On Wed, Jun 04, 2025 at 09:15:08AM +0900, Michael Paquier wrote:
> On Tue, Jun 03, 2025 at 03:34:16PM -0400, Andres Freund wrote:
>> I'm somewhat doubtful this is is the right direction. Tests that require
>> injection points before consistency also can't wait for injection points using
>> the SQL interface or such, so most of the stuff has to be written in C
>> anyway. And if so, you also can attach to injection points in the relevant
>> shared_preload_libraries entry.
>
> Hmm. I'm wondering about an alternate approach here: a postmaster GUC
> in injection_points that can take in input a list of
> name/library/function where the module would load them when
> initializing. That's a bit artistic, perhaps, still it would work
> without having to worry about the flush and reload steps.
[A couple of weeks later]
I am still not sure what's the right course of action here, so I'd
agree to just reject the patch for now, but keep it in mind once we
have more cases where this could be useful. The case I've found is
perhaps not enough. Updating the CF app to reflect on this feedback
now, thanks all for the input.
--
Michael
