Question

I have the SQL query:

select title, scale / next_scale, c
from ( select title, scale, scale*D0 AS c, 
              lead(scale) over(partition by title order by scale asc) as next_scale,
              row_number() over(partition by title order by scale asc) as agg_row
       from signatures
     ) agg
where agg_row = 1;

and it works as expected. However, what I really want the sorting “scale” value to be an arithmetic operation between several columns, so I tried using an AS clause (shown above) and modify the query to:

select title, scale / next_scale, c
from ( select title, scale, scale*D0 AS c, 
              lead(scale) over(partition by title order by c asc) as next_scale,
              row_number() over(partition by title order by c asc) as agg_row
       from signatures
     ) agg
where agg_row = 1;

However, it fails at the ORDER BY c. Why is this? I can substitute ORDER BY scale*D0 and it works just fine. However, I will eventually want to use a term like: scale*D0*D1*D2*...*D100; and I don’t want to have to calculate that 3 different times – not to mention the physical length of the query. I am hoping to have scale*D0*D1*D2*...*D100 AS c and then ORDER BY c.

Is this possible?

I am using PostgreSQL.

Many thanks,
Brett

Answer 1

Calculate c in a subquery:

select title, scale / next_scale, c
from ( select title, scale, c, 
              lead(scale) over(partition by title order by c asc) as next_scale,
              row_number() over(partition by title order by c asc) as agg_row
       from (select title, scale, scale * D0 AS c from signatures) signatures_calc
     ) agg
where agg_row = 1;