Question

我一直在尝试编写一个sql查询来从bugzilla获取一些统计信息。这是查询

select bugs.bug_id AS bug_id, 
       COUNT(map_pingpong.bug_when) AS re_open, 
       MAX(map_closetime.bug_when) AS closed_date 
  from bugs 
         LEFT JOIN bugs_activity AS map_pingpong 
                 ON ((map_pingpong.bug_id = bugs.bug_id 
                        and map_pingpong.fieldid=15)) 
         LEFT JOIN bugs_activity AS map_closetime 
                 ON ((bugs.bug_id = map_closetime.bug_id 
                        and map_closetime.fieldid=8 
                        and bugs.bug_status = 'CLOSED' )) 
 where (bugs.assigned_to = 480) 
 GROUP BY bugs.bug_id 
 ORDER BY bug_id;

所以，查询应该返回两件事

1）发生事件的计数 2）事件发生的日期

因此，当我将查询分成两个不同的查询时，它们返回正确的值。如果我如上所述运行它，计数值是错误的（虽然日期是正确的）。我不应该在同一张桌子上运行两个连接？当你使用join时，或者count不应该在那里？

Answer 1

正如Barmar所说，你必须分离结果集以获得正确的计数：

SELECT 
    bugs.bug_id AS bug_id, 
    map_pingpong.cnt AS re_open, 
    map_closetime.mx AS closed_date
FROM bugs 
    LEFT JOIN (
        SELECT bug_id, COUNT(bug_when) AS cnt
        FROM bugs_activity
        WHERE fieldid = 15
        GROUP BY bug_id
    ) AS map_pingpong ON map_pingpong.bug_id = bugs.bug_id
    LEFT JOIN (
        SELECT ba.bug_id, MAX(ba.bug_when) AS mx
        FROM bugs_activity ba JOIN bugs ON bugs.bug_status = 'CLOSED' AND ba.bug_id = bug.bug_id AND bugs.assigned_to = 480
        WHERE ba.fieldid = 8
        GROUP BY ba.bug_id
    ) AS map_closetime ON bugs.bug_id = map_closetime.bug_id
WHERE bugs.assigned_to = 480
GROUP BY bugs.bug_id 
ORDER BY bug_id;

从功能上来说这是正确的，在性能方面可能是一场彻底的灾难，所以要小心......

在mysql join中使用count

1 个答案: