Hi Tom,
Thanks for the answer.
It's clearer now, I might avoid this behavior ordering by id at last.
I still have the impression that is a bit buggy or misleading.
Even with simple query such as :
SELECT c.id
FROM public.testme c
WHERE c.status = ANY ('{5}')
ORDER BY c.status DESC
OFFSET 150
LIMIT 200
the last is always the same but not the penultimate. Even if the ordering is
underdetermined it shouldn't change then LIMIT/OFFSET shouldn't return the
same value for 2 non overlapping ranges.
I find it difficult to understand how it works completely, but my knowledge
are quite limited.
Thanks anyway. :)
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/BUG-8629-Strange-resultset-when-using-CTE-or-a-subselect-tp5780187p5780519.html
Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.