On Wed, 7 Jan 2026 at 09:37, Amit Langote <amitlangote09@gmail.com> wrote:
>
> Yes, I think the approach in your patch seems better for the reason
> you mentioned, at least for back-patching sanity.
>
> I intended all of these relid sets to account for prunable RELATION RTEs only.
Yes, I think that makes sense.
> Thanks Tender and Bernice for the additional analysis. I prefer Dean's
> fix-the-executor approach for back-patching. Bernice, are there other
> related issues you're aware of beyond this rowmark bug? Want to make
> sure Dean's patch covers them too.
It looks to me as though either approach would work, so I'm happy for
you to decide which approach fits best with your design.
> Thanks for the patch! Do you intend to commit and back-patch this
> yourself, or would you like me to handle it?
It's your code, and you're more familiar with it than me, so I'm happy
to leave it to you :-)
Regards,
Dean
