ORA-06502 (character string buffer too small) on import application via script

Question

Running apex 4.2.2.00.11
oracle 10.2.0.5
I'm using a script to import my development application to a few production applications.
It worked great on apex 4.1.1, but now I get an 'ORA-06502 PL/SQL numeric of value error: character string buffer too small'
In the command window, I see that the script fails after it starts on the plugins.
My question: is this a bug in 4.2.2?
I could be a incompatible plugin, but how can I found out which plugin?
My script is quite simple:
In a command window I run
sqlplus SCHEMA/PASSWORD@SPIEU10 @_IMPORT_F102.sql
The script is
declare
l_workspace_id number;
begin
    -- determine the workspace id for the workspace FM in the target system
    select workspace_id into l_workspace_id from apex_workspaces where workspace = 'BY_APEX_SHARED';
    -- set the context for the target workspace
    apex_application_install.set_workspace_id( l_workspace_id );
    -- override the original application id 102
    apex_application_install.set_application_id ( 102 );
    -- set a different application name and alias, it should be unique within an APEX instance
    apex_application_install.set_application_alias( 'MY_ALIAS' );
    apex_application_install.set_application_name( 'MY_APP_NAME' );
    -- Use fixed offset
    apex_application_install.set_offset( p_offset => 0);
    -- override the original parsing schema FM with a different value
    apex_application_install.set_schema( 'BY_SCHEMA' );
end;
-- install the original application export, now the values for the workspace
@f102.sql
exit;
Error message from command window:
...ui types
...plugins
begin
ERROR at line 1:
ORA-06502: PL/SQL: numeric or value error: character string buffer too small
ORA-06512: at line 3
Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.5.0 - 64
bit Production
With the Partitioning, Data Mining and Real Application Testing options

Answer 1

ListAGG throws an ORA-01489 for varchar2 > 4000 bytes
OP has an ORA-06502 --> the IR is running into a 4k/32k limit.
quick solution: wrap USER_COMMENTS ListAGG in a substr( ...., 1, 4000)
Longer solution, use a subquery to modify the comments based on ROW_NUMBER() (ie after nth row, change it to NULL)
with
-- simulating data
t as (select task_id, sysdate - lv as date_entered
  ,round(dbms_random.value(1,24)) hours
  , '-*' || lv || '.' || task_id || '*-' as user_comments
from ( select level as task_id from dual connect by level <=10 ), (select level lv from dual connect by level < 1000)
-- modify data
modified_data as (
  select task_id, hours, date_entered
    ,case
      when row_number() over (partition by task_id order by date_entered desc) < 5
        then user_comments
      else null
     end USER_COMMENTS
    from t)
select task_id, sum(hours) total_hours,
  listagg( user_comments  ) within group (order by date_entered desc)
  || case when count(*) >= 5 then '! MORE COMMENTS !' else null end
    as user_comments
from modified_data
group by task_id;