When a Pipeline object executes, it commits updates to the destination table according to your specifications in the Data Pipeline painter. You do not need to write any COMMIT statements in your application’s scripts (unless you specified the value None for the Pipeline object’s Commit property).
For instance, both of the Pipeline objects in the order entry application (pipe_sales_extract1 and pipe_sales_extract2) are defined in the Data Pipeline painter to commit all rows. As a result, the Start function (or the Repair function) will pipe every appropriate row and then issue a commit.
You might want instead to define a Pipeline object that periodically issues commits as rows are being piped, such as after every 10 or 100 rows.
A related topic is what happens with committing if your application calls the Cancel function to stop a pipeline that is currently executing. In this case too, the Commit property in the Data Pipeline painter determines what to do, as shown in Table 17-3.
If your Commit value is |
Then Cancel does this |
|---|---|
All |
Rolls back every row that was piped by the current Start function (or Repair function) |
A particular number of rows (such as 1, 10, or 100) |
Commits every row that was piped up to the moment of cancellation |
This is the same commit/rollback behavior that occurs when a pipeline reaches its Max Errors limit (which is also specified in the Data Pipeline painter).
For more information on controlling commits and rollbacks for a Pipeline object, see the PowerBuilder Users Guide.
