How does postgresql insert multiple rows with default values for all columns?

502    Asked by DavidEDWARDS in SQL Server , Asked on Mar 10, 2023

I can insert multiple rows into a table with default values for all columns the RBAR way:

create table course(course_id serial primary key);

do $$
begin
  for i in 1..100000 loop
    insert into course default values;
  end loop;
end;$$;
Is there a way of doing the same with a single SQL statement?
Answered by David Edmunds

Postgresql insert multiple rows with default values for all columns- Using generate_series() and ctes. Tested in rextester.com:

create table t
( tid serial primary key,
  i int default 0,
  name text default 'Jack'
) ;

with ins as

  (insert into t (i, name)               -- all the columns except any serial
   values (default, default)
   returning i, name
  )
insert into t
  (i, name)
select
  ins.i, ins.name

from

  ins cross join generate_series(1, 9); -- one less than you need

For the case when there is only one column and it's a serial, I see no way to use the default. Using the generate_series is straight-forward:

insert into course

  (course_id)
select
  nextval('course_course_id_seq')
from
  generate_series(1, 10);

If there are other, more "peculiar" default values, like a UUID function or the non-standard clock_timestamp(), the statement will have to be adjusted accordingly, like the serial case. An idea to improve came from the similar question: Inserting dummy data into an empty table having a primary key integer field GENERATED ALWAYS AS IDENTITY, using the OVERRIDING USER VALUE option in the INSERT statement. This is available only in versions 10+, not in 9.6 and previous.

Note: it does NOT work if the table has a single column which is serial.

Tested in dbfiddle.uk:
create table t
( tid serial primary key,
  i int default 0,
  name text default 'Jack'
) ;
insert into t (i)
  OVERRIDING USER VALUE
select null
from generate_series(1, 10) as gs(i) ;


Your Answer

Interviews

Parent Categories