两年之间的SQL日期减法

Question

I am having an issue with a query I am trying to convert from MS Access. The query flags record for removal when it is older than 90 days but when I convert this query to sql server is is removing too many records.

UPDATE  DT.SM_T_CountTotals
       SET  IsActive = 0
WHERE Convert(varchar, DT.SM_T_CountTotals.PostDate, 101) <
        Convert(varchar, GetDate()- 90, 101) 

When I run this query in MS Access I get a total of 3793 records that are flagged but in SQL server I get 69061 records that are flagged for removal. The GetDate()-90 value is correct at 10/26/2010 but it is flagging everything from this year to be removed.

I am sure it is something easy that I am overlooking. Help please?

I figured it out:

UPDATE  DT.SM_T_CountTotals
   SET  IsActive = 0
WHERE DT.SM_T_CountTotals.PostDate < Convert(varchar, GetDate()- 90, 101) 

Answer 1

You're comparing VARCHAR values, not DATEs.

101 converts to MM/DD/YY, so you're comparing month, then day, then year.

You should be using 112 (yymmdd)

Answer 2

Calculations between two dates can be easily done in the native data type rather than convert it to string. One can (and you have) get incorrect answers from such conversions.

Use DateDiff in the where clause to get the records that are more than 90 days old.

http://msdn.microsoft.com/en-us/library/ms189794.aspx

UPDATE  DT.SM_T_CountTotals
SET  IsActive = 0
WHERE ABS (DATEDIFF (dd, Getdate(), DT.SM_T_CountTotals.PostDate)) > 90