Get value of a specific column of a row without primary key - sql

Question

This is purely out of curiosity.
create table test (ename varchar(50))
insert into test values ('abcd')
insert into test values ('pqrs')
insert into test values ('lmno')
insert into test values ('xxxx')
insert into test values ('tops')
I want the value of 3rd row from this table in a variable. i.e "lmno"
If I do this :
Declare #value varchar(50)
Select #value = ename from
(
select Row_number() over(order by ename) Rowno, * from test
) X where Rowno=3
print #value
I will get pqrs.
I cannot use this:
Declare #value varchar(50)
Select #value = ename from
(
select Row_number() over(order by 1) Rowno, * from test
) X where Rowno=3
because
Windowed functions do not support integer indices as ORDER BY clause
expressions.
Any options?
EDIT :
If I query it as
Select * from test
I do get records in the order in which they were inserted. That means somewhere there is a record as to how they were inserted. I just want to capture this sequence.

Answer 1

You are making a very very poor assumption about RDBMS's. The order that a RDBMS stores records, or they order they are written to the table is 100% absolutely inconsequential. It means nothing. It's arbitrary and you can't rely on it.
You will need to either add a new column to be the 'order' that you desire, or you will have to better define why you want pqrs in your recordset since 3rd record is meaningless in this sense.
To your edit: There is no record of the order which the records were inserted. There is an order by which the records are returned to the record set, and that they naturally lay in the DB's structure underneath, but it is arbitrary. The reason you get them back in the order in which they were written is because you have a tiny little table on a RDBMS that stores data in a single spot. This fails as soon as you scale your architecture up. You can not and should never ever rely on the order that your RDBMS retrieves records.

Answer 2

Let's look at it step by step.
select ename from test order by ename;
This orders by ename.
select ename from test order by 1;
Here 1 is an alias for the 1st element in your select clause, which is ename. So you order by ename again.
select Row_number() over(order by ename) Rowno, * from test
The row_number function works on records ordered by ename.
select Row_number() over(order by 1) Rowno, * from test
What is 1 supposed to mean here? We are inside an over clause and there is no first element the 1 could refer to. So it is not allowed to use a number here (it would only be confusing, as it could only mean a literal 1 for every record which doesn't order anything).
As to "I do get records in the order in which they were inserted. That means somewhere there is a record as to how they were inserted. I just want to capture this sequence.": No, that isn't the case. Right now you happen to get the records in the order they were inserted, but this is in no way guaranteed. The only way to guarantee an order is to have fields to represent the desired order and use them in ORDER BY.

Answer 3

Try This code
create table test (ename varchar(50))
insert into test values ('abcd')
insert into test values ('pqrs')
insert into test values ('lmno')
insert into test values ('xxxx')
insert into test values ('tops')
SELECT *FROM (
SELECT ROW_NUMBER() over (order by HH) AS RNO,ename FROM (SELECT ENAME,'' AS HH FROM test) T)T1
WHERE RNO=3

Answer 4

WITH MyCte AS
(
SELECT *, row_number() OVER(ORDER BY (SELECT 0)) ID FROM test
)
SELECT *
FROM MyCte
WHERE ID = 3

Answer 5

Try:
create table #test (ename varchar(50))
insert into #test values ('abcd')
insert into #test values ('pqrs')
insert into #test values ('lmno')
insert into #test values ('xxxx')
insert into #test values ('tops')
CREATE TABLE #Temp(RowID INT PRIMARY KEY IDENTITY(1,1), ename VARCHAR(10))
INSERT INTO #temp(ename)
SELECT ename from #test
SELECT T2.*
FROM #temp T1
JOIN #test T2 ON T1.ename = T2.ename
WHERE T1.RowID = 3