I am trying to update a table using a composite key, and can't figure out the syntax. I have a table defined as:
create schema test;
create type test.ra_dec as (f1 double precision, f2 double precision);
create table test.pos_vel(
xp double precision,
yp double precision,
xv double precision,
yv double precision
);
insert into test.pos_vel(xp,yp,xv,yv) values
(1,2,5,10),
(2,3,10,20),
(3,4,20,40);
I can verify my define type works:
gsh=# select cast( (100,200) as test.ra_dec);
row
-----------
(100,200)
(1 row)
But I can't figure out how to update my table.
gsh=# update test.pos_vel set (xv,yv) = (select cast( (100,200) as test.ra_dec));
ERROR: syntax error at or near "select"
LINE 1: update test.pos_vel set (xv,yv) = (select cast( (100,200) as...
I know for this trival example I can do it without using a composite type, but for speed reasons I want to define a function in pl/python and have it return a composite type, so I need to figure out the proper syntax.
Can anyone help please?
-- drop schema if exists test cascade;
create schema test;
create type test.ra_dec as (f1 double precision, f2 double precision);
-- Your function here
create function test.f() returns test.ra_dec language sql as $$
select (100,200)::test.ra_dec $$;
create table test.pos_vel(
xp double precision,
yp double precision,
xv double precision,
yv double precision
);
insert into test.pos_vel(xp,yp,xv,yv) values
(1,2,5,10),
(2,3,10,20),
(3,4,20,40);
update test.pos_vel set (xv,yv) = (select * from test.f());
Update:
Because sub-select for such cases was introduced in the PostgreSQL 9.5 it is not working in earlier versions. There are two workarounds:
update test.pos_vel set (xv,yv) = ((test.f()).f1, (test.f()).f2);
update test.pos_vel set (xv,yv) = (t.f1, t.f2) from (select * from test.f()) t;
Tested on 8.4: http://dbfiddle.uk/?rdbms=postgres_8.4&fiddle=0a4927ed804b76442a69f8259cca1929