This chapter covers how one creates the database structures that will hold one's data. In a relational database, the raw data is stored in tables, so the majority of this chapter is devoted to explaining how tables are created and modified and what features are available to control what data is stored in the tables. Subsequently, we discuss how tables can be organized into namespaces, and how privileges can be assigned to tables. Finally, we will briefly look at other features that affect the data storage, such as views, functions, and triggers. Detailed information on these topics is found in &cite-programmer;. A table in a relational database is much like a table on paper: It consists of rows and columns. The number and order of the columns is fixed, and each column has a name. The number of rows is variable -- it reflects how much data is stored at a given moment. SQL does not make any guarantees about the order of the rows in a table. When a table is read, the rows will appear in random order, unless sorting is explicitly requested. This is covered in . Furthermore, SQL does not assign unique identifiers to rows, so it is possible to have several completely identical rows in a table. This is a consequence of the mathematical model that underlies SQL but is usually not desirable. Later in this chapter we will see how to deal with this issue. Each column has a data type. The data type constrains the set of possible values that can be assigned to a column and assigns semantics to the data stored in the column so that it can be used for computations. For instance, a column declared to be of a numerical type will not accept arbitrary text strings, and the data stored in such a column can be used for mathematical computations. By contrast, a column declared to be of a character string type will accept almost any kind of data but it does not lend itself to mathematical calculations, although other operations such as string concatenation are available. PostgreSQL includes a sizable set of built-in data types that fit many applications. Users can also define their own data types. Most built-in data types have obvious names and semantics, so we defer a detailed explanation to . Some of the frequently used data types are integer for whole numbers, numeric for possibly fractional numbers, text for character strings, date for dates, time for time-of-day values, and timestamp for values containing both date and time. To create a table, you use the aptly named CREATE TABLE command. In this command you specify at least a name for the new table, the names of the columns and the data type of each column. For example: CREATE TABLE my_first_table ( first_column text, second_column integer ); This creates a table named my_first_table with two columns. The first column is named first_column and has a data type of text; the second column has the name second_column and the type integer. The table and column names follow the identifier syntax explained in . The type names are also identifiers, but there are some exceptions. Note that the column list is comma-separated and surrounded by parentheses. Of course, the previous example was heavily contrived. Normally, you would give names to your tables and columns that convey what kind of data they store. So let's look at a more realistic example: CREATE TABLE products ( product_no integer, name text, price numeric ); (The numeric type can store fractional components, as would be typical of monetary amounts.) When you create many interrelated tables it is wise to choose a consistent naming patter for the tables and columns. For instance, there is a choice of using singular or plural nouns for table names, both of which are favored by some theorist or other. There is a limit on how many columns a table can contain. Depending on the column types, it is between 250 and 1600. However, defining a table with anywhere near this many columns is highly unusual and often a questionable design. If you don't need a table anymore, you can remove it using the DROP TABLE command. For example: DROP TABLE my_first_table; DROP TABLE products; Attempting to drop a table that does not exist is an error. Nevertheless, it is common in SQL script files to unconditionally try to drop each table before creating it, ignoring the error messages. If you need to modify a table that already exists look into later in this chapter. With the tools discussed so far you can create fully functional tables. The remainder of this chapter is concerned with adding features to the table definition to ensure data integrity, security, or convenience. If you are eager to fill your tables with data now you can skip ahead to and read the rest of this chapter later. A column can be assigned a default value. When a new row is created and no values are specified for some of the columns, the columns will be filled with their respective default values. A data manipulation command can also request explicitly that a column be set to its default value, without knowing what this value is. (Details about data manipulation commands are in the next chapter.) If no default value is declared explicitly, the null value is the default value. This usually makes sense because a null value can be thought to represent unknown data. In a table definition, default values are listed after the column data type. For example: CREATE TABLE products ( product_no integer PRIMARY KEY, name text, price numeric DEFAULT 9.99 ); The default value may be a scalar expression, which well be evaluated whenever the default value is inserted (not when the table is created). Data types are a way to limit the kind of data that can be stored in a table. For many applications, however, the constraint they provide is too coarse. For example, a column containing a product price should probably only accept positive values. But there is no data type that accepts only positive numbers. Another issue is that you might want to constrain column data with respect to other columns or rows. For example, in a table containing product information, there should only be one row for each product number. To that end, SQL allows you to define constraints on columns and tables. Constraints give you as much control over the data in your tables as you wish. If a user attempts to store data in a column that would violate a constraint, an error is raised. This applies even if the value came from the default value definition. A check constraint is the most generic constraint type. It allows you to specify that the value in a certain column must satisfy an arbitrary expression. For instance, to require positive product prices, you could use: CREATE TABLE products ( product_no integer, name text, price numeric CHECK (price > 0) ); As you see, the constraint definition comes after the data type, just like default value definitions. Default values and constraints can be listed in any order. A check constraint consists of the key word CHECK followed by an expression in parentheses. The check constraint expression should involve the column thus constrained, otherwise the constraint would not make too much sense. You can also give the constraint a separate name. This clarifies error messages and allows you to refer to the constraint when you need to change it. The syntax is: CREATE TABLE products ( product_no integer, name text, price numeric CONSTRAINT positive_price CHECK (price > 0) ); To specify a named constraint, use the key word CONSTRAINT followed by an identifier followed by the constraint definition. A check constraint can also refer to several columns. Say you store a regular price and a discounted price, and you want to ensure that the discounted price is lower than the regular price. CREATE TABLE products ( product_no integer, name text, price numeric CHECK (price > 0), discounted_price numeric CHECK (discounted_price > 0), CHECK (price > discounted_price) ); The first two constraints should look familiar. The third one uses a new syntax. It is not attached to a particular column, instead it appears as a separate item in the comma-separated column list. In general, column definitions and constraint definitions can be listed in mixed order. We say that the first two are column constraints, whereas the third one is a table constraint because it is written separately from the column definitions. Column constraints can also be written as table constraints, while the reverse is not necessarily possible. The above example could also be written as CREATE TABLE products ( product_no integer, name text, price numeric, CHECK (price > 0), discounted_price numeric, CHECK (discounted_price > 0), CHECK (price > discounted_price) ); or even CREATE TABLE products ( product_no integer, name text, price numeric CHECK (price > 0), discounted_price numeric, CHECK (discounted_price > 0 AND price > discounted_price) ); It's a matter of taste. It should be noted that a check constraint is satisfied if the check expression evaluates to true or the null value. To ensure that a column does not contain null values, the not-null constraint described in the next section should be used. A not-null constraint simply specifies that a column must not assume the null value. A syntax example: CREATE TABLE products ( product_no integer NOT NULL, name text NOT NULL, price numeric ); A not-null constraint is always written as a column constraint. A not-null constraint is equivalent to creating a check constraint CHECK (column_name IS NOT NULL), but in PostgreSQL creating an explicit not-null constraint is more efficient. The drawback is that you cannot give explicit names to not-null constraints created that way. Of course, a column can have more than one constraint. Just write the constraints after one another: CREATE TABLE products ( product_no integer NOT NULL, name text NOT NULL, price numeric NOT NULL CHECK (price > 0) ); The order doesn't matter. It does not necessarily affect in which order the constraints are checked. The NOT NULL constraint has an inverse: the NULL constraint. This does not mean that the column must be null, which would surely be useless. Instead, this simply defines the default behavior that the column may be null. The NULL constraint is not defined in the SQL standard and should not be used in portable applications. (It was only added to PostgreSQL to be compatible with other database systems.) Some users, however, like it because it makes it easy to toggle the constraint in a script file. For example, you could start with CREATE TABLE products ( product_no integer NULL, name text NULL, price numeric NULL ); and then insert the NOT key word where desired. In most database designs the majority of columns should be marked not null. Unique constraints ensure that the data contained in a column or a group of columns is unique with respect to all the rows in the table. The syntax is CREATE TABLE products ( product_no integer UNIQUE, name text, price numeric ); when written as a column constraint, and CREATE TABLE products ( product_no integer, name text, price numeric, UNIQUE (product_no) ); when written as a table constraint. If a unique constraint refers to a group of columns, the columns are listed separated by commas: CREATE TABLE example ( a integer, b integer, c integer, UNIQUE (a, c) ); It is also possible to assign names to unique constraints: CREATE TABLE products ( product_no integer CONSTRAINT must_be_different UNIQUE, name text, price numeric ); In general, a unique constraint is violated when there are (at least) two rows in the table where the values of each of the corresponding columns that are part of the constraint are equal. However, null values are not considered equal in this consideration. That means, in the presence of a multicolumn unique constraint it is possible to store an unlimited number of rows that contain a null value in at least one of the constrained columns. This behavior conforms to the SQL standard, but we have heard that other SQL databases may not follow this rule. So be careful when developing applications that are intended to be portable. Technically, a primary key constraint is simply a combination of a unique constraint and a not-null constraint. So, the following two table definitions accept the same data: CREATE TABLE products ( product_no integer UNIQUE NOT NULL, name text, price numeric ); CREATE TABLE products ( product_no integer PRIMARY KEY, name text, price numeric ); Primary keys can also constrain more than one column; the syntax is similar to unique constraints: CREATE TABLE example ( a integer, b integer, c integer, PRIMARY KEY (a, c) ); A primary key indicates that a column or group of columns can be used as a unique identifier for rows in the table. (This is a direct consequence of the definition of a primary key. Note that a unique constraint does not, in fact, provide a unique identifier because it does not exclude null values.) This is useful both for documentation purposes and for client applications. For example, a GUI application that allows modifying row values probably needs to know the primary key of a table to be able to identify rows uniquely. A table can have at most one primary key (while it can have many unique and not-null constraints). Relational database theory dictates that every table must have a primary key. This rule is not enforced by PostgreSQL, but it is usually best to follow it. A foreign key constraint specifies that the values in a column (or a group of columns) must match the values in some other column. We say this maintains the referential integrity between two related tables. Say you have the product table that we have used several times already: CREATE TABLE products ( product_no integer PRIMARY KEY, name text, price numeric ); Let's also assume you have a table storing orders of those products. We want to ensure that the orders table only contains orders of products that actually exist. So we define a foreign key constraint in the orders table that references the products table: CREATE TABLE orders ( order_id integer PRIMARY KEY, product_no integer REFERENCES products (product_no), quantity integer ); Now it is impossible to create orders with product_no entries that do not appear in the products table. We say that in this situation the orders table is the referencing table and the products table is the referenced table. Similarly, there are referencing and referenced columns. You can also shorten the above command to CREATE TABLE orders ( order_id integer PRIMARY KEY, product_no integer REFERENCES products, quantity integer ); because in absence of a column list the primary key of the referenced table is used as referenced column. A foreign key can also constrain and reference a group of columns. As usual, it then needs to be written in table constraint form. Here is a contrived syntax example: CREATE TABLE t1 ( a integer PRIMARY KEY, b integer, c integer, FOREIGN KEY (b, c) REFERENCES other_table (c1, c2) ); Of course, the number and type of constrained columns needs to match the number and type of referenced columns. A table can contain more than one foreign key constraint. This is used to implement many-to-many relationships between tables. Say you have tables about products and orders, but now you want to allow one order to contain possibly many products (which the structure above did not allow). You could use this table structure: CREATE TABLE products ( product_no integer PRIMARY KEY, name text, price numeric ); CREATE TABLE orders ( order_id integer PRIMARY KEY, shipping_address text, ... ); CREATE TABLE order_items ( product_no integer REFERENCES products, order_id integer REFERENCES orders, quantity integer, PRIMARY KEY (product_no, order_id) ); Note also that the primary key overlaps with the foreign keys in the last table. We know that the foreign keys disallow creation of orders that don't relate to any products. But what if a product is removed after an order is created that references it? SQL allows you to specify that as well. Intuitively, we have a few options: Disallow deleting a referenced product Delete the orders as well Something else? To illustrate this, let's implement the following policy on the many-to-many relationship example above: When someone wants to remove a product that is still referenced by an order (via order_items), we disallow it. If someone removes an order, the order items are removed as well. CREATE TABLE products ( product_no integer PRIMARY KEY, name text, price numeric ); CREATE TABLE orders ( order_id integer PRIMARY KEY, shipping_address text, ... ); CREATE TABLE order_items ( product_no integer REFERENCES products ON DELETE RESTRICT, order_id integer REFERENCES orders ON DELETE CASCADE, quantity integer, PRIMARY KEY (product_no, order_id) ); Restricting and cascading deletes are the two most common options. RESTRICT can also be written as NO ACTON and it's also the default if you don't specify anything. There are two other options for what should happen with the foreign key columns when a primary key is deleted: SET NULL and SET DEFAULT. Note that these do not excuse you from observing any constraints. For example, if an action specifies SET DEFAULT but the default value would not satisfy the foreign key, the deletion of the primary key wil fail. Analogous to ON DELETE there is also ON UPDATE which is invoked when a primary key is changed (updated). The possible actions are the same. More information about updating and deleting data is in . Finally, we should mention that a foreign key must reference columns that are either a primary key or form a unique constraint. If the foreign key references a unique constraint, there are some additional possibilities regarding how null values are matched. These are explained in the CREATE TABLE entry in &cite-reference;. This section needs to be rethought. Some of the information should go into the following chapters. Let's create two tables. The capitals table contains state capitals which are also cities. Naturally, the capitals table should inherit from cities. CREATE TABLE cities ( name text, population float, altitude int -- (in ft) ); CREATE TABLE capitals ( state char(2) ) INHERITS (cities); In this case, a row of capitals inherits all attributes (name, population, and altitude) from its parent, cities. The type of the attribute name is text, a native PostgreSQL type for variable length ASCII strings. The type of the attribute population is float, a native PostgreSQL type for double precision floating-point numbers. State capitals have an extra attribute, state, that shows their state. In PostgreSQL, a table can inherit from zero or more other tables, and a query can reference either all rows of a table or all rows of a table plus all of its descendants. The inheritance hierarchy is actually a directed acyclic graph. For example, the following query finds the names of all cities, including state capitals, that are located at an altitude over 500ft: SELECT name, altitude FROM cities WHERE altitude > 500; which returns: name | altitude -----------+---------- Las Vegas | 2174 Mariposa | 1953 Madison | 845 On the other hand, the following query finds all the cities that are not state capitals and are situated at an altitude over 500ft: SELECT name, altitude FROM ONLY cities WHERE altitude > 500; name | altitude -----------+---------- Las Vegas | 2174 Mariposa | 1953 Here the ONLY before cities indicates that the query should be run over only cities and not tables below cities in the inheritance hierarchy. Many of the commands that we have already discussed -- SELECT, UPDATE and DELETE -- support this ONLY notation. In some cases you may wish to know which table a particular tuple originated from. There is a system column called TABLEOID in each table which can tell you the originating table: SELECT c.tableoid, c.name, c.altitude FROM cities c WHERE c.altitude > 500; which returns: tableoid | name | altitude ----------+-----------+---------- 139793 | Las Vegas | 2174 139793 | Mariposa | 1953 139798 | Madison | 845 (If you try to reproduce this example, you will probably get different numeric OIDs.) By doing a join with pg_class you can see the actual table names: SELECT p.relname, c.name, c.altitude FROM cities c, pg_class p WHERE c.altitude > 500 and c.tableoid = p.oid; which returns: relname | name | altitude ----------+-----------+---------- cities | Las Vegas | 2174 cities | Mariposa | 1953 capitals | Madison | 845 In previous versions of PostgreSQL, the default was not to get access to child tables. This was found to be error prone and is also in violation of SQL99. Under the old syntax, to get the sub-tables you append * to the table name. For example SELECT * from cities*; You can still explicitly specify scanning child tables by appending *, as well as explicitly specify not scanning child tables by writing ONLY. But beginning in version 7.1, the default behavior for an undecorated table name is to scan its child tables too, whereas before the default was not to do so. To get the old default behavior, set the configuration option SQL_Inheritance to off, e.g., SET SQL_Inheritance TO OFF; or add a line in your postgresql.conf file. A limitation of the inheritance feature is that indexes (including unique constraints) and foreign key constraints only apply to single tables, not to their inheritance children. Thus, in the above example, specifying that another table's column REFERENCES cities(name) would allow the other table to contain city names but not capital names. This deficiency will probably be fixed in some future release. When you create a table and you realize that you made a mistake, then you can drop the table and create it again. But this is not a convenient option if the table is already filled with data, or if the table is referenced by other database objects (for instance a foreign key constraint). Therefore PostgreSQL provides a family of commands to make modifications on existing tables. You can Add columns, Add constraints, Remove constraints, Change default values, Rename a column, Rename the table. In the current implementation you cannot Remove a column, Change the data type of a column. These may be possible in a future release. OK, now explain how to do this. There's currently so much activity on ALTER TABLE that I'm holding off a bit. to be filled in Tables are the central objects in a relational database structure, because they hold your data. But they are not the only objects that exist in a database. Many other kinds of objects can be created to make the use and management of the data more efficient or convenient. They are not discussed in this chapter, but we give you a list here so that you are aware of what is possible. Views Functions, operators, data types, domains Triggers and rewrite rules When you create complex database structures involving many tables with foreign key constraints, views, triggers, functions, etc. you will implicitly create a net of dependencies between the objects. For instance, a table with a foreign key constraint depends on the table it references. To ensure the integrity of the entire database structure, PostgreSQL makes sure that you cannot drop objects that other objects still depend on. For example, attempting to drop the products table we had considered in , with the orders table depending on it, would result in an error message such as this: DROP TABLE products; NOTICE: constraint $1 on table orders depends on table products ERROR: Cannot drop table products because other objects depend on it Use DROP ... CASCADE to drop the dependent objects too The error message contains a useful hint: If you don't want to bother deleting all the dependent objects individually, you can run DROP TABLE products CASCADE; and all the dependent objects will be removed. Actually, this doesn't remove the orders table, it only removes the foreign key constraint. All drop commands in PostgreSQL support specifying CASCADE. Of course, the nature of the possible dependencies varies with the type of the object. You can also write RESTRICT instead of CASCADE to get the default behavior which is to restrict drops of objects that other objects depend on. According to the SQL standard, specifying either RESTRICT or CASCADE is required. No database system actually implements it that way, but the defaults might be different.