On Fri, Apr 17, 2026, at 6:48 PM, Tom Lane wrote:
>
> I think running a test promptly after a new patch submission is
> useful, even for small patches. I agree that the periodic re-tests
> for bit-rot could be scaled back a lot.
>
That's my opinion too. This is particularly important for first-time
contributors that may not know about the Postgres development process. For
regular contributors, I expect that they have a CI setup and submit a new
version only after the patch passes CI in its own repository.
I'm not sure about restricting the CI runs to small patches. Although it is a
minority, there are small patches that has a big potential to break things.
Maybe an alternative to small and/or high-frequency patches is to not run them
automatically but have a mechanism to trigger them manually once detected. The
author or even one of the reviewers can trigger it.
--
Euler Taveira
EDB https://www.enterprisedb.com/
