Re: DROP OWNED BY fails to clean out pg_init_privs grants

Daniel Gustafsson <daniel(at)yesql(dot)se> writes:
> On 21 Apr 2024, at 23:08, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> So the meson animals are not running the test that sets up the
>> problematic data.

> I took a look at this, reading code and the linked thread. My gut feeling is
> that Stephen is right in that the underlying bug is these privileges ending up
> in pg_init_privs to begin with. That being said, I wasn't able to fix that in
> a way that doesn't seem like a terrible hack.

Hmm, can't we put the duplicate logic inside recordExtensionInitPriv?
Even if these calls need a different result from others, adding a flag
parameter seems superior to having N copies of the logic.

A bigger problem though is that I think you are addressing the
original complaint from the older thread, which while it's a fine
thing to fix seems orthogonal to the failure we're seeing in the
buildfarm. The buildfarm's problem is not that we're recording
incorrect pg_init_privs entries, it's that when we do create such
entries we're failing to show their dependency on the grantee role
in pg_shdepend. We've missed spotting that so far because it's
so seldom that pg_init_privs entries reference any but built-in
roles (or at least roles that'd likely outlive the extension).

regards, tom lane