Postgresql：仅当两个字段分别唯一时才选择字段

我需要选择10个独特的位置，每个位置应该来自不同的用户。 因此，user_id和location_id中的值应该是唯一的。 尝试了很多方法，但不知怎么的，都没有成功

不考虑uniques的查询如下所示：

SELECT
    b.user_id, 
    l.location_id,
    ROUND(distance(transform(PointFromText('POINT(4.0 52.0)', 4326), 26986),transform("spatialPoint", 26986))) AS distance
 FROM  locations AS l
    JOIN b_user_location AS b USING(location_id)
 ORDER BY ROUND(distance(transform(PointFromText('POINT(4.0 52.0)', 4326) ASC
 LIMIT 10 OFFSET 0

看来我需要这样的东西：

SELECT
   DISTINCT ON(user_id)
   DISTINCT ON(location_id)
   b.user_id,
   .....

SELECT DISTINCT ON(location_id) * 
FROM (
SELECT DISTINCT ON(b.user_id)
    b.user_id, 
    l.location_id,
    ROUND(distance(transform(PointFromText('POINT(4.0 52.0)', 4326), 26986),transform("spatialPoint", 26986))) AS distance
 FROM  locations AS l
    JOIN b_user_location AS b USING(location_id)
 ORDER BY ROUND(distance(transform(PointFromText('POINT(4.0 52.0)', 4326) ASC
)
 LIMIT 10 OFFSET 0

但这是不允许的

谁能给我指一下正确的方向吗

[编辑] 链接到SqlFiddle

Sql查询：

CREATE TABLE users (user_id SERIAL, username VARCHAR(255), CONSTRAINT users_pkey PRIMARY KEY (user_id));
CREATE TABLE locations(location_id SERIAL, "spatialPoint" INT,name VARCHAR(255),CONSTRAINT locations_pkey PRIMARY KEY (location_id));
CREATE TABLE b_user_location(user_id INT, location_id INT, CONSTRAINT user_id_fkey FOREIGN KEY(user_id) REFERENCES users(user_id),CONSTRAINT location_id_fkey FOREIGN KEY(location_id) REFERENCES locations(location_id));

INSERT INTO users(username)VALUES('user_1'),('user_2'),('user_3');
INSERT INTO locations("spatialPoint", name)VALUES(1,'location_1'), (2, 'location_2'), (3,'location_3');

INSERT INTO b_user_location(user_id, location_id)VALUES(1,1),(1,2),(1,3),(2,1),(2,2),(2,3),(3,1),(3,2),(3,3);

SELECT
  b.user_id, 
  l.location_id,
  l."spatialPoint" AS distance
FROM  locations AS l
JOIN b_user_location AS b USING(location_id)
ORDER BY l."spatialPoint" ASC
LIMIT 10 OFFSET 0

[/edit]

尝试以下方法：

SELECT
   DISTINCT ON(user_id)
   DISTINCT ON(location_id)
   b.user_id,
   .....

SELECT DISTINCT ON(location_id) * 
FROM (
SELECT DISTINCT ON(b.user_id)
    b.user_id, 
    l.location_id,
    ROUND(distance(transform(PointFromText('POINT(4.0 52.0)', 4326), 26986),transform("spatialPoint", 26986))) AS distance
 FROM  locations AS l
    JOIN b_user_location AS b USING(location_id)
 ORDER BY ROUND(distance(transform(PointFromText('POINT(4.0 52.0)', 4326) ASC
)
 LIMIT 10 OFFSET 0

---

它将为您提供10行，其中包含唯一的

user\u id

和唯一的

location\u id

，但这些行可能没有最低的

距离
示例数据或SQLFiddle在这里很有用。添加了Sql和SQLFiddle，tnx；）是的，它很接近，但顺序是必需的，我可以将它包装在另一个子查询中，但我必须看看这在性能方面是如何进行的。