Question

I have a (rather complicated) SQL statement where I select data from lots of different tables, and to cope with a bad legacy data structure, I have a couple of custom columns that get their values based on values from other columns. I have currently solved this with CASE statements:

 SELECT
     ...,
     CASE channel
         WHEN 1 THEN channel_1
         WHEN 2 THEN channel_2
         ...
         ELSE 0
     END AS ChannelValue,
     CASE channelu
         WHEN 1 THEN channelu_1
         WHEN 2 THEN channelu_2
         ...
         ELSE '0'
     END AS ChannelWithUnit,
     ...
 FROM 
     ... 
 --rest of statement continues with multiple joins and where/and clauses...

I get all the results I expect when executing the query in MS SQL Server Management Studio, and the column names are listed as I have specified in my AS clauses. However, for some reason I’m not allowed to use the conditional values in a WHERE statement. If I add

AND ChannelValue > Limit * p.Percentage / 100

at the end of the query, I get an error on that line saying

Msg 207, Level 16, State 1, Line 152
Invalid column name ‘ChannelValue’

Why is this not allowed? What should I do instead?

Answer 1

The only part of the SQL Statement where it is valid to use an alias declared in the SELECT list is the ORDER BY clause. For other parts of the query you just have to repeat the whole CASE expression and trust the optimiser to recognise it is the same.

If you are on SQL2005+ you can use a CTE to avoid this issue which sometimes helps with readability.

WITH YourQuery As
(

 SELECT
     Limit, 
     Percentage,
     CASE channel
         WHEN 1 THEN channel_1
         WHEN 2 THEN channel_2
         ...
         ELSE 0
     END AS ChannelValue,
     CASE channelu
         WHEN 1 THEN channelu_1
         WHEN 2 THEN channelu_2
         ...
         ELSE '0'
     END AS ChannelWithUnit,
     ...
 FROM 
)

select ...
FROM YourQuery WHERE
ChannelValue > Limit * Percentage / 100