"Sergey E. Koposov" <math@sai.msu.ru> writes:
> I see the very strange behaviour with the following set of queries:

> wsdb=# select na,nb, na::double precision as da, nb::double precision as db from ( select random()::numeric as na,random()::numeric as nb from generate_series(1,2)) as xx;

The planner "flattens" this query to a single level of SELECT, so what
you effectively have is

select random()::numeric as na,
random()::numeric as nb,
random()::numeric::double precision as da,
random()::numeric::double precision as db
from generate_series(1,2) as xx;

There's been some talk about prohibiting flattening if there are any
volatile functions in the subselect's targetlist, but nothing's been
done about that. Disabling flattening is a sufficiently big hit to
the planner's optimization ability that it shouldn't be done lightly.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match