On 2013-08-30 17:15:32 -0400, Alvaro Herrera wrote:
> Andres Freund wrote:
>
> > No need, found the bug. And I think can build a testcase myself.
> >
> > ExecReScanMergeAppend resets ms_initialized, but doesn't clear the
> > binaryheap. Thus no new elements fit.
>
> Um. Are we missing a binaryheap_clear() method?
In the patch I am patch-to-be it's binaryheap_reset(), but yes ;). Are
you already patching it, or do you want me to finish it? I have a not so
nice testcase and I can confirm that this is the issue and that a
binaryheap_reset() fixes it:
SELECT (SELECT g.i FROM ((SELECT random()::int ORDER BY 1 OFFSET 0) UNION ALL (SELECT random()::int ORDER BY 1 OFFSET
0))f(i) ORDER BY f.i LIMIT 1) FROM generate_series(1, 10) g(i);
Greetings,
Andres Freund
--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services