On Mon, Apr 19, 2021 at 12:37 PM Mark Dilger <mark.dil...@enterprisedb.com> wrote: > > OK, so let's fix it. If amcheck is going to stay in contrib then ISTM > > pg_amcheck should move there. I can organize that if there's agreement. > > Or else let's move amcheck as Alvaro suggests. > > Ah, no. I wrote pg_amcheck in contrib originally, and moved it to src/bin as > requested during the v14 development cycle.
Yeah, I am not that excited about moving this again. I realize it was never committed anywhere else, but it was moved at least one during development. And I don't see that moving it to contrib really fixes anything anyway here, except perhaps conceptually. Maybe inventing src/extensions is the right idea, but there's no real need to do that at this point in the release cycle, and it doesn't actually fix anything either. The only thing that's really needed here is to either (a) teach the test script to install amcheck as a separate step or (b) teach pg_amcheck to install amcheck in a user-specified schema. If we do that, AIUI, this issue is fixed regardless of whether we move any source code around, and if we don't do that, AIUI, this issue is not fixed no matter how much source code we move. -- Robert Haas EDB: http://www.enterprisedb.com
