On Mon, Feb 27, 2023 at 12:50 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
> > On Mon, Feb 27, 2023 at 11:20 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> >> Well, that's a user error not pg_dump's fault. Particularly so for hash
> >> partitioning, where there is no defensible reason to make the partitions
> >> semantically different.
>
> > I am still of the opinion that you're going down a dangerous path of
> > redefining pg_dump's mission from "dump and restore the database, as
> > it actually exists" to "dump and restore the database, unless the user
> > did something that I think is silly".
>
> Let's not attack straw men, shall we? I'm defining pg_dump's mission
> as "dump and restore the database successfully". Failure to restore
> does not help anyone, especially if they are in a disaster recovery
> situation where it's not possible to re-take the dump. It's not like
> there's no precedent for having pg_dump tweak things to ensure a
> successful restore.
Sure, but I was responding to your assertion that there's no case in
which --load-via-partition-root could cause a restore failure. I'm not
sure that's accurate. The fact that the case is something that nobody
is especially likely to do doesn't mean it doesn't exist, and ISTM we
have had more than a few pg_dump bug reports that come down to someone
having done something which we didn't foresee.
--
Robert Haas
EDB: http://www.enterprisedb.com