Chapter 52 – Module Binding Style


You are reading a digital copy of SQL-99 Complete, Really, a book that documents the SQL-99 standard.

The book and the standard does not reflect the features of CrateDB, we are just publishing it as a service for the community and for reference purposes.

On the one hand, CrateDB does not implement the SQL-99 standard thoroughly, on the other hand, it extends the standard for implementing cluster features and others.

For more information specific to CrateDB, check out the CrateDB Reference documentation.

SQL DBMSs communicate with SQL applications through a common programming language interface that is invoked through one of the SQL Standard-defined binding styles, or interface options. There are three main approaches to writing complete programs with SQL:

  1. With embedded SQL, you can put SQL statements directly into host programs. We described this binding style earlier.

  2. With SQL/CLI, you can call a SQL DBMS’s library from a host program. This binding style was the subject of the last several chapters.

  3. With the Module language, you can dispense with host programs and write entire Modules in SQL. You’ll still have to call these Modules from one of the standard host languages though. The Module language binding style is the subject of this chapter.

Because we believe that SQL/CLI will quickly become the SQL interface of choice, this chapter omits large amounts of detail, in favour of providing the necessary detail in our chapters on SQL/CLI.


The SQL Standard actually describes three kinds of SQL Module, each of which has certain characteristics and contains various kinds of Module Objects (principally, routines). An SQL-client Module contains only externally-invoked procedures, an SQL-session Module contains only SQL statements prepared in that SQL-session and an SQL-server Module – the SQL/PSM type – is a Schema Object that contains only SQL-invoked routines.

Table of Contents

SQL-client Modules

SQL-client Modules are programming modules that contain externally-invoked procedures – that is, SQL procedures that are invoked by a host language. Generally speaking, this methodology is really the basic SQL binding style: all of the binding styles – direct SQL, embedded SQL, CLI or SQL-client Modules – at least conceptually, involve a program module. Before SQL/PSM came on the scene, SQL “module language” was the only way to identify SQL Modules. But now, with the advent of SQL/PSM and (far better) of SQL/CLI, we believe this method of utilizing SQL will rapidly become obsolete and so we give only a brief description here.

An SQL-client Module is an Object – a programming module – that you define with SQL’s module language: a subset of SQL that allows you to write database routines in pure SQL. It contains SQL statements that will operate on your SQL-data, and you link it (in some implementation-defined way) with modules of code in one or more of the Standard’s host languages. The database routines are called externally-invoked procedures because they are invoked by the host language program to which you link the Module they belong to.

MODULE Statement

The MODULE statement defines an SQL-client Module. The required syntax for the MODULE statement is:

<SQL-client Module definition> ::=
MODULE [ <SQL-client Module name> ]
 NAMES ARE <Character set name>
 <Module authorization clause>
 [ PATH <Schema name> {,<Schema name>}... ]
 [ TRANSFORM GROUP {<group name> | {<group name> FOR TYPE <UDT name>} , ...} ]
 [ DECLARE TABLE statement(s) ]
 <Module contents>...

      <Module authorization clause> ::=
      SCHEMA <Schema name> |
      AUTHORIZATION <AuthorizationID> |
      SCHEMA <Schema name> AUTHORIZATION <AuthorizationID>

      <Module contents> ::=
      DECLARE CURSOR statement(s) |
      PROCEDURE statement(s)

An SQL-client Module doesn’t have to be named (unless the Module’s LANGUAGE clause specifies ADA, in which case you must give the Module a valid Ada library unit name); your SQL-environment can contain multiple unnamed SQL- client Modules. If you do name an SQL-client Module, though, you must give it a unique name (for all SQL-client Modules) in your SQL-environment. A <SQL-client Module name> is a <regular identifier> or a <delimited identifier>.

The optional NAMES ARE clause provides the name of the Module’s default Character set: the Character set that your DBMS will use for any character strings in the Module that don’t include an explicit Character set specification. If you omit this clause, the Module’s default Character set is chosen by your DBMS, and must contain at least every <SQL language character>.

The LANGUAGE clause provides the name of the host language that will invoke the routines this Module contains.

The Module authorization clause provides either the Module’s default Schema, the Module’s <AuthorizationID> or both (at least one must be included).

  • SCHEMA <Schema name> provides an explicit <Schema name> – this will be the default <Schema name> qualifier for any Objects referred to in the Module without explicit qualifiers. If you omit this clause, the default <Schema name> defaults to the value in the AUTHORIZATION clause.

  • AUTHORIZATION <AuthorizationID> provides an explicit <AuthorizationID> to be the owner of the Module – this will be the <AuthorizationID> whose Privileges will be checked when the Module’s SQL statements are executed. If you omit this clause, your DBMS will treat the SQL-session <AuthorizationID> as the Module’s owner at runtime.

The optional PATH clause provides a list of <Schema name>s that will be used as the Module’s default path – that is, the qualifying <Schema name>s that will be used for any unqualified <Routine name>s in this Module. You can name zero or more Schemas in this clause (your DBMS will pick the one that matches the unqualified routine best at runtime); each Schema in the list must belong to the same Catalog that this Module’s default Schema belongs to. If you omit this clause, it will default to a list of Schemas, containing at least this Module’s default Schema, chosen by your DBMS.

The optional TRANSFORM GROUP clause provides a <group name> for each UDT parameter that has no locator; see our chapter on UDTs.

You can declare zero or more temporary Tables for the Module. Each will be visible only to the Module you declare them in.

The <Module contents> clause is the meat of the Module – it contains the SQL statements that do the work you need done on your SQL-data. You can declare zero or more Cursors here (see our chapter on embedded SQL), as well as one or more externally-invoked procedures (with the PROCEDURE statement, see below).

If you want to restrict your code to Core SQL, don’t use the NAMES ARE clause, the PATH clause, the TRANSFORM GROUP clause or any DECLARE TABLE statements in a MODULE statement.


The PROCEDURE statement defines an externally-invoked procedure. The required syntax for the PROCEDURE statement is:

PROCEDURE <procedure name>
{(<host parameter declaration> [ {,<host parameter declaration>}... ] ) |
<host parameter declaration>...};
SQL procedure statement;

   <host parameter declaration> ::=
   :<host parameter name> <data type> [ AS LOCATOR ] |

An externally-invoked procedure is an SQL procedure that is called from a host language program. It belongs to an SQL-client Module, and must have a <procedure name> that is unique (for all procedures) within that Module. A <procedure name> is a <regular identifier> or a <delimited identifier> and should conform to the host language you’ll be calling it from.

Each procedure has to contain a list of one or more parameter declarations, terminated with a semicolon – SQLSTATE is always mandatory. Your list should be enclosed in parentheses, with each parameter declaration separated from the next by a comma (though our syntax diagram shows that both the parentheses and the commas are currently optional, this is a deprecated feature in the Standard, so avoid it). Other than SQLSTATE, each parameter has a name (preceded by a colon), a <data type> and – if it is a BLOB, CLOB, NCLOB, UDT or ARRAY – an optional AS LOCATOR indicator. A call of an externally-invoked procedure has to supply the same number of arguments as the parameter declarations in the procedure.

Each procedure has to contain exactly one SQL procedure statement, terminated with a semicolon. This is the SQL statement that gets executed when the procedure is called. An SQL procedure statement is any executable SQL statement – this includes all the SQL-Schema statements, the SQL-data statements, the SQL-control statements, the SQL-transaction statements, the SQL-Connection statements, the SQL-session statements and the SQL diagnostics statement.

If you calling the procedure from:

Ada – then use only these <data type>s in your <host parameter declaration>s: CHARACTER, BIT, SMALLINT, INTEGER, REAL, DOUBLE PRECISION, declare SQLSTATE's base type as SQL_STANDARD.SQLSTATE_TYPE, and identify the procedure by its <procedure name>, as if it was declared within an Ada library unit specification that has a name equal to the name of the SQL-client Module that contains the procedure.

C – then use only these <data type>s in your <host parameter declaration>s: CHARACTER, CHARACTER VARYING, BIT, INTEGER, SMALLINT, REAL, DOUBLE PRECISION, and declare SQLSTATE as a C char with length 6.

COBOL – then use only these <data type>s in your <host parameter declaration>s: CHARACTER, BIT, NUMERIC, INTEGER, SMALLINT, and declare SQLSTATE as a COBOL PICTURE X(5).

Fortran – then use only these <data type>s in your <host parameter declaration>s: CHARACTER, BIT, INTEGER, REAL, DOUBLE PRECISION, and declare SQLSTATE as a Fortran CHARACTER with length 5.

MUMPS – then use only these <data type>s in your <host parameter declaration>s: CHARACTER VARYING, INTEGER, DECIMAL, REAL, and declare SQLSTATE as a MUMPS character with maximum length greater than or equal to 5.

Pascal – then use only these <data type>s in your <host parameter declaration>s: CHARACTER, BIT, INTEGER, REAL, and declare SQLSTATE as a Pascal PACKED ARRAY [1..5] OF CHAR.

PL/I – then use only these <data type>s in your <host parameter declaration>s: CHARACTER, CHARACTER VARYING, BIT, BIT VARYING, DECIMAL, INTEGER, SMALLINT, FLOAT, and declare SQLSTATE as a PL/I CHARACTER(5).

For a list of the correspondences between SQL <data type>s and host data types, see “Host Variables” in our chapter on embedded SQL.

Here’s an example of an SQL-client Module:

MODULE module_1
   SCHEMA catalog_1.schema_1 AUTHORIZATION bob
   PROCEDURE proc_1 (
      :var_1 INTEGER, :var_2 CHARACTER VARYING(25));
      INSERT INTO Table_1 (column_1, column_2)
         VALUES (:var_1, :var_2);
   PROCEDURE proc_2 (
      :var_1 INTEGER);
      UPDATE Table_2 SET
         column_3 = column_3*:var_1 WHERE column_3 IN
           (SELECT column_1 FROM Table_1;