Ok, I see. This is actually an interesting question. I don't understand which approach to use. I am using pd_dump and pg_restore of the receiving server. It might make sense to use pg_dump source, pg_restore sink
сб, 20 нояб. 2021 г. в 22:33, Tom Lane <t...@sss.pgh.pa.us>: > Adrian Klaver <adrian.kla...@aklaver.com> writes: > > On 11/20/21 04:27, Дмитрий Иванов wrote: > >> Why does pg_dump generate an insertion script in the generated columns? > > > Actually I think it should be DEFAULT. > > It should be, and it is when I try this example. I get output like > > -- > -- Data for Name: object; Type: TABLE DATA; Schema: bpd; Owner: postgres > -- > > INSERT INTO bpd.object (id, id_class, id_position, bquantity, > id_position_root, id_conception, barcode_unit, id_unit_conversion_rule, > "timestamp", on_freeze, timestamp_class, name, id_class_root, id_group, > id_group_root, id_object_carrier, "desc", id_class_prop_object_carrier, > id_pos_temp_prop, is_inside, mc) VALUES (51253, 1015, 461, 1, 461, 84, > 2020000512530, 14, '2021-11-14 08:40:31.381', false, '2021-02-19 > 11:01:28.402', 'NFC метка самоклеющаяся', 1013, 138, 138, -1, '', -1, -1, > DEFAULT, 1); > > I wonder what version of pg_dump is actually being used there. > > regards, tom lane >
