create rum index without "WITH" cause an error when select use “ORDER BY”

[Wenbo94]

Use postgres REL_12_STABLE branch

postgres=# CREATE TABLE tsts (id int, t tsvector, d timestamp);
CREATE TABLE
postgres=# \copy tsts from '/rum/data/tsts.data'
COPY 508
postgres=# CREATE INDEX tsts_idx ON tsts USING rum (t rum_tsvector_addon_ops, d);
CREATE INDEX
postgres=# SELECT id, d, d <=> '2016-05-16 14:21:25' FROM tsts WHERE t @@ 'wr&qh' ORDER BY d <=> '2016-05-16 14:21:25' LIMIT 5;
ERROR: cannot order without attribute 2 in WHERE clause
postgres=# DROP INDEX tsts_idx;
DROP INDEX
postgres=# SELECT id, d, d <=> '2016-05-16 14:21:25' FROM tsts WHERE t @@ 'wr&qh' ORDER BY d <=> '2016-05-16 14:21:25' LIMIT 5;
id | d | ?column?
-----+----------------------------+---------------
355 | 2016-05-16 14:21:22.326724 | 2.673276
354 | 2016-05-16 13:21:22.326724 | 3602.673276
371 | 2016-05-17 06:21:22.326724 | 57597.326724
406 | 2016-05-18 17:21:22.326724 | 183597.326724
415 | 2016-05-19 02:21:22.326724 | 215997.326724
(5 rows)

[sokolcati]

Hi! This is the expected behavior of the index.

In the case of WITH, we attach one attribute to another, so if you specify one of the columns in ORDER BY, the sort order will be chosen unambiguously.
If the index is created on several fields without attach-to, then you will need to specify the remaining fields in ORDER BY. In the case of your example, you need to add the t-attribute after d <=> '2016-05-16 14:21:25':

SELECT id, d, d <=> '2016-05-16 14:21:25' FROM tsts WHERE t @@ 'wr&qh'
ORDER BY d <=> '2016-05-16 14:21:25', t LIMIT 5;

This query works fine with an index created without WITH.

But I agree that the error message is misleading. The second field is expected not in WHERE , but in ORDER BY . The error text will be changed in the #133.