Question

I have a table Actions, schema blow:

[Actions]
ActionID
Date
Status <--Nullable, a delta column, only stores value when status changes

Now I want to retrieve the latest record, however it is very likely that Stutus for that record is null, therefore I want to get its last status change(ranked by Date).

Here is an example:

ActionID | Date | Status
------------------------
1        | 04/12| 'Bon'
2        | 04/13| NULL
3        | 04/14| NULL
4        | 04/15| NULL

and my latest record should look like: ActionID: 4, Date: 04/15, Status: 'Bon'

I know it’s possible to do with nested select statements, but in my real table, I have about 10 of these columns, it will drastically affect the performance when a lot of queries like these are made. I wonder if there is a simpler way to do it?

Answer 1

Not sure if I understood your rules, but try this:

SELECT TOP 1
ActionID,
Date,
(SELECT TOP 1 Status FROM Actions WHERE Status IS NOT NULL ORDER BY ActionID DESC) AS Status
FROM Actions
ORDER BY ActionID DESC

---

You said you have 10 columns.. How it works?

Like scenario A)

 ActionID | Date  | StatusA | StatusB | StatusC
        1 | 04/11 | DELTA_A | NULL    | DELTA_C
        2 | 04/12 | NULL    | DELTA_B | DELTA_C
        3 | 04/13 | DELTA_A | NULL    | NULL 

..then multiple SELECT TOP 1 subqueries is still best choise I think..

but if it’s like scenario B)

 ActionID | Date  | StatusA | StatusB | StatusC
        1 | 04/11 | NULL    | NULL    | NULL
        2 | 04/12 | DELTA_A | DELTA_B | DELTA_C
        3 | 04/13 | NULL    | NULL    | NULL 

..then you may “reverse” your query like this:

SELECT TOP 1
(SELECT TOP 1 ActionID FROM Actions WHERE Status IS NOT NULL ORDER BY ActionID DESC) AS ActionID,
(SELECT TOP 1 Date FROM Actions WHERE Status IS NOT NULL ORDER BY ActionID DESC) AS Date,
StatusA,
StatusB,
StatusC
FROM Actions
WHERE StatusA IS NOT NULL -- then StatusB and StatusC are also NOT NULL
ORDER BY ActionID DESC

..but be aware you may get empty result if there is no row with StatusA = NOT NULL.