Gregory Stark <stark@enterprisedb.com> writes: > You're almost there: > CREATE FUNCTION getfoo (IN int, OUT int, OUT int) returns setof record(int,int) AS $$ > SELECT fooid, foosubid FROM foo WHERE fooid = $1; > $$ LANGUAGE SQL; Not quite --- it's just "returns setof record". The output column types are defined by the OUT parameters. The only reason you need the returns clause is to have a place to stick the "setof" specification ... regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match

Gregory Stark <stark@enterprisedb.com> writes: > "Tom Lane" <tgl@sss.pgh.pa.us> writes: >> Not quite --- it's just "returns setof record". > I did test my example before posting it: > postgres=# postgres=# CREATE or replace FUNCTION getfoo (IN int, OUT int, OUT int) returns setof record(int,int)AS $$ > SELECT 1,2 union all select 2,3; > $$ LANGUAGE SQL; Interesting --- if you try it in anything older than 8.3, it will fail. What is happening here is that the "(int,int)" is being taken as a typmod (per Teodor's work to allow typmods for all data types), and apparently in this path we never check to see if it's a *valid* typmod. Now typmods are always discarded from function argument and result types, but it seems like we'd better validate that they're legal for the datatype anyway. Otherwise there will be confusion of just this sort. Comments, objections? regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match