SQL单组分组函数错误/非分组表达式

Question

I'm trying to create a view, but when i include the "group by" function, it says that there is an error at line 2 saying that it is not a group-by expression. If i comment out the "group-by" function, then it says theres an single-group group function error in the view. how can i fix this?

CREATE or REPLACE VIEW bestseller_view
  AS SELECT e.empl_name,
            NVL((sum((si.amt_paid+a.ao_price)-fs.purchase_cost+a.ao_purchase_price)*se.salary_comm),0) sales_commission,
            COUNT(si.s_empl_id) vehicles_sold
 FROM for_sale fs
 JOIN sales_invoice si ON fs.vehicle_id = si.vehicle_id
 JOIN sales_empl se    ON si.s_empl_id = se.empl_id
 JOIN purchased_ao pao ON si.sales_invoice_id=pao.sales_invoice_id
 JOIN add_ons a        ON pao.ao_id=a.ao_id
 JOIN employee e       ON se.empl_id=e.empl_id;
 GROUP BY si.s_empl_id;

Answer 1

The non-aggregate column in your select is e.empl_name , so you need to group by that, not si.s_empl_id; :

... SELECT e.empl_name,
  NVL((sum(...),0) sales_commission,
  COUNT(...) vehicles_sold
...
GROUP BY e.empl_name;

But since empl_name presumably isn't guaranteed to be unique, grouping by the ID makes more sense. You need to either select and group by both columns - though using e.empl_id might look more logical - and just ignore the ID you get back if it isn't relevant, in this case by just never selecting it from the view; or use this as an inner query and don't include it in the outer one. Seems like it makes more sense to have it in the view in case anyone does want/need it one day.

... SELECT e.empl_id, e.empl_name,
  NVL((sum(...),0) sales_commission,
  COUNT(...) vehicles_sold
...
GROUP BY e.empl_id, e.empl_name;

You also have your parentheses in the wrong place in your sum :

NVL((sum((si.amt_paid+a.ao_price)-fs.purchase_cost+a.ao_purchase_price)
     *se.salary_comm),0) sales_commission,

... is calculating the sum of all the (si.amt_paid+a.ao_price)-fs.purchase_cost+a.ao_purchase_price values, and then multiplying that sum by the se.salary_comm , which makes that column non-aggregate as well; I'm pretty sure that's not what you meant to do, so that should be:

NVL(sum((si.amt_paid+a.ao_price-fs.purchase_cost+a.ao_purchase_price)
  *se.salary_comm),0) sales_commission,

Or if there is a one-to-one relationship between e and se , which looks likely, you could leave that as it is and add se.salary_comm to the group by clause as @ypercube suggested.

Answer 2

 SELECT e.empl_name, si.s_empl_id
 ...
 GROUP BY e.empl_name, si.s_empl_id