Exception handling v16

By default, any error occurring in an SPL program stops the program from executing. You can trap errors and recover from them by using a BEGIN block with an EXCEPTION section.

Syntax

The syntax is an extension of the normal syntax for a BEGIN block:

[ DECLARE
    <declarations> ]
  BEGIN
    <statements>
  EXCEPTION
    WHEN <condition> [ OR <condition> ]... THEN
      <handler_statements>
  [ WHEN <condition> [ OR <condition> ]... THEN
      <handler_statements> ]...
END;

Error handling process

• If no error occurs, this form of block executes all the statements, and then control passes to the next statement after END.
• If an error occurs in the statements, further processing of the statements is abandoned, and control passes to the EXCEPTION list. The list is searched for the first condition matching the error.
• If a match is found, the corresponding handler_statements are executed, and then control passes to the next statement after END.
• If no match is found, the error propagates out as though the EXCEPTION clause wasn't there. You can catch the error by an enclosing block with EXCEPTION. If there is no enclosing block, it aborts the subprogram.

The special condition named OTHERS matches every error type. Condition names aren't case sensitive.

If a new error occurs in the selected handler_statements, this EXCEPTION clause can't catch it, but it is propagated out. A surrounding EXCEPTION clause might catch it.

Performance implications

Heavy use of the EXCEPTION clause can have visible performance consequences. An EXCEPTION clause establishes a sub-transaction internally in the database server. If the code in the EXCEPTION clause completes without an error, the implicit sub-transaction commits. If an error occurs, it rolls back.

This process has some overhead. If the code protected by the EXCEPTION clause modifies the database, the transaction acquires another transaction ID (XID), increasing the overhead considerably. XID consumption is one factor determining how frequently each table in the database must be subject to VACUUM, so a very high rate of XID consumption leads to more VACUUM activity. Also, when any individual session has more than 64 XIDs assigned simultaneously, some tuple visibility checks incur more overhead.

List of condition names to use

• On this page
• Syntax
• Error handling process
• Performance implications
• List of condition names to use