Usa PostgreSQL
extensión DISTINCT ON
:
SELECT DISTINCT ON (documents.id) *
FROM document
JOIN updates
ON updates.document_id = document_id
ORDER BY
documents.id, updates.date DESC
Esto tomará la primera fila de cada document.id
clúster en ORDER BY
orden.
Script de prueba para comprobar:
SELECT DISTINCT ON (documents.id) *
FROM (
VALUES
(1, 'Test Title'),
(2, 'Test Title 2')
) documents (id, title)
JOIN (
VALUES
(1, 1, '2006-01-01'::DATE),
(2, 1, '2007-01-01'::DATE),
(3, 1, '2008-01-01'::DATE),
(4, 2, '2009-01-01'::DATE),
(5, 2, '2010-01-01'::DATE)
) updates (id, document_id, date)
ON updates.document_id = documents.id
ORDER BY
documents.id, updates.date DESC