Pipe nested object type - plsql

Question

I am just trying to achieve pipe row nested type. There are tons of examples around but none that I am able to apply.
My types are:
create type t1_row as object ( a1 number, a2 varchar2(10) );
create type t1_tab as table of t1_row;
create type t2_row as object ( b1 number, b2 varchar2(10), b3 t1_tab );
create type t2_tab as table of t2_row;
I've tried to create a function in so many ways, but none of them are able to compile successfully.
One example:
create or replace function fn (r in number) return t2_row pipelined is
l_row1 t1_tab;
l_row2 t2_tab;
begin
for i in 1..r loop
for j in 1..r loop
l_row1(j).a1 := j;
l_row1(j).a2 := 'a2 ' || j;
end loop;
l_row2(i) := (i,l_row1);
PIPE ROW (l_row2);
end loop;
return;
end;
This code produces the following errors:
[Error] PLS-00630 (1: 12): PLS-00630: pipelined functions must have a supported collection return type
[Error] PLS-00382 (10: 22): PLS-00382: expression is of wrong type
Any help advice or any similar example can be useful.
Version: Oracle 11g Release 11.2.0.1.0

Answer 1

Your construction for t2_row was missing the 2nd parameter, and you are
returning the wrong type. Try this:
create or replace function fn (r in number) return t2_tab pipelined is
l_row1 t1_tab := t1_tab();
l_row2 t2_tab := t2_tab();
begin
for i in 1..r loop
for j in 1..r loop
l_row1.extend(1);
l_row1(j) := t1_row(j,'a2 ' || j);
end loop;
l_row2.extend(1);
l_row2(i) := t2_row(i,'TEST',l_row1);
PIPE ROW (l_row2(i));
end loop;
return;
end;
Also, read this tutorial on how to use pl/sql object collections.

Answer 2

Your code has two syntax errors and a logic error.
The first syntax error is that the function's return type should be a collection not a row type, so
return t2_tab pipelined
The second syntax error is that you need to include the type when instantiating an object, and the number of arguments must match the signature of the type. So the outer loop assignment should be:
l_row2(i) := t2_row(i, 'T2', l_row1);
The logic error is that we don't need to maintain a collection variable for the output. We just need a row variable.
Also the indexed counts seem a bit confused, so my code may differ from your intention.
create or replace function fn (r in number)
return t2_tab pipelined
is
l_tab1 t1_tab;
l_row2 t2_row;
begin
for i in 1..r loop
l_tab1 := new t1_tab();
l_tab1.extend(r);
for j in 1..r loop
l_tab1(j) := t1_row(j*i, 'a2 ' || j);
end loop;
l_row2 := t2_row(i, 'T2', l_tab1);
PIPE ROW (l_row2);
end loop;
return;
end;
/
Here is the run:
SQL> select * from table(fn(3));
B1 B2 B3(A1, A2)
----- --- ---------------------------------------------------------------
1 T2 T1_TAB(T1_ROW(1, 'a2 1'), T1_ROW(2, 'a2 2'), T1_ROW(3, 'a2 3'))
2 T2 T1_TAB(T1_ROW(2, 'a2 1'), T1_ROW(4, 'a2 2'), T1_ROW(6, 'a2 3'))
3 T2 T1_TAB(T1_ROW(3, 'a2 1'), T1_ROW(6, 'a2 2'), T1_ROW(9, 'a2 3'))
SQL>