Provide proper SQL auditing user/timestamps

IBM does not intend to provide a solution to this request at this time, so it is being closed.
While we agree with the request, we do not currently plan to implement it because the cost of implementing a new version of user would not be worth the benefit.

The CLIENT_USERID and SESSION_USER can be and frequently are different values especially when using connection pooling on the client side. Both values are important and we will not change either definition.

The ON UPDATE part of the row change timestamp syntax semantically includes the initial "update" which happens on an INSERT. This attribute already does what you are requesting.

The syntax for row change timestamp is verbose, but this is what the Db2 family agreed to implement. We do not intend to introduce an alternate syntax variation.

The "Generated Always For Each Row on Update as Row Change" issue does not seem to be an issue - when ran a CREATE TABLE now with such a timestamp, it used the current timestamp, so there was not a need to have it on insert - now if you change the DEFAULT on the column, that is another story.

Here are the statements I ran -

CREATE OR REPLACE TABLE dp2test/table1 (
column1 INTEGER NOT NULL,
column2 VARCHAR(100) ALLOCATE (20),
change_ts GENERATED ALWAYS FOR EACH ROW ON UPDATE AS ROW CHANGE TIMESTAMP NOT NULL
);

INSERT INTO dp2test/table1 (
column1,
column2
)
VALUES
(
25,
'this is varchar'
);

SELECT *
FROM dp2test/table1;

And the result -

25 this is varchar 2019-08-20 08:56:53.569561

So I can't support the addition of automatic timestamp on insert - the other request could perhaps have its own RFE, for easier tracking by IBM and the community.

Vern (CAAC member)