On Thu, Jul 29, 2021 at 02:28:55PM -0400, Bruce Momjian wrote: > On Thu, Jul 29, 2021 at 10:00:12AM -0700, David G. Johnston wrote: > > I'm warming up to "should" but maybe add a "why" such as "the old versions > > are > > considered unsupported on the newer server". > > > > I dislike "usually via operating system commands", just offload this to the > > extension, i.e., "must be installed in the new cluster via installation > > procedures specific to, and documented by, each extension (for contrib it is > > usually enough to ensure the -contrib package was chosen to be installed > > along > > with the -server package for your operating system.)" > > > > I would simplify the first two sentences to just: > > > > If the old cluster used extensions those same extensions must be installed > > in > > the new cluster via installation procedures specific to, and documented by, > > each extension. For contrib extensions it is usually enough to install the > > -contrib package via the same method you used to install the PostgreSQL > > server.
Oh, and you can't use the same installation procedures as when you installed the extension because that probably included CREATE EXTENSION. This really highlights why this is tricky to explain --- we need the binaries, but not the SQL that goes with it. -- Bruce Momjian <br...@momjian.us> https://momjian.us EDB https://enterprisedb.com If only the physical world exists, free will is an illusion.
