H2 database - Rank（）和Row_Num（）的替代方案

Question

我在我的java應用程序中使用H2數據庫。它似乎不支持rank()和row_num()功能，因爲他們仍然在他們的roadmap list

下面是我試圖執行

select * from (
    select *, 
      rank() over(partition by MSISDN order by PORT_IN_DATE desc) rank 
    from TEST_PORTIN 
) s 
where rank = 1 
    AND PORT_IN_DATE > '2012-01-16 23:20:27' 
ORDER BY PORT_IN_DATE 

Answer 1

你不需要RANK()爲您的特定查詢的查詢。你可以把它改寫如下：

SELECT * 
FROM TEST_PORTIN t1 
WHERE PORT_IN_DATE > '2012-01-16 23:20:27' 
AND NOT EXISTS (
    SELECT * 
    FROM TEST_PORTIN t2 
    WHERE t1.MSISDN = t2.MSISDN -- PARTITION BY clause translation 
    AND t1.PORT_IN_DATE > t2.PORT_IN_DATE -- ORDER BY clause translation 
) 
ORDER BY PORT_IN_DATE 

或者說，有點票友（但不一定更快），使用量化比較謂詞

SELECT * 
FROM TEST_PORTIN t1 
WHERE PORT_IN_DATE > '2012-01-16 23:20:27' 
AND PORT_IN_DATE >= ALL (
    SELECT PORT_IN_DATE 
    FROM TEST_PORTIN t2 
    WHERE t1.MSISDN = t2.MSISDN 
) 
ORDER BY PORT_IN_DATE 

[SQL秩和分區（HTTP的