Tom Lane writes:
>> On 08/01/2015 05:59 PM, Tom Lane wrote:
>>> Well, I certainly think all of these represent bugs:
>>>> 1 | ERROR: could not find pathkey item to sort
>
> Hmm ... I see no error with these queries as of today's HEAD or
> back-branch tips. I surmise that this was triggered by one of the other
> recently-fixed bugs, though the connection isn't obvious offhand.
I still see this error in master as of b8cbe43, but the queries are
indeed a pita to reproduce. The one below is the only one so far that
is robust against running ANALYZE on the regression db, and also
reproduces when I run it as an EXTRA_TEST with make check.
regards,
Andreas
select rel_217088662.a as c0, rel_217088554.a as c1, rel_217088662.b as c2, subq_34235266.c0 as c3, rel_217088660.id2
asc4, rel_217088660.id2 as c5
from public.clstr_tst as rel_217088554 inner join (select rel_217088628.a as c0 from
public.rtest_vview3as rel_217088628 where (rel_217088628.b !~ rel_217088628.b) and ((((rel_217088628.b ~~*
rel_217088628.b) or (rel_217088628.b ~* rel_217088628.b)) or (rel_217088628.b <> rel_217088628.b)) or
(rel_217088628.b= rel_217088628.b))) as subq_34235266 inner join public.num_exp_mul as rel_217088660 inner join
public.onek2as rel_217088661 on (rel_217088660.id1 = rel_217088661.unique1 ) on (subq_34235266.c0 = rel_217088660.id1
) inner join public.main_table as rel_217088662 on (rel_217088661.unique2 = rel_217088662.a ) on
(rel_217088554.b= rel_217088660.id1 )
where rel_217088554.d = rel_217088554.c
fetch first 94 rows only;