Aggregate functions are used to collapse values from a grouping of rows into a
single result value. This can be done with a whole table, as is common with
the aggregate function
count(*), or it
can be done with groupings of rows from a
BY clause, as is common with something like
sum(). Aggregate functions are used to summarize, or
aggregate, all of the individual row values into some single representative
SQL aggregate functions are created using the same
that is used to create scalar functions (See Scalar Functions). When defining a scalar function,
you pass in a C function pointer in the sixth parameter and set the
seventh and eighth parameter to NULL. When defining an aggregate
function, the sixth parameter is set to NULL (the scalar function
pointer) and the seventh and eighth parameters are used to pass in two
C function pointers.
The first C function is a “step” function. It is called once for each row in an aggregate group. It acts similarly to an scalar function, except that it does not return a result (it may return an error, however).
The second C function is a “finalize” function. Once all the SQL rows have been stepped over, the finalize function is called to compute and set the final result. The finalize function doesn’t take any SQL parameters, but it is responsible for setting the result value.
The two C functions work together to implement the SQL aggregate function. Consider ...