如何使用子查詢與IBM的Informix 10.0

Question

RDBMS是INFORMIX 10.0

SELECT owner FROM systables WHERE TABNAME= ' VERSION'; 

返回9.50C1（這相當於版本10.x）

的子查詢工作得很好：

SELECT acct.fund_acct_nbr, acct.bin, prod.issuer_id, COUNT(*) 
FROM fund_acct AS acct 
JOIN products AS prod ON acct.cusip = prod.cusip 
WHERE prod.issuer_id = 'xxxx' 
AND SUBSTR(acct.bin, 1, 1) = 'x' 
GROUP BY acct.fund_acct_nbr, acct.bin, prod.issuer_id; 

我想將此查詢用作子查詢。

So my question is:

How do I get this done on IBM INFORMIX v10.0?
Should be simple and straight forward right?

任何嘗試使用該查詢作爲子查詢生成此有意義的消息：

「失敗查詢=> 2」

INFORMIX - 否「WITH」子句所以這將不起作用：

WITH issuer_accts AS 
(
SELECT acct.fund_acct_nbr, acct.bin, prod.issuer_id, COUNT(*) 
FROM fund_acct AS acct 
JOIN products AS prod ON acct.cusip = prod.cusip 
WHERE prod.issuer_id = 'xxxx' 
AND SUBSTR(acct.bin, 1, 1) = 'x' 
GROUP BY acct.fund_acct_nbr, acct.bin, prod.issuer_id 
) 
SELECT issuer_accts.issuer_id, COUNT(*) 
FROM issuer_accts 
GROUP BY issuer_accts.issuer_id; 

試圖以此爲榜樣(IBM documentation):

SELECT issuer_id, COUNT(*) 
FROM 
(
SELECT acct.fund_acct_nbr, acct.bin, prod.issuer_id, COUNT(*) 
FROM fund_acct AS acct 
JOIN products AS prod ON acct.cusip = prod.cusip 
WHERE prod.issuer_id = 'xxxx' 
AND SUBSTR(acct.bin, 1, 1) = 'x' 
GROUP BY acct.fund_acct_nbr, acct.bin, prod.issuer_id 
) issuer_accts 
GROUP BY issuer_id; 

試圖以此爲榜樣(specific to INFORMIX v10.0)

SELECT issuer_id, COUNT(*) 
FROM 
TABLE(MULTISET(
SELECT acct.fund_acct_nbr, acct.bin, prod.issuer_id, COUNT(*) 
FROM fund_acct AS acct 
JOIN products AS prod ON acct.cusip = prod.cusip 
WHERE prod.issuer_id = 'xxxx' 
AND SUBSTR(acct.bin, 1, 1) = 'x' 
GROUP BY acct.fund_acct_nbr, acct.bin, prod.issuer_id 
)) 
GROUP BY issuer_id; 

Answer 1

因爲Informix v10的這個早期版本不支持「Sub Query」/「Inline Que ry「/」內部查詢「，那麼常規做法是使用TEMP表來支持複雜的sql語句。因此，這是IBM的Informix的這個版本V10工作答案：

SELECT acct.fund_acct_nbr, acct.bin, prod.issuer_id, COUNT(*) 
FROM fund_acct AS acct 
JOIN products AS prod ON acct.cusip = prod.cusip 
WHERE prod.issuer_id = 'xxxx' 
AND SUBSTR(acct.bin, 1, 1) = 'x' 
GROUP BY acct.fund_acct_nbr, acct.bin, prod.issuer_id 
INTO TEMP temp_issuer_accts WITH NO LOG; 

SELECT issuer_id, COUNT(*) account_count 
FROM temp_issuer_accts 
GROUP BY issuer_id; 

DROP TABLE temp_issuer_accts; 

臨時表生存只爲會話的持續時間 - 所以請確保您連接保持在多個語句的執行開放。

相關參考這裏：

INTO TEMP clause

Using the WITH NO LOG option

Explicit inserts with SELECT...INTO TEMP statements

Duration of temporary tables

Answer 2

我想你只需要一個別名：

SELECT issuer_id, COUNT(*) 
FROM (SELECT acct.fund_acct_nbr, acct.bin, prod.issuer_id, COUNT(*) as cnt 
     FROM fund_acct AS acct JOIN 
      products AS prod 
      ON acct.cusip = prod.cusip 
     WHERE prod.issuer_id = 'xxxx' AND SUBSTR(acct.bin, 1, 1) = 'x' 
     GROUP BY acct.fund_acct_nbr, acct.bin, prod.issuer_id 
    ) x 
GROUP BY issuer_id; 

但是，根據有關數據最合理的假設，你也可以句話此爲：

SELECT issuer_id, COUNT(DISTINCT acct.bin || ' ' || prod.issuer_id) 
FROM fund_acct AS acct JOIN 
    products AS prod 
    ON acct.cusip = prod.cusip 
WHERE prod.issuer_id = 'xxxx' AND SUBSTR(acct.bin, 1, 1) = 'x' 
GROUP BY issuer_id;