Utilisez ROW_NUMBER()
:
SELECT
name,
ROW_NUMBER() OVER (ORDER BY name) AS id
FROM people;
MODIF :
Différence entre ORDER BY 1
vs ORDER BY column_name
SELECT
name,
ROW_NUMBER() OVER (ORDER BY name) AS id
FROM people;
/* Execution Plan */
QUERY PLAN WindowAgg (cost=83.37..104.37 rows=1200 width=38)
-> Sort (cost=83.37..86.37 rows=1200 width=38)
**Sort Key: name**
-> Seq Scan on people (cost=0.00..22.00 rows=1200 width=38)
SELECT
name,
ROW_NUMBER() OVER (ORDER BY 1) AS id
FROM people;
/* Execution Plan */
QUERY PLAN WindowAgg (cost=0.00..37.00 rows=1200 width=38)
-> Seq Scan on people (cost=0.00..22.00 rows=1200 width=38)
Dans le second cas, il n'y a pas d'opération de tri.
Vous pouvez également écrire la deuxième requête comme :
SELECT
name,
ROW_NUMBER() OVER () AS id
FROM people;
Pourquoi les gens écrivent ORDER BY 1
dans les fonctions fenêtrées ?
Parce que dans certains dialectes c'est obligatoire et ORDER BY 1
agit comme un espace réservé.
Oracle :
SELECT
name,
ROW_NUMBER() OVER (ORDER BY 1) AS id
FROM people;
TSQL :
SELECT
name,
ROW_NUMBER() OVER (ORDER BY (SELECT 1)) AS id
FROM people;