Filtering Nulls from multiple columns - sql

Question

Using SQL Server 2005.
Data is in 2 separate tables and I have only been given write permissions.
Data looks like:
DateTime1 | DateTime2
-----------------------
2012-06-01 | 2012-06-01
2012-06-02 | 2012-06-02
2012-06-04 | 2012-06-05
2012-06-02 | NULL
NULL | 2012-06-05
2012-06-04 | 2012-06-05
NULL | NULL
What I am trying to do is be able to count values in which DateTime1 and DateTime2 contain values, DateTime1 contains a date and DateTime2 is NULL, DateTime1 is NULL and DateTime2 contains values.
Overall Im trying to avoid DateTime1 being Null and DateTime2 being null.
My where statement looks like this:
Where (DateTime1 is not null or DateTime2 is not null)
The only problem is it is still showing where both are null values. Anyone know why this might be happening or how to solve it?
Thanks
EDIT
Full Query as requested by #Lamak
;With [CTE] As (
Select
TH.ID
,AMT
,Reason
,EffDate
,DateReq
,CS_ID
,ROW_NUMBER()
Over (Partition By ID Order By [pthPrimeKey] Desc) as [RN]
From
DateTime1Table as [MC] (nolock)
Left Join History as [TH] (nolock) on [TH].[ID] = [MC].[ID]
Left Join Trans as [SUB] (nolock) on [SUB].TransactionReasonCode = [TH].Reason
Left Join Renew as [RM] (nolock) on [MC].ID = [RM].ID
Where
([MC].[DateTime1] is not null or [RM].[DateTime2] is not null)
And [PostingDate] = DATEADD(dd, datediff(dd, 1, GetDate()),0)
)
SELECT
[ID]
,[AMT] as [Earned]
,[Reason] as [Reason]
,[EffDate] as [Eff]
,[DateReq] as [Date_Cancel_Req]
,[pthUserId_Number] as [CSR]
FROM [CTE]
Where RN <= 1

Answer 1

The following will allow rows to be included if
only DateTime1 has a value
only DateTime2 has a value
both have values
It will exclude rows where both values are NULL. Is that what you're after? (I tried to follow the conversations but got lost, and wish you'd have a simpler repro with sample data - I think the CTE and all the other joins and logic really take away from the actual problem you're having.)
WHERE COALESCE([MC].[DateTime1], [RM].[DateTime2]) IS NOT NULL
However, since you're performing a LEFT OUTER JOIN, this may belong in the ON clause for [RM] instead of WHERE. Otherwise you won't know if a row is excluded because the value in a matching row was NULL, or because there was no matching row. And maybe that's ok, just thought I would mention it.
EDIT
Of course, that clause provides the exact same results as ...
WHERE ([MC].[DateTime1] is not null or [RM].[DateTime2] is not null)
Want proof?
DECLARE #a TABLE(id INT, DateTime1 DATETIME);
DECLARE #b TABLE(id INT, DateTime2 DATETIME);
INSERT #a SELECT 1, '20120602' ; INSERT #b SELECT 1, NULL;
INSERT #a SELECT 2, NULL ; INSERT #b SELECT 2, '20120605';
INSERT #a SELECT 3, '20120604' ; INSERT #b SELECT 3, '20120605';
INSERT #a SELECT 4, NULL ; INSERT #b SELECT 4, NULL;
INSERT #a SELECT 5, '20120602' ; INSERT #b SELECT 9, NULL;
INSERT #a SELECT 6, NULL ; INSERT #b SELECT 10, '20120605';
INSERT #a SELECT 7, '20120604' ; INSERT #b SELECT 11, '20120605';
INSERT #a SELECT 8, NULL ; INSERT #b SELECT 12, NULL;
SELECT * FROM #a AS a LEFT OUTER JOIN #b AS b
ON a.id = b.id
WHERE COALESCE(a.DateTime1, b.DateTime2) IS NOT NULL;
SELECT * FROM #a AS a LEFT OUTER JOIN #b AS b
ON a.id = b.id
WHERE a.DateTime1 IS NOT NULL OR b.DateTime2 IS NOT NULL;
Both queries yield:
id DateTime1 id DateTime2
-- ---------- ---- ----------
1 2012-06-02 1 NULL -- because left is not null
2 NULL 2 2012-06-05 -- because right is not null
3 2012-06-04 3 2012-06-05 -- because neither is null
5 2012-06-02 NULL NULL -- because of no match
7 2012-06-04 NULL NULL -- because of no match
So as I suggested in the comment, if you're not seeing the rows you expect, you need to look at other parts of the query. If you provide sample data and desired results, we can try to help you narrow that down. As it is, I don't think we know enough about your schema and data to determine where the problem is.