At Wed, 17 Jan 2018 22:26:15 +0300, Sergei Kornilov <sk@zsrv.org> wrote in <412861516217175@web38o.yandex.ru> > Hello > I can reproduce on actual 9.6.6, 10.1 and fresh master build > (9c7d06d60680c7f00d931233873dee81fdb311c6 commit). I did not check > earlier versions > > set enable_indexonlyscan to off ; > postgres=# SELECT w FROM words WHERE w LIKE '%e%'; > w > ------- > Lorem > Index scan result is correct. Affected only index only scan, > > PS: i find GIST(w gist_trgm_ops, w); some strange idea, but result > is incorrect in any case.
The cause is that gist_trgm_ops lacks "fetch" method but planner is failing to find that.
Index only scan is not usable in the case since the first index column cannot be rechecked but check_index_only makes wrong decision by the second occurance of "w'. There may be a chance that recheck is not required but we cannot predict that until actually acquire a tuple during execution.
I didn't get this point. Same column is indexed twice using two
different opclasses. The first opclass doesn't support index-only scan,
while the second opclass does support it. So, if the first opclass needs
recheck then it can be rechecked using the value got from the second
opclass. Thus, I think we can potentially support index-only scan
in this case.
Another thing is that it's probably hard to do in our current optimizer/executor/am
infrastructure. And assuming that use-case is quite narrow, and it looks
OK to just disable such feature as bug fix.
------ Alexander Korotkov Postgres Professional: http://www.postgrespro.com The Russian Postgres Company
