Cursors

Dynamic SQL

Chapter 5: Data Structures

All of the Server Express SQL preprocessors make use of the SQL Communications Area (SQLCA) and the SQL Descriptor Area (SQLDA) data structures.

5.1 SQL Communications Area

After each embedded SQL statement is executed, error and status information is returned in the SQL Communications Area (SQLCA).

The SQLCA data structure is shown below:

01 SQLCA. 
  03 SQLCAID     PIC X(8) VALUE "SQLCA".
  03 SQLCABC     PIC S9(9) COMP-5 VALUE 136.
  03 SQLCODE     PIC S9(9) COMP-5 VALUE 0.
  03 SQLERRM.
    49 SQLERRML  PIC S9(4) COMP-5.
    49 SQLERRMC  PIC X(70).
  03 SQLERRP     PIC X(8).
  03 SQLERRD     PIC S9(9) COMP-5 OCCURS 6 VALUE 0.
  03 SQLWARN.
    05 SQLWARN0  PIC X.
    05 SQLWARN1  PIC X.
    05 SQLWARN2  PIC X.
    05 SQLWARN3  PIC X.
    05 SQLWARN4  PIC X.
    05 SQLWARN5  PIC X.
    05 SQLWARN6  PIC X.
    05 SQLWARN7  PIC X.
  03 FILLER      PIC X(3).
  03 SQLSTATE    PIC X(5).

Oracle, Sybase and Informix all have different versions of the SQLCA. The SQLCA presented above is for OpenESQL and the DB2 ECM. The Oracle, Sybase and Informix SQLCA's all have a SQLCODE, SQLERRML, SQLERRMC and a SQLWARN. The sizes and the positions of these fields can differ between the precompilers.

The table below describes the contents of the SQLCA data structure:

The SQLCA contains two variables (sqlcode and sqlstate) plus a number of warning flags which are used to indicate whether an error has occurred in the most recently executed SQL statement.

COBSQL
For COBSQL, SQLSTATE is a separate data item. For the currently supported versions of Oracle and Sybase, the SQLCA should be used in preference to the SQLSTATE variable. The SQLSTATE variable will eventually supersede the SQLCA as the preferred method of passing data between the database and the client application, but this is not yet the case.

5.1.1 The SQLCODE Variable

Testing the value of sqlcode is the most common way of determining the success or failure of an embedded SQL statement. The possible values for sqlcode are:

OpenESQL
If you are using OpenESQL, the following error codes are defined for SQLCODE.

COBSQL and DB2
For COBSQL and DB2, it is possible to get other positive values. This means that the SQL statement has executed but produced a warning.

COBSQL

• For details about the range of positive values that SQLCODE can be set to, consult your Oracle, Sybase or Informix Error Messages manual. The above SQLCODES are produced by OpenESQL. The values for SQLCODE and the errors reported by Oracle, Sybase and Informix are all different. Please refer to Oracle, Sybase or Informix Error Messages manuals for more information on the errors returned.

• The value +100 is the ANSI standard for 'data not found'. Oracle can return another value for 'data not found'. To get Oracle to return the value +100 for 'data not found', either of the Oracle precompiler directives MODE=ANSI or END_OF_FETCH=100 must be set. This will affect other aspects of the way the Oracle precompiler handles SQL statements. For more details on the Oracle precompiler MODE or END_OF_FETCH=+100 directives, refer to the Programmer's Guide to the ORACLE Precompilers.

• Even when SQLCODE contains zero, a warning may have been generated. The values of the sqlwarn flags should be checked to determine the type of warning. For Oracle, Sybase and Informix, sqlwarn0 will always be set when the database server has sent a warning back to the application.

5.1.2 The SQLSTATE Variable

The SQLSTATE variable was introduced in the SQL-92 standard and is the recommended mechanism for future applications. It is divided into two components:

• The first two characters are called the class code. Any class code that begins with the letters A through H or the digits 0 through 4 indicates a SQLSTATE value that is defined by the SQL standard or another standard.

• The last three characters are called the subclass code.

A value of "00000" indicates that the previous embedded SQL statement executed successfully.

For specific details of the values returned in SQLSTATE when using Oracle, Sybase or Informix, refer to the relevant Database Error Messages manual.

Full details of SQLSTATE values are given below:

DB2
DB2 Universal Database returns SQL-92 compliant SQLSTATE values. DB2 Version 2.1 does not.

5.1.3 The Warning Flags

Some statements may cause warnings to be generated. To determine the type of warning, your application should examine the contents of the SQLWARN flags.

Each flag returns one of the following values:

• W: The flag has generated a warning.

• blank (space): The flag has not generated a warning.

Each SQLWARN flag has a specific meaning. For more information on the meaning of the SQLWARN flags, refer to the section SQL Communications Area.

5.1.4 The WHENEVER Statement

To check explicitly the value of SQLCODE or SQLSTATE after each embedded SQL statement can involve writing a lot of code. As an alternative, check the status of the SQL statement by using a WHENEVER statement in your application.

The WHENEVER statement is not an executable statement. The WHENEVER statement is a directive to the Compiler to generate automatically the code that handles errors after each executable embedded SQL statement.

The WHENEVER statement allows one of three default actions (CONTINUE, GOTO or PERFORM) to be registered for each of the following conditions:

A WHENEVER statement for a particular condition replaces all previous WHENEVER statements for that condition.

The scope of a WHENEVER statement is related to its physical position in the source program, not its logical position in the run sequence. For example, in the following code if the first SELECT statement does not return anything, paragraph A is performed, not paragraph C:

    EXEC SQL
       WHENEVER NOT FOUND PERFORM A
    END-EXEC.
    PERFORM B.
    EXEC SQL
       SELECT col1 into :host-var1 FROM table1 
         WHERE col2 = :host-var2
    END-EXEC.
 A.
    DISPLAY "First item not found".
 B.
    EXEC SQL
       WHENEVER NOT FOUND PERFORM C.
    END-EXEC.
 C.
    DISPLAY "Second item not found".

5.1.4.1 COBSQL

For Oracle, Sybase and Informix, setting SQLWARN0 to W triggers the SQLWARNING clause.

Oracle
When no data is returned from a SELECT or FETCH statement, the condition NOT FOUND is triggered, regardless of the setting of the Oracle precompiler directive MODE.

Informix
Informix allows you to perform a STOP or a CALL from within a WHENEVER statement. These are additions to the ANSI standard and are documented in the Informix ESQL/COBOL programmers manual.

5.1.5 SQLERRM

The SQLERRM data area is used to pass error messages to the application from the database server. The SQLERRM data area is split into two parts:

• SQLERRML:

  SQLERRML holds the length of the error message

• SQLERRMC:

  SQLERRMC holds the error text.

Within an error routine, the following code can be used to display the SQL error message:

IF (SQLERRML > ZERO) and (SQLERRML < 80)
   DISPLAY 'Error Message: ', SQLERRMC(1:SQLERRML)
ELSE
   DISPLAY 'Error Message: ', SQLERRMC
END-IF.

5.1.6 SQLERRD

The SQLERRD data area is an array of six integer status values.

5.1.6.1 COBSQL

Oracle, Sybase and Informix may set one (or more) of the six values within the SQLERRD array. These indicate how many rows were affected by the SQL statement just executed. For example, SQLERRD(3) holds the total number of rows returned by a SELECT or a series of FETCH statements.

5.1.6.2 DB2

The third element of SQLERRD in the SQLCA, SQLERRD(3), records the number of rows processed for INSERT, UPDATE, DELETE and SELECT INTO statements. For FETCH statements, it records the cumulative sum of rows processed.

SQLERRD(3)
For DB2, SQLERRD(3) contains the following:

• If PREPARE is invoked and successful, an estimate of the number of rows that will be returned.
  
  
• After INSERT, UPDATE and DELETE, the actual number of rows affected.
  
  
• If compound SQL is invoked, a count of the number of successful substatements.
  
  
• If CONNECT is invoked, either 1 if the database can be updated or 2 if the database is read-only.
  
  
• If an error occurs when processing a host array, the last row that was processed successfully.

SQLERRD(4)
For DB2, SQLERRD(4) contains the following:

• If PREPARE is invoked and successful, a relative cost estimate of the resources required to process the statement.
  
  
• If compound SQL is invoked, a count of the number of successful substatements.
  
  
• If CONNECT is invoked, 0 for a one-phase commit from a down-level client; 1 for a one-phase commit; 2 for a one-phase, read-only commit; 3 for a two-phase commit.

SQLERRD(5)
For DB2, SQLERRD(5) contains the following:

• The total number of rows deleted, inserted, or updated as a result of both:
  
  
  • The enforcement of constraints after a successful delete operation.
    
    
  • The processing of triggered SQL statements from activated triggers.
    
    
• If compound SQL is invoked, an accumulation of the number of such rows for all substatements. In some cases, when an error is encountered, this field contains a negative value that is an internal error pointer.
  
  
• If CONNECT is invoked, an authentication type value of 0 for a server authentication; 1 for client authentication; 2 for authentication using DB2 Connect; 3 for DCE security services authentication; 255 for unspecified authentication.

5.2 The SQL Descriptor Area (SQLDA)

The SQLDA is unique to each precompiler. The Oracle SQLDA is not compatible with that used by Sybase, OpenESQL or DB2 and vice versa.

When either the number of parameters to be passed, or their data types, are unknown at compilation time, you can use an SQL Descriptor Area (SQLDA) instead of host variables.

An SQLDA contains descriptive information about each input parameter or output column. It contains the column name, data type, length, and a pointer to the actual data buffer for each input or output parameter. An SQLDA is ordinarily used with parameter markers to specify input values for prepared SQL statements but you can also use an SQLDA with the DESCRIBE statement (or the INTO option of a PREPARE statement) to receive data from a prepared SELECT statement.

Although you cannot use an SQLDA with static SQL statements, you can use a SQLDA with a cursor FETCH statement.

The following table describes the contents of the SQLDA data structure.

5.2.1 COBSQL

For Oracle, Sybase and Informix, the SQLDA is only required if your program uses dynamic SQL.

Oracle, Sybase and Informix do not allow the SQLDA to be included in your program using the following syntax statement:

EXEC SQL
   INCLUDE SQLDA
END-EXEC

For Oracle, Sybase and Informix, the SQLDA must be defined as a standard COBOL copyfile.

5.2.1.1 Oracle

Oracle provides an extra copyfile, ORACA, for use with dynamic SQL. This can be included in your program using the following syntax:

EXEC SQL
   INCLUDE ORACA
END-EXEC

You must set the Oracle precompiler option, ORACA=YES before you can use the ORACA copyfile. For more information on setting Oracle precompiler options, refer to the Programmer's Guide to the Oracle Precompilers.

Oracle does not supply an SQLDA. For a clearer explanation of this and the ORACA copyfile, refer to the Programmer's Guide to the Oracle Precompilers.

5.2.1.2 Sybase

Sybase does not supply an SQLDA copyfile. The Sybase precompiler documentation describes the layout of the SQLDA and how to assign values to the various items within it. The documentation also describes how to get Sybase to convert between COBOL and Sybase data types.

5.2.1.3 Informix

Informix does not supply an SQLDA copyfile. The Informix precompiler documentation describes the layout of the data items that need to be defined to be able to use Dynamic SQL with Informix.

5.2.2 OpenESQL

The SQLDA structure is supplied in the file sqlda.cpy in the source directory under your Server Express base installation directory. You can include it in your COBOL program by adding the following statement to your Data Division:

EXEC SQL
   INCLUDE SQLDA
END-EXEC

The SQLDA data structure is shown below:

01 SQLDA sync.
  05 SQLDAID           PIC X(8) VALUE "SQLDA  ".
  05 SQLDABC           PIC S9(9) COMP-5 value 0.
  05 SQLN              PIC S9(4) COMP-5 value 0.
  05 SQLD              PIC S9(9) COMP-5 value 0.
  05 SQLVAR OCCURS 0 to 1489 TIMES DEPENDING ON SQLN.
    10 SQLTYPE         PIC S9(4) COMP-5.
    10 SQLLEN          PIC S9(4) COMP-5.
    10 SQLDATA         USAGE POINTER.
    10 SQLIND          USAGE POINTER.
    10 SQLNAME.
      15 SQLNAMEL      PIC S9(4) COMP-5.
      15 SQLNAMEC      PIC X(30).

5.2.3 Valid Values for SQLTYPE

Odd-numbered code values indicate that null values are allowed.

In the table below:

(1) - These types can be returned in COBOL by a PREPARE INTO or DESCRIBE statement.

(2) - These types can be set by an application using Dynamic SQL.

(3) - These types are supported for COBOL host variables.

5.2.4 Using the SQLDA

Before an SQLDA structure is used, your application must initialize the following fields:

5.2.4.1 The PREPARE and DESCRIBE Statements

You can use the DESCRIBE statement (or the PREPARE statement with the INTO option) to enter the column name, data type, and other data into the appropriate fields of the SQLDA structure.

Before the statement is executed, the SQLN and SQLDABC fields should be initialized as described above.

After the statement has been executed, the SQLD field will contain the number of parameters in the prepared statement. A SQLVAR record is set up for each of the parameters with the SQLTYPE and SQLLEN fields completed.

If you do not know how big the value of SQLN should be, you can issue a DESCRIBE statement with SQLN set to 1 and SQLD set to 0. No column detail information is moved into the SQLDA structure, but the number of columns in the results set is inserted into SQLD.

5.2.4.2 The FETCH Statement

Before performing a FETCH statement using an SQLDA structure, follow the procedure below:

1. The application must initialize SQLN and SQLDABC as described above.

2. The application must then insert, into the SQLDATA field, the address of each program variable that will receive the data from the corresponding column. (The SQLDATA field is part of SQLVAR).

3. If indicator variables are used, SQLIND must also be set to the corresponding address of the indicator variable.

The data type field (SQLTYPE) and length (SQLLEN) are filled with information from a PREPARE INTO or a DESCRIBE statement. These values can be overwritten by the application prior to a FETCH statement.

5.2.4.3 The OPEN or EXECUTE Statements

To use an SQLDA structure to specify input data to an OPEN or EXECUTE statement, your application must supply the data for the fields of the entire SQLDA structure, including the SQLN, SQLD, SQLDABC, and SQLTYPE, SQLLEN, and SQLDATA fields for each variable.

If the value of the SQLTYPE field is an odd number, the address of the indicator variable must also be supplied in SQLIND.

5.2.4.4 The DESCRIBE Statement

After a PREPARE statement, you can execute a DESCRIBE statement to retrieve information about the data type, length and column name of each column returned by the specified prepared statement. This information is returned in the SQL Descriptor Area (SQLDA):

EXEC SQL
   DESCRIBE stmt1 INTO :sqlda
END-EXEC

If you want to execute a DESCRIBE statement immediately after a PREPARE statement, you can use the INTO option on the PREPARE statement to perform both steps at once:

EXEC SQL
   PREPARE stmt1 INTO :sqlda FROM :stmtbuf
END-EXEC

Cursors

Dynamic SQL