在内置函数中使用SQL列别名？

Question

I'm trying to order a selection based on the alias but I can't figure out how. Here is an example:

select distinct top 100 id, 
                    col1, 
                    col2, 
                    CASE WHEN @orderFormat = 'this' then col1
                         WHEN @orderFormat = 'that' then col2
                    END as orderby
from table
where col1 = like '%'
order by Len(orderby) asc, orderby asc

Whenever I pass my alias 'orderby' as an argument, it is reported as an invalid column.

My goal is to be able to order a variable column alphanumerically. I know that 'order by Len(orderby) asc, orderby asc works, but just not with an alias.

Anybody know a good way around this or if I'm doing something wrong?

Thanks!

EDIT:

I've managed to strip the select function down to this:

select top 200 Clip_Name as orderby
               from Clips
order by       Len(orderby) asc, orderby asc

Clip_Name is declared as column Clip_Name(nvarchar, not null) . The error from Microsoft SQL Server 2008 R2 Edition is Msg 207, Level 16, State 1, Line 1 Invalid column name 'orderby' .

However, this works (without the alias):

select top 200 Clip_Name 
               from Clips 
order by len(FLE_ID) desc, FLE_ID desc

Answer 1

When you're using DISTINCT , you can only order by expressions that are actually in the SELECT list. You can't reference columns, aliases or expressions that aren't there. Here's one possible workaround, though it might actually be better to simply remove the DISTINCT (if you have two rows with the same id then there is something seriously wrong with either your schema or at least the name of that column).

select distinct top 100 id, 
                    col1, 
                    col2, 
                    CASE WHEN @orderFormat = 'this' then col1
                         WHEN @orderFormat = 'that' then col2
                    END as orderby,
    len(CASE WHEN @orderFormat = 'this' then col1
             WHEN @orderFormat = 'that' then col2
        END) AS ignore_this_column
from table
where col1 like '%'
order by ignore_this_column, orderby;

Expressed much simpler so you don't have to repeat the expression (and also without the unnecessary DISTINCT ):

;WITH x AS 
(
  SELECT id, col1, col2, 
    orderby = CASE @orderFormat
      WHEN 'this' THEN col1
      WHEN 'that' THEN col2
    END
  FROM dbo.table
  WHERE col1 LIKE '%' -- necessary?
)
SELECT id, col1, col2
  FROM x
  ORDER BY LEN(orderby), orderby;

Answer 2

Based on you're first query, and following discussion that DISTINCT needs to be removed, this will work:

select top 100 id, 
        col1, 
        col2, 
        CASE WHEN @orderFormat = 'this' then col1
             WHEN @orderFormat = 'that' then col2
        END as orderby
from t
order by Len(CASE WHEN @orderFormat = 'this' then col1
                  WHEN @orderFormat = 'that' then col2
             END) asc, orderby asc