On Mon, Apr 01, 2019 at 10:53:27AM +0200, Peter Eisentraut wrote:
> On 2019-03-31 15:22, Jeff Janes wrote:
>> I can't do a same-major-version pg_upgrade across this commit, as the
>> new pg_dump is trying to dump a nonexistent attgenerated column from the
>> old database. Is that acceptable collateral damage? I thought we
>> usually avoided that breakage within the dev branch, but I don't know
>> how we do it.
>
> We don't.
Really? I thought on the contrary that it should work. This reminds
me of commit 59a884e9, which has been discussed here:
https://www.postgresql.org/message-id/20160212001846.GB29511@momjian.us
--
Michael