Andrew Dunstan <andrew@dunslane.net> writes:
> On 2024-02-28 We 17:36, Tom Lane wrote:
>> Hmm ... pg_dump has never used any DO blocks in its output, and
>> I'm not sure it's a great idea to start. Seems like the kind of
>> decision we could regret down the road, given the possible need
>> for pg_restore to parse the output.
> This is only for pg_dumpall. pg_restore shouldn't care.
Yeah, fair point. I've long thought that we should get pg_dumpall
to emit something more structured than "big SQL script", but I'm
not holding my breath for that to happen.
Still, I'm a bit leery of the idea. Isn't plpgsql supposed to be
optional/droppable? I guess we could tell people who don't want it
that they can't drop it until after restoring their old data,
but still ...
regards, tom lane