do we have to alter the numeric columns in our table schema like its mentioned in the documentation?

[Rammurthy5]

In the documentation, its mentioned the below. if i dont create the default 4 tables and create different tables and schema, do i still have to alter the numeric columns like this? how does it help? can someone please clarify? 🙇🏻

alter the numeric columns, setting LOG_OLD_VALUE equal to true. This will make these numeric fields conflict-free delta-apply columns, ensuring that the value replicated is the delta of the committed changes (the old value plus or minus any new value) to a given record:

ALTER TABLE pgbench_accounts ALTER COLUMN abalance SET (LOG_OLD_VALUE=true);
ALTER TABLE pgbench_branches ALTER COLUMN bbalance SET (LOG_OLD_VALUE=true);
ALTER TABLE pgbench_tellers ALTER COLUMN tbalance SET (LOG_OLD_VALUE=true);

[luss]

Hi Ram, You only want to alter numeric columns that are running sums. Such as the monetary sum on and order header table that represents the sum of the detail lines.

…

--Denis

On Mon, Sep 11, 2023 at 10:00 AM Ram ***@***.***> wrote: In the documentation, its mentioned the below. if i dont create the default 4 tables and create different tables and schema, do i still have to alter the numeric columns like this? how does it help? can someone please clarify? 🙇🏻 alter the numeric columns, setting LOG_OLD_VALUE equal to true. This will make these numeric fields conflict-free delta-apply columns, ensuring that the value replicated is the delta of the committed changes (the old value plus or minus any new value) to a given record: ALTER TABLE pgbench_accounts ALTER COLUMN abalance SET (LOG_OLD_VALUE=true); ALTER TABLE pgbench_branches ALTER COLUMN bbalance SET (LOG_OLD_VALUE=true); ALTER TABLE pgbench_tellers ALTER COLUMN tbalance SET (LOG_OLD_VALUE=true); — Reply to this email directly, view it on GitHub <#28>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMWOHT7MLIGYLOFD6N6RO3XZ4KOVANCNFSM6AAAAAA4TMHIWM> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>