SQL Server中的Isnull和数据类型

Question

Suppose I have a table like:

Tab(MyDate datetime null, MyIs bit null, ...)

Then I have SQL Like

    select Isnull(MyDate, 1), IsNull(MyIs, 999) from Tab;

If MyDate , MyIs value is null in DB, the result is:

    MyDate                   MyIs  
    1900-01-02 00:00:00.000  1

Not the expected value like

MyDate        MyIs  
1             999

So it looks like the return value is based on data type of the column. How to resolve this issue?

For my testing case as above, I Only got one row from the table.

Answer 1

The way you're looking for, you don't.

All the values in the same column must be the same data type. If your data had NOT NULL values in some rows (returning the column's data-type) and NULL values in other rows (returning a different data type, as per your CASE statement) , you would break this requirement.

You could have two extra fields...

select
  MyDate,
  CASE WHEN MyDate IS NULL THEN 1   ELSE 0 END As MyDateWibble,
  MyIs,
  CASE WHEN MyIs   IS NULL THEN 999 ELSE 0 END AS MyIsWibble
FROM
  Tab;

You could CAST() everything to VARCHAR() and return strings instead of DATETIME , BIT or INT .

But, more probably, I'd review the reason you want to do this in the first place and work out an alternative design.

Answer 2

You are mixing up the datatypes.

For example, when you test for the data you return integer 1. This will get translated to the data where sql server starts to count from + 1 (day)

Answer 3

you can do

select 
    case when mydate is null
    then '1' else cast(mydate as varchar(40))
    end,
    case when MyIs is null
    then 999 else MyIs
    end
from tab

Answer 4

SQL servers "zero" date or base date is 1900-01-01 so you are just converting MYDate null with 1 so it is adding 1 day in base date as

declare @d datetime = 1
select @d

1900-01-02 00:00:00.000

set @d = 0
select @d

1900-01-01 00:00:00.000

MyIs is a bit type so anything except 0 is '1' like

declare @b bit = 0
select @b

set @b = 1
select @b

set @b = -1
select @b

Above behavior of columns because of different data types