Re: test/isolation/expected/stats_1.out broken for me

Hi,

On 2022-04-07 18:31:35 -0400, Tom Lane wrote:
> Andres Freund  writes:
> > On 2022-04-07 09:57:09 -0700, Andres Freund wrote:
> >> Yea :(. I tested debug_discard_caches, but not -DRELCACHE_FORCE_RELEASE
> >> -DCATCACHE_FORCE_RELEASE.
> >>
> >> Not quite sure what to do about it - it's intentionally trying to test the
> >> case of no invalidations being processed, as that's an annoying edge case with
> >> functions.  Perhaps wrapping the function call of the "already dropped"
> >> function in another function that catches the error would do the trick? It'd
> >> be more easily silently broken, but still be better than not having the test.
>
> > Anybody got a better idea?
>
> Maybe if the wrapper function checks for exactly the two expected
> behaviors, it'd be robust enough?

Seems to work. If I break the code it's trying to test, it still fails... Of
course only when none of debug_discard_caches, RELCACHE_FORCE_RELEASE,
CATCACHE_FORCE_RELEASE are used, but that seems unavoidable / harmless. Let's
see what the buildfarm thinks.

Greetings,

Andres Freund