如果添加

SET SESSION sql_mode = 'ONLY_FULL_GROUP_BY';

然后，您将看到为什么其他RDBMS不允许使用此语法的原因:

Then you'll see the why other RDBMS won't allow this syntax:

• SELECT中3个未聚合的列，而GROUP BY中1个
• ORDER BY列不在GROUP BY/SELECT中，并且未聚合

• 3 un-aggregated columns in SELECT but one in GROUP BY
• ORDER BY column is not in GROUP BY/SELECT and not aggregated

如果要分组而不是DISTINCT，则需要对SELECT中的所有列进行分组

If you want to GROUP BY rather then DISTINCT, then you need to GROUP BY all column in the SELECT

SELECT s.serviceid, s.servicenameit, s.servicenameen 
FROM services s, kpi k, checklist_has_kpi chk 
WHERE s.serviceid=k.serviceid AND k.kpiid=chk.kpiid AND k.inreport='yes' AND chk.checklistid=61 
GROUP BY s.serviceid, s.servicenameit, s.servicenameen

但是，无论使用GROUP BY还是DISTINCT，您都没有chk.order可以进行排序

But then you have no chk.order to order by, whether using GROUP BY or DISTINCT

那怎么办，完全忽略重复项?

So what about this, ignoring duplicates completely?

SELECT s.serviceid, s.servicenameit, s.servicenameen 
FROM services s, kpi k, checklist_has_kpi chk 
WHERE s.serviceid=k.serviceid AND k.kpiid=chk.kpiid AND k.inreport='yes' AND chk.checklistid=61 
ORDER BY chk.order ASC

或者将此命令按3x services列的最早顺序排序

Or this to ORDER BY the earliest order per 3x services columns

SELECT s.serviceid, s.servicenameit, s.servicenameen 
FROM services s, kpi k, checklist_has_kpi chk 
WHERE s.serviceid=k.serviceid AND k.kpiid=chk.kpiid AND k.inreport='yes' AND chk.checklistid=61 
GROUP BY s.serviceid, s.servicenameit, s.servicenameen
ORDER BY MIN(chk.order)