J'utiliserais insert .. on duplicate key update ...
déclaration. Quelque chose comme ça :
insert into highscore set
name = :name,
score = :new_score
on duplicate key update
score = greatest(score, :new_score)
name
la colonne doit être indexée comme unique
.
Scénario de test :
create table player (
name varchar(32) primary key,
score int not null default 0
);
-- create new players
insert into player set name = 'foo', score = 100
on duplicate key update score = greatest(score, 100);
insert into player set name = 'bar', score = 100
on duplicate key update score = greatest(score, 100);
insert into player set name = 'baz', score = 100
on duplicate key update score = greatest(score, 100);
-- update score of existing player
insert into player set name = 'bar', score = 200
on duplicate key update score = greatest(score, 200);
Sortie de select * from player
:
+------+-------+
| name | score |
+------+-------+
| bar | 200 |
| baz | 100 |
| foo | 100 |
+------+-------+