATA table:

   SELECT DEFAULT_COLLATION_NAME
   FROM information_schema.SCHEMATA
   WHERE SCHEMA_NAME='db1';
   

   +------------------------+
   | DEFAULT_COLLATION_NAME |
   +------------------------+
   | utf8_general_ci        |
   +------------------------+
   

2. Create a table in this database, which has at least one column which contains characters. This column could be a data type such as VARCHAR, CHAR, TEXT, or any of the aliases for those types:

   CREATE TABLE db1.tab (
      id int PRIMARY KEY,
      str varchar(50)
   );
   

3. You can confirm that the table's default collation is the same as the database's default collation by querying the information_schema.TABLES table:

   SELECT TABLE_COLLATION
   FROM information_schema.TABLES
   WHERE TABLE_SCHEMA='db1'
      AND TABLE_NAME='tab';
   

   +-----------------+
   | TABLE_COLLATION |
   +-----------------+
   | utf8_general_ci |
   +-----------------+
   

4. You can confirm that the str column is using the database's default collation by querying the information_schema.COLUMNS table:

   SELECT COLLATION_NAME
   FROM information_schema.COLUMNS
   WHERE TABLE_SCHEMA='db1'
      AND TABLE_NAME='tab'
      AND COLUMN_NAME='str';
   

   +-----------------+
   | COLLATION_NAME  |
   +-----------------+
   | utf8_general_ci |
   +-----------------+
   

Checking a Table's Default Collation

You can check a table's default collation by querying the information_schema.TABLES table:

SELECT TABLE_COLLATION
FROM information_schema.TABLES
WHERE TABLE_SCHEMA='db1'
   AND TABLE_NAME='tab';

+-----------------+
| TABLE_COLLATION |
+-----------------+
| utf8_general_ci |
+-----------------+

Setting a Table's Default Collation

A table's default collation can be set at creation by executing the CREATE TABLE statement and specifying the DEFAULT COLLATE table option:

CREATE TABLE db1.tab (
   id int PRIMARY KEY,
   str varchar(50)
) DEFAULT COLLATE 'utf8mb4_bin';

The new table's default character set will be the character set associated with the specified collation.

Changing a Table's Default Collation

A table's default collation can be set by executing the ALTER TABLE statement and specifying the DEFAULT COLLATE table option:

ALTER TABLE db1.tab
   DEFAULT COLLATE 'utf8mb4_bin';

The table's default character set will be the character set associated with the specified collation.

Effect of a Table's Default Collation

A table's default collation is used to determine the default collation of new columns within the table, unless a different collation is specified for a given column. To demonstrate this effect:

1. Check the table's default collation by querying the information_schema.TABLES table:

   SELECT TABLE_COLLATION
   FROM information_schema.TABLES
   WHERE TABLE_SCHEMA='db1'
      AND TABLE_NAME='tab';
   

   +-----------------+
   | TABLE_COLLATION |
   +-----------------+
   | utf8_general_ci |
   +-----------------+
   

2. Add a new column in this table by executing the ALTER TABLE ... ADD COLUMN statement. This column could be a data type such as VARCHAR, CHAR, TEXT, or any of the aliases for those types:

   ALTER TABLE db1.tab
      ADD COLUMN new_str VARCHAR(50);
   

3. You can confirm this column is using the table's default collation by querying the information_schema.COLUMNS table:

   SELECT COLLATION_NAME
   FROM information_schema.COLUMNS
   WHERE TABLE_SCHEMA='db1'
      AND TABLE_NAME='tab'
      AND COLUMN_NAME='new_str';
   

   +-----------------+
   | COLLATION_NAME  |
   +-----------------+
   | utf8_general_ci |
   +-----------------+
   

Checking a Column's collation

You can check a column's collation by querying the information_schema.COLUMNS table:

SELECT COLLATION_NAME
FROM information_schema.COLUMNS
WHERE TABLE_SCHEMA='db1'
   AND TABLE_NAME='tab'
   AND COLUMN_NAME='str';

+-----------------+
| COLLATION_NAME  |
+-----------------+
| utf8_general_ci |
+-----------------+

Setting a Column's Collation

A column's collation can be set at table creation by executing the CREATE TABLE statement and specifying the COLLATE column option:

CREATE TABLE db1.tab (
   id int PRIMARY KEY,
   str varchar(50) COLLATE 'utf8mb4_bin'
);

The column's character set will be the character set associated with the specified collation.

Changing a Column's Collation

A column's collation can be changed by executing the ALTER TABLE statement and specifying the MODIFY clause with the COLLATE column option:

ALTER TABLE db1.tab
   MODIFY str varchar(50)
   COLLATE 'utf8mb4_bin';

The column's character set will be the character set associated with the specified collation.

Binary Column Collations

For binary column collations, the collation compares the column's character data by comparing the binary values of each character.

For example, to see how the utf8mb4_bin collation would handle sorting strings that use the utf8mb4 character set:

1. Create a table with a column that uses the utf8mb4_bin collation by executing the CREATE TABLE statement and specifying the COLLATE column option:

   CREATE TABLE db1.tab (
      id int PRIMARY KEY,
      str varchar(50) COLLATE 'utf8mb4_bin'
   );
   

2. Insert some data into the table:

   INSERT INTO db1.tab VALUES
      (1, 'SkySQL'),
      (2, 'skysql'),
      (3, 'SKYSQL'),
      (4, 'MariaDB'),
      (5, 'mariadb'),
      (6, 'MARIADB');
   

3. Execute a query that eliminates duplicates and sorts the data:

   SELECT DISTINCT str
   FROM db1.tab
   ORDER BY str;
   

   +---------+
   | str     |
   +---------+
   | MARIADB |
   | MariaDB |
   | SKYSQL  |
   | SkySQL  |
   | mariadb |
   | skysql  |
   +---------+
   

   The results show that lower case characters and upper case characters are considered to be different, and upper case characters are sorted before lower case characters, because upper case characters have lower code points (numeric values) than lower case characters.

Case-Insensitive Column Collations

For case-insensitive column collations, the collation compares the column's character data by treating upper case and lower case characters of the same character as equal.

For example, to see how the utf8mb4_unicode_ci collation would handle sorting strings that use the utf8mb4 character set:

1. Create a table with a column that uses the utf8mb4_unicode_ci collation by executing the CREATE TABLE statement and specifying the COLLATE column option:

   CREATE TABLE db1.tab (
      id int PRIMARY KEY,
      str varchar(50) COLLATE 'utf8mb4_unicode_ci'
   );
   

2. Insert some data into the table:

   INSERT INTO db1.tab VALUES
      (1, 'SkySQL'),
      (2, 'skysql'),
      (3, 'SKYSQL'),
      (4, 'MariaDB'),
      (5, 'mariadb'),
      (6, 'MARIADB');
   

3. Execute a query that eliminates duplicates and sorts the data:

   SELECT DISTINCT str
   FROM db1.tab
   ORDER BY str;
   

   +---------+
   | str     |
   +---------+
   | mariadb |
   | skysql  |
   +---------+
   

   The results show that lower case characters and upper case characters are considered to be the same, so many of the strings were considered to be duplicates.

Case-Sensitive Column Collations

For case-sensitive column collations, the collation compares the column's character data by treating upper case and lower case characters of the same character as different.

For example, to see how the latin1_general_cs collation would handle sorting strings that use the latin1 character set:

1. Create a table with a column that uses the latin1_general_cs collation by executing the CREATE TABLE statement and specifying the COLLATE column option:

   CREATE TABLE db1.tab (
      id int PRIMARY KEY,
      str varchar(50) COLLATE 'latin1_general_cs'
   );
   

2. Insert some data into the table:

   INSERT INTO db1.tab VALUES
      (1, 'SkySQL'),
      (2, 'skysql'),
      (3, 'SKYSQL'),
      (4, 'MariaDB'),
      (5, 'mariadb'),
      (6, 'MARIADB');
   

3. Execute a query that eliminates duplicates and sorts the data:

   SELECT DISTINCT str
   FROM db1.tab
   ORDER BY str;
   

   +---------+
   | str     |
   +---------+
   | MARIADB |
   | MariaDB |
   | mariadb |
   | SKYSQL  |
   | SkySQL  |
   | skysql  |
   +---------
   

   The results show that lower case characters and upper case characters are considered to be different, and strings are sorted in lexicographical order.