Re: BUG #14289: Potential bug: "invalid attribute number" when dblink result is assigned in PL/PGSQL

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: m(dot)overmeyer(at)yahoo(dot)ca
Cc: pgsql-bugs(at)postgresql(dot)org
Subject: Re: BUG #14289: Potential bug: "invalid attribute number" when dblink result is assigned in PL/PGSQL
Date: 2016-08-18 18:51:57
Message-ID: 19568.1471546317@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

m(dot)overmeyer(at)yahoo(dot)ca writes:
> I believe I may have found a bug in PL/PGSQL or dblink.

So you did. Fixed, thanks.

For the archives' sake, the test case looked like

------
create extension dblink;

create table test_table
(
username text primary key
);

INSERT INTO test_table VALUES ('test_user');

CREATE OR REPLACE FUNCTION test_assignment(primary_key_attnums int2vector)
RETURNS text AS
$BODY$
DECLARE
results text;
BEGIN
-- This doesn't work:
results := dblink_build_sql_delete('test_table', primary_key_attnums, 1, ARRAY['test_user']::text[]);

-- But this one (use value instead of parameter) does work:
--results := dblink_build_sql_delete('test_table', '1'::int2vector, 1, ARRAY['test_user']::text[]);

--And so does this one:
-- SELECT dblink_build_sql_delete('test_table', primary_key_attnums, 1, ARRAY['test_user']::text[]) INTO results;

return results;
END;
$BODY$
LANGUAGE plpgsql VOLATILE
COST 100;

SELECT test_assignment('1'::int2vector);
-----

and the problem was that plpgsql thought it could convert the int2vector
argument into an "expanded array". But int2vector doesn't allow toasting
so a fortiori it can't handle being expanded.

regards, tom lane

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message rotten 2016-08-18 19:06:23 BUG #14290: materialized view refresh doesn't use temp_tablespace
Previous Message Joe Conway 2016-08-18 18:25:27 Re: BUG #14289: Potential bug: "invalid attribute number" when dblink result is assigned in PL/PGSQL