Re: make -C src/test/isolation failure in index-killtuples due to btree_gist

Hi,

On 2025-08-18 11:38:02 -0400, Tom Lane wrote:
> Andres Freund <andres(at)anarazel(dot)de> writes:
> > On 2025-08-18 08:57:13 +0900, Michael Paquier wrote:
> >> The following command fails, because btree_gist is not installed in
> >> the context of the isolation tests:
> >> make -C src/test/isolation/
> >> ...
> >> and it does not strike me as a good idea to begin doing that in the main
> >> isolation test suite, so perhaps the best thing to do here is just move this
> >> test to contrib/btree_gist/?
>
> > No, it makes absolutely no sense to test e.g. hash killtuples support in
> > btree_gist.
>
> I think the complaint is that nothing has been done to ensure that
> these modules have been installed.

Yes, clearly that needs to be fixed. Not sure why I missed that. My point in
my above - perhaps to flippant - response was just that moving it to
btree_gist doesn't make sense.

> You created a new dependency that developers have to work around, rather
> than teaching the build system to handle it. As a comparison point, all of
> the tests in src/test/recovery, src/test/authentication, etc take care to
> install required modules when you say "make check" in those directories.
> You broke that for src/test/isolation, and you should fix it. It shouldn't
> be much harder than setting EXTRA_INSTALL in the Makefile case; I dunno
> about meson.

That would be the easiest fix - but I'm starting to wonder if it shouldn't
just be its own test module, as annoying as the boilerplate for that is.

While the test improved code coverage for the various indexes noticeably, I
did subsequently realize that the new test doesn't end up testing the recovery
path :(. Better than nothing, but having any coverage of those paths might be
worth the boilerplate and the runtime overhead of a test module :/

Greetings,

Andres Freund