> On Jun 10, 2024, at 16:01, Michael Paquier <[email protected]> wrote: > > External Email > > From: Michael Paquier <[email protected]> > Subject: Re: Proposal to add page headers to SLRU pages > Date: June 10, 2024 at 16:01:50 GMT+8 > To: Bertrand Drouvot <[email protected]> > Cc: "Li, Yong" <[email protected]>, Jeff Davis <[email protected]>, Aleksander > Alekseev <[email protected]>, PostgreSQL Hackers > <[email protected]>, "Bagga, Rishu" <[email protected]>, > Robert Haas <[email protected]>, Andrey Borodin <[email protected]>, > "Shyrabokau, Anton" <[email protected]> > > > On Mon, Jun 10, 2024 at 07:19:56AM +0000, Bertrand Drouvot wrote: >> On Tue, Mar 19, 2024 at 06:48:33AM +0000, Li, Yong wrote: >>> Unfortunately, the test requires a setup of two different versions of PG. I >>> am not >>> aware of an existing test infrastructure which can run automated tests >>> using two >>> PGs. I did manually verify the output of pg_upgrade. >> >> I think there is something in t/002_pg_upgrade.pl (see >> src/bin/pg_upgrade/TESTING), >> that could be used to run automated tests using an old and a current version. > > Cluster.pm relies on install_path for stuff, where it is possible to > create tests with multiple nodes pointing to different installation > paths. This allows mixing nodes with different build options, or just > different major versions like pg_upgrade's perl tests. > — > Michael > >
Thanks for pointing this out. Here is what I have tried: 1. Manually build and install PostgreSQL from the latest source code. 2. Following the instructions from src/bin/pg_upgrade to manually dump the regression database. 3. Apply the patch to the latest code, and build from the source. 4. Run “make check” by following the instructions from src/bin/pg_upgrade and setting up the olddump and oldinstall to point to the “old” installation used in step 2. All tests pass. Yong
