Postgres insists on Primary Key in Group By statement whereas SQLite does not - postgresql

Question

I am having a problem with Postgres insisting the primary key is part of the “group by” clause, the frustration being that the syntax I have works perfectly with SQLITE.
The error message is ‘column "attendees.id" must appear in the GROUP BY clause or be used in an aggregate function’
My question is can anyone provide the fix such that Postgres will produce the output that SQLite does (SQL statement, schema, data and desired output provided below)
My table schema:
CREATE TABLE Attendees
(
id INTEGER PRIMARY KEY AUTOINCREMENT,
activitynumber int,
activitydate DATE,
club int,
usernum int,
username ,
activityname ,
cost int,
received int,
update_time DATETIME,
CONSTRAINT FK_activitynumber
FOREIGN KEY (activitynumber)
REFERENCES Activity(activitynumber)
CONSTRAINT FK_club
FOREIGN KEY (club)
REFERENCES "club"(clubnum)
CONSTRAINT FK_usernum
FOREIGN KEY (usernum)
REFERENCES "users"(usernum)
);
CREATE INDEX ix_attendees_group ON Attendees
(club,activitydate,activitynumber);
(Please note, I added the index to try and solve the problem but it didn’t).
activities = Attendees.query.filter_by(club=current_user.club).order_by(Attendees.activitdate.desc()).group_by(Attendees.activitydate,Attendees.activitynumber)
The data I have is:-
id activitynumber activitydate club usernum username activityname cost received update_time
1 3 15-10-19 1002 1002000001 susan Monday Swim 200 0 58:43.8
2 3 15-10-19 1002 1002000002 triblokerich Monday Swim 200 0 58:49.9
3 4 17-10-19 1002 1002000001 susan Thursday Swim 200 0 59:04.5
4 4 17-10-19 1002 1002000015 craig Thursday Swim 200 0 59:09.9
5 6 16-10-19 1002 1002000001 susan Dunton 200 0 00:06.5
6 6 16-10-19 1002 1002000002 triblokerich Dunton 200 0 00:17.6
7 3 16-10-19 1002 1002000001 susan Monday Swim 300 0 58:28.1
8 3 16-10-19 1002 1002000002 triblokerich Monday Swim 300 0 58:33.7
9 3 16-10-19 1002 1002000015 craig Monday Swim 300 0 58:37.7
10 3 16-10-19 1002 1002000016 craig2 Monday Swim 300 0 01:41.8
11 3 19-10-19 1002 1002000001 susan Monday Swim 300 0 07:56.4
12 3 19-10-19 1002 1002000002 triblokerich Monday Swim 300 0 08:04.8
and the output I am trying to get (which works with SQLite) is:
Date Activity
2019-10-19 Monday Swim
2019-10-17 Thursday Swim
2019-10-16 Monday Swim
2019-10-16 Dunton
2019-10-15 Monday Swim

Answer 1

If I understand SQL Alchemy's syntax correctly you're looking for something like this, hopefully I'm not far off (broken into lines to show better):
activities = Attendees.query\
.with_entities(Attendees.activitydate, Attendees.activityname)\
.filter_by(club=current_user.club)\
.order_by(Attendees.activitdate.desc())\
.group_by(Attendees.activitydate,Attendees.activityname)
You always need to have the not-grouped-by columns in the result set as aggregates (min, max, sum, etc) because there's no way to show them otherwise. Which ID for example would be taken from the grouping of five rows? SQLite may be one of the inaccurate databases that just throws any of the rows back but PostgreSQL is very strict about the query and will never do this.
In this query it is defined that the result should contain only the columns that are in the group_by, since those are what you need. It also has the activityname as a grouped column rather than activitynumber since they need to match. This way there's no un-aggregated non-grouped columns in the result, only the grouped ones and the query should work fine.