Dans de nombreux autres SGBD (Oracle, SQL-Server, Postgres), vous pouvez utiliser des fonctions de fenêtre :
SELECT id, file, folder, added
FROM
( SELECT id, file, folder, added,
DENSE_RANK() OVER (ORDER BY added DESC) AS d_rank,
ROW_NUMBER() OVER (PARTITION BY added ORDER BY id DESC) AS row_no
FROM AviationImages
) d
WHERE d_rank <= 5 -- limit number of dates
AND row_no <= 5 ; -- limit number of images per date
Dans MySQL, vous n'avez pas le luxe de la fonction de fenêtre et de OVER
clause :
SELECT i.id, i.file, i.folder, i.added
FROM
( SELECT DISTINCT added
FROM AviationImages
ORDER BY added DESC
LIMIT 5
) AS da
JOIN
AviationImages AS i
ON i.added = da.added
AND i.id >= COALESCE(
( SELECT ti.id
FROM AviationImages AS ti
WHERE ti.added = da.added
ORDER BY ti.id DESC
LIMIT 1 OFFSET 4
), -2147483647) ; -- use 0 if the `id` is unsigned int
Un index sur (added, id)
aidera l'efficacité - et si la table utilise InnoDB et le id
est la clé primaire, puis juste un index sur (added)
suffira.