Am 14.01.24 um 12:17 schrieb Yongtao Huang:
My question is why scan all columns in PG 16.01?
If `select distinct c1`, scan the column `c1` is enough, like PG 9.4.
good question, I think because the seq scan always reads the column. If
you create an index on c1, this changes:
postgres=# create index idx1 on t1(c1);
CREATE INDEX
postgres=# set enable_seqscan to off;
SET
postgres=# explain (costs, verbose) select distinct c1 from t1;
QUERY PLAN
-------------------------------------------------------------------------------------
Unique (cost=0.15..63.93 rows=200 width=4)
Output: c1
-> Index Only Scan using idx1 on public.t1 (cost=0.15..61.10
rows=1130 width=4)
Output: c1
(4 rows)
now we scan only the index and not the heap.
Regards, Andreas
--
Andreas Kretschmer
CYBERTEC PostgreSQL Services and Support
