"Philip Crotwell" <crotwell@seis.sc.edu> writes:
> It seems to me that square root and cube root should be the same in return
> type.
They are, or at least one form of them matches:
regression=# \df sqrt
List of functions
Schema | Name | Result data type | Argument data types
------------+------+------------------+---------------------
pg_catalog | sqrt | double precision | double precision
pg_catalog | sqrt | numeric | numeric
(2 rows)
regression=# \df cbrt
List of functions
Schema | Name | Result data type | Argument data types
------------+------+------------------+---------------------
pg_catalog | cbrt | double precision | double precision
(1 row)
I think your request really amounts to "we should require every numeric
function to have a float8 counterpart and vice versa". Life's a bit too
short for that, though.
regards, tom lane