Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> Hmm. Well, as I said, maybe this part of the test isn't worth much
> anyway. Rather than spending time trying to figure out why isn't this
> triggering the WAL overwriting, I compared the coverage report for
> running only the first test to the coverage report of running only the
> second test. It turns out that there's no relevant coverage increase in
> the second test. So I propose just removing that part.
Seems reasonable. We don't need to spend buildfarm cycles forevermore
on a test that's not adding useful coverage.
regards, tom lane