Re: Minimal logical decoding on standbys

Hi,

On 12/13/22 5:37 PM, Drouvot, Bertrand wrote:
> Hi,
>
> On 12/13/22 2:50 PM, Robert Haas wrote:
>> On Tue, Dec 13, 2022 at 5:49 AM Drouvot, Bertrand
>
>> It seems kind of unfortunate to have to add payload to a whole bevy of
>> record types for this feature. I think it's worth it, both because the
>> feature is extremely important,
>
> Agree and I don't think that there is other option than adding some payload in some WAL records (at the very beginning the proposal was to periodically log a new record
> that announces the current catalog xmin horizon).
>
>> and also because there aren't any
>> record types that fall into this category that are going to be emitted
>> so frequently as to make it a performance problem.
>
> +1
>
> If no objections from your side, I'll submit a patch proposal by tomorrow, which:
>
> - get rid of IndexIsAccessibleInLogicalDecoding
> - let RelationIsAccessibleInLogicalDecoding deals with the index case
> - takes care of the padding where the new bool is added
> - convert this new bool to a flag for the xl_heap_visible case (adding a new bit to the already existing flag)
> - Add my proposed wording above to the commit message
> - Add your proposed wording above in a README

Please find attached v31 with the changes mentioned above (except that I put your wording into the commit message instead of a README: I think it helps to make
clear what the "design" for the patch series is).

Regards,

--
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com