NULL 连接表的值缺失

Question

这是一个简单的例子： sqlfiddle.com /#!9/8087c5/1

我有以下查询：

select
    contracts.ID as contractID,
    departments.ID as departmentID,
    max(invoices.timestamp) as timestamp
from contracts
    left join departments on departments.ID = contracts.departmentID and departments.accountingType > 0
    left join invoices on contracts.ID = invoices.contractID
group by contractID
having departmentID is not null;

我期望得到如下结果：

+------------+--------------+---------------------+
| contractID | departmentID | timestamp           |
+------------+--------------+---------------------+
| 101        | 301          | NULL                |
+------------+--------------+---------------------+
| 102        | 302          | NULL                |
+------------+--------------+---------------------+
| 103        | 303          | 2020-05-01 11:11:00 |
+------------+--------------+---------------------+

相反，我得到：

+------------+--------------+---------------------+
| contractID | departmentID | timestamp           |
+------------+--------------+---------------------+
| 103        | 303          | 2020-05-01 11:11:00 |
+------------+--------------+---------------------+

我不明白为什么查询会削减最后一个左连接表的 NULL 值。 我错过了一些非常简单的东西吗？

Answer 1

好的 - 问题是group by contractID正确应该是group by contracts.ID 。 按连接发票 talbes 列 contractID 分组的group by contractID分组，这就是缺少列的原因。