Re: Counting booleans in GROUP BY sections
"David G. Johnston" <[email protected]>
| Newsgroups | gmane.comp.db.postgresql.general |
|---|---|
| Message-ID | <CAKFQuwaZD_L5_BdELtXHpPekZGBPnKJbbKK5zS6zgPHmRTTzQg@mail.gmail.com> |
On Fri, Nov 29, 2019 at 12:48 PM Alexander Farber < [email protected]> wrote: > > if I remove GROUP BY mid, then I get the error: > > ERROR: 42803: column "words_moves.mid" must appear in the GROUP BY clause > or be used in an aggregate function > LINE 18: mid, > ^ > LOCATION: check_ungrouped_columns_walker, parse_agg.c:1369 > > Yes, you need to decide whether you want to output GROUPS (in which case any detail more specific than your desired group needs to be aggregated) or NOT (in which case you can probably use WINDOW functions to accomplish your goal - count(...) OVER (PARTITION BY <the level of grouping you desire knowledge about - year month it seems in this case>)) David J.