Using CAST/COLLECT with a function

Question

I have a select statement that gathers comments into a report by user ID and Project Name.
This works really well until I get to a point where the total comments exceeds 4000 characters, then I get a "character string buffer too small".
Is there another way to gather the comments and report on them in APEX?
Here is the Select Statement:
/** START SELECT
SELECT
PROJECT_NAME, USER_ID, COLLECT_FUNC(CAST(COLLECT (USER_NOTES) AS VARCHAR2_T)) as COMMENTS
FROM
TBL_PROJECT
where PROJ_DATE >= to_date('5/1/2011','MM/DD/YYYY') and PROJ_DATE <= to_date('6/30/2011','MM/DD/YYYY')
PROJECT_NAME, USER_ID
order by USER_ID;
/** END SELECT
Where the VARCHAR2_T is:
CREATE OR REPLACE TYPE "VARCHAR2_T" as table of varchar2(4000)
and the function that collects the comments is:
/*** START CODE
create or replace function collect_func (t in varchar2_t)
return varchar2
as
ret varchar(4000) := '';
i number;
begin
i := t.first;
while i is not null loop
if ret is not null then
ret := ret || ' -- <BR> ';
end if;
ret := ret || t(i);
i := t.next(i);
end loop;
return ret;
end;
/*** END CODE

Answer 1

SleepDeprivedInSeattle wrote:
I created a workspace at http://apex.oracle.com
The user name is: TEST_USER
Password is: myPassword1!You need to tell us the workspace name as well.
If you go to the SQL Workshop, SQL Commands window and try to execute:
SELECT
PROJECT_NAME, USER_ID, COLLECT_FUNC(CAST(COLLECT (COMMENTS_FLD) AS VARCHAR2_T)) as COMMENTS
FROM
TBL_PROJECT
group by PROJECT_NAME, USER_ID
order by USER_ID;
It will work for 10 row. But if you select 1000 rows, it fails because one of the projects has more than 4000 characters in the COMMENTS_FLD field. I expanded the VARCHAR to 6000 and it still fails.The VARCHAR2maximum size of <tt>VARCHAR2</tt> value that can be returned in Oracle SQL is 4000 bytes.
To return more than 4000 bytes, modify <tt>collect_func</tt> to return a CLOB, or use a different string aggregation technique. I normally (ab)use XMLDB for this as it:
<li>allows for either VARCHAR2 or CLOB output
<li>doesn't require any additional objects or privileges
<li>produces clean and properly structured XHTML without a lot of tedious and unreadable concatenations (my colleagues point out that it introduces a lot of tedious and unreadable XMLDB stuff instead)
SQL> select
  2            d.deptno
  3          , d.dname
  4          , xmlserialize(
  5             content
  6             xmlelement(
  7                 "ul"
  8               , xmlagg(
  9                  xmlelement("li", e.ename)
10                  order by e.ename
11                 )
12             )
13             as clob
14             indent size=2
15            ) employees
16  from
17            dept d
18           join emp e on d.deptno = e.deptno
19  group by
20            d.deptno
21          , d.dname
22  order by
23            d.deptno
24*          , d.dname
SQL> /
    DEPTNO DNAME       EMPLOYEES
     10 ACCOUNTING       <ul>
                   <li>CLARK</li>
                   <li>KING</li>
                   <li>MILLER</li>
                 </ul>
     20 RESEARCH       <ul>
                   <li>ADAMS</li>
                   <li>JONES</li>
                   <li>SCOTT</li>
                   <li>SMITH</li>
                 </ul>
     30 SALES       <ul>
                   <li>ALLEN</li>
                   <li>BLAKE</li>
                   <li>JAMES</li>
                   <li>MARTIN</li>
                   <li>TURNER</li>
                   <li>WARD</li>
                 </ul>
     40 OPERATIONS       <ul>
                   <li>FORD</li>
                 </ul>Note that there is a 32K limit on the total size of a report row in APEX, including all data and mark-up.