SQL How can I order the data using distinct? - sql

Question

I have the following Table 'tbl1'.
declare #tbl1 table ( col1 varchar(32))
insert into #tbl1
values ( 'C1' )
, ( 'B1' )
, ( 'X1' )
, ( 'A1' )
, ( 'B1' )
, ( 'C1' )
, ( 'B1' )
, ( 'A1' )
, ( 'X1' )
, ( 'C1' )
, ( 'D1' )
I tried the following query
select distinct col1
from #tbl1
order by col1
The output should come in the following order, and remove all the duplicate value
C1
B1
X1
A1
D1

Answer 1

You need to specify the order for your items, as others have noted.
Also you need a subquery and grouping to be able to sort over the minimal order column.
declare #source table (id int not null identity primary key, name nvarchar(3));
insert into #source
values ( 'C1' )
, ( 'B1' )
, ( 'X1' )
, ( 'A1' )
, ( 'B1' )
, ( 'C1' )
, ( 'B1' )
, ( 'A1' )
, ( 'X1' )
, ( 'C1' )
, ( 'D1' )
;
with grouped as
(
select min(id) as minId, name from #source
group by name
)
select name from grouped order by minId;
The query could be rewritten without CTE:
select grouped.name from
(select min(id) as minId, name from #source group by name) grouped
order by grouped.minId;
This yields exactly the result you requested.

Answer 2

To repeat what others have pointed out in comments: if you specify no ORDER, then the order of results is not guaranteed. The fact that you get your results in a certain order currently should be treated as coincidental. If you want a certain ordering in your results, you have to be explicit about it!
As an interesting note, in my experience this is especially important if you're doing a DISTINCT in your query, because depending on the statistics for your tables, the engine may or may not decide that ordering the data to execute the DISTINCT is in fact the best possible plan.
Given that you mention a very explicit ordering requirement...
The output should come in the following order, and remove all the duplicate value
C1
B1
X1
A1
D1
...you should make that explicit in your query:
SELECT DISTINCT
*,
CASE
WHEN Col1 = 'C1' THEN 0
WHEN Col1 = 'B1' THEN 1
WHEN Col1 = 'X1' THEN 2
WHEN Col1 = 'A1' THEN 3
WHEN Col1 = 'D1' THEN 4
ELSE 5
END AS SortColumn
FROM tbl1
ORDER BY SortColumn

Answer 3

(extension of Jeroen response)
If your "key" values are dynamic, you normally should have a separate table with the order of the keys...
declare #sort table (id varchar(10), ord int)
insert into #sort
values ( 'C1', 1 )
, ( 'X1', 2 )
, ( 'B1', 3 )
, ( 'A1', 4 )
, ( 'D1', 5 )
then you join/subquery on that table to calculate the SortColumn
-- join
SELECT DISTINCT
t.*,
s.ord SortColumn
FROM #tbl1 t
LEFT JOIN #sort s ON S.id = t.col1
ORDER BY SortColumn
-- subquery
SELECT DISTINCT
t.*,
(SELECT s.ord FROM #sort s WHERE s.id = t.col1) SortColumn
FROM #tbl1 t
ORDER BY SortColumn

Answer 4

declare #tbl1 table ( col1 varchar(32))
insert into #tbl1
values ( 'Basic Salary' )
, ( 'HRA' )
, ( 'OtherAllowance' )
, ( 'PF' )
, ( 'Basic Salary' )
, ( 'HRA' )
, ( 'Other Allowance' )
, ( 'PF' )
, ( 'ESIC' )
, ( 'Basic Salary' )
, ( 'HRA' )
, ( 'Other Allowance' )
, ( 'PF' )
, ( 'Basic Salary' )
select distinct col1
from #tbl1
order by col1