Postgres 14

By using db<>fiddle, you agree to license everything you submit by Creative Commons CC0.

create table histodureeconnexion ( id serial primary key, iduser int, connexion timestamp, deconnexion timestamp, duree_seconde integer );
insert into histodureeconnexion values 
(1,1,'2021-10-20 08:05:00','2021-10-20 08:15:40',640),
(2,1,'2021-10-20 09:25:00','2021-10-20 09:35:00',600),
(3,1,'2021-10-21 08:05:00','2021-10-20 09:35:00',600),
(4,1,'2021-10-21 09:00:00','2021-10-20 09:30:00',1800);
4 rows affected

SELECT *, SUM(duree_seconde) OVER (PARTITION BY connexion::date)
FROM histodureeconnexion;


id
iduser
connexion
deconnexion
duree_seconde
sum


1
1
2021-10-20 08:05:00
2021-10-20 08:15:40
640
1240

2
1
2021-10-20 09:25:00
2021-10-20 09:35:00
600
1240

3
1
2021-10-21 08:05:00
2021-10-20 09:35:00
600
2400

4
1
2021-10-21 09:00:00
2021-10-20 09:30:00
1800
2400


SELECT *, SUM(duree_seconde) OVER w
FROM histodureeconnexion
WINDOW w AS (PARTITION BY connexion::date ORDER BY connexion
             RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW);


id
iduser
connexion
deconnexion
duree_seconde
sum


1
1
2021-10-20 08:05:00
2021-10-20 08:15:40
640
640

2
1
2021-10-20 09:25:00
2021-10-20 09:35:00
600
1240

3
1
2021-10-21 08:05:00
2021-10-20 09:35:00
600
600

4
1
2021-10-21 09:00:00
2021-10-20 09:30:00
1800
2400

id	iduser	connexion	deconnexion	duree_seconde	sum
1	1	2021-10-20 08:05:00	2021-10-20 08:15:40	640	1240
2	1	2021-10-20 09:25:00	2021-10-20 09:35:00	600	1240
3	1	2021-10-21 08:05:00	2021-10-20 09:35:00	600	2400
4	1	2021-10-21 09:00:00	2021-10-20 09:30:00	1800	2400