Reputation: 4173
I have a query that returns a lot of rows. I would need the difference between the values in the last row and the first row. TimeCol
is a DateTime
column, all other columns are of type float
.
My current (simplified) query is this:
select
timeCol, TotSoap_W1, TotSoap_W2, TotEnergy
from
TotEnergy
where
TimeCol >= '2019-03-01 00:00:00' and timeCol < '2019-03-02 00:00:00'
order by
TimeCol asc
Upvotes: 0
Views: 2042
Reputation: 131374
You can use the FIRST_VALUE and LAST_VALUE analytic functions :
select top 1 LAST_VALUE(TotSoap_W1) OVER (ORDER BY TimeCol ROWS
BETWEEN UNBOUNDEDPRECEDING AND UNBOUNDED FOLLOWING)
- FIRST_VALUE(TotSoap_W1)
from TotEnergy
where TimeCol >= '2019-03-01 00:00:00' and timeCol < '2019-03-02 00:00:00'
order by TimeCol asc
LAST_VALUE
returnes the last value in each group/result set according to the sorting order. In this query, it would be the same as returning each row's TotSoap_W1
value. To make it return the last value in all the results, we need to add the ROWS BETWEEN UNBOUNDEDPRECEDING AND UNBOUNDED FOLLOWING
clause
Upvotes: 1
Reputation: 809
This might not be the best way of doing but should do the job:
SELECT B.TotSoap_W1-A.TotSoap_W1 as TotSoap_W1, B.TotSoap_W2-A.TotSoap_W2 as TotSoap_W2, B.TotEnergy-A.TotEnergy as TotEnergy
FROM
(select TOP(1) timeCol, TotSoap_W1, TotSoap_W2, TotEnergy
from TotEnergy
where TimeCol >= '2019-03-01 00:00:00' and timeCol < '2019-03-02 00:00:00'
order by TimeCol ASC) A,
(select TOP(1) timeCol, TotSoap_W1, TotSoap_W2, TotEnergy
from TotEnergy
where TimeCol >= '2019-03-01 00:00:00' and timeCol < '2019-03-02 00:00:00'
order by TimeCol DESC
) B
Upvotes: 1