我有2张桌子。
1)汽车表:
2)参数表:
我需要提取具有 owner 参数匹配且在 insurance 中没有区别的重复汽车(这必须是相同的,或者两者都不存在< / strong>)。
我已成功使用ANSI语法中的LEFT JOIN
执行查询:
SELECT owner.name, owner.value, COALESCE (insur.name, 'insurance'), insur.value, count(*)
FROM car INNER JOIN param owner ON car.id = owner.car_id
LEFT JOIN param insur ON car.id = insur.car_id AND insur.name = 'insur'
WHERE owner.name = 'owner'
GROUP BY owner.name, owner.value, insur.name, insur.value
HAVING count(*) > 1
SQL Fiddle with properly working ANSI-syntax
但是当我用Oracle语法中的(+)
符号代替LEFT JOIN
重写此查询时,我得到了不同的结果:
SELECT owner.name, owner.value, COALESCE (insur.name, 'insurance'), insur.value, count(*)
FROM car,
param owner,
param insur
WHERE car.id = owner.car_id
AND owner.name = 'owner'
AND car.id (+) = insur.car_id -- key change
AND insur.name = 'insur'
GROUP BY owner.name, owner.value, insur.name, insur.value
HAVING count(*) > 1
SQL Fiddle with unexpected result of (+) instead LEFT JOIN
此查询的结果为空。我不明白如何用Oracle语法重写它以得到相同的查询结果。
答案 0 :(得分:0)
那是
SQL> select owner.name, owner.value,
2 coalesce (insur.name, 'insurance') in_name,
3 insur.value, count(*)
4 from car, param owner, param insur
5 where car.id = owner.car_id
6 and car.id = insur.car_id (+)
7 and insur.name (+) = 'insur'
8 and owner.name = 'owner'
9 group by owner.name, owner.value, insur.name, insur.value
10 having count(*) > 1;
NAME VALUE IN_NAME VALUE COUNT(*)
-------- -------- -------------------- -------- ----------
owner John insurance 2
SQL>
但是,为什么为什么要使用 old Oracle外连接语法?与ANSI连接相比,它仅具有缺点,而没有优点(至少,我没有想到)。实际上,我知道一个-如果您在旧的Forms&Reports 6i中使用外部联接(甚至更旧?我认为现在没有人使用这些版本),则它们的内置 PL / SQL引擎可能不会说 ANSI外部联接,所以您注定要使用旧的(+)
外部联接运算符。除此之外...不,不知道。