Problem in executing Stored Procedure from Trigger

Question

Hi,
Case1.
I am executing a stored procedure form a trigger. The stored procedure is not executing fully.
Case 2.
But when when I execute Stored Procedure alone it is executing.
CREATE OR REPLACE TRIGGER mhubadmin.call_proc_ratesheet_new
after INSERT OR delete ON mhubadmin.pvt_br_ratesheet
FOR EACH ROW
declare
var_orgid number;
begin
if inserting then
select org_child_id into var_orgid from business_relationship where br_id=:new.br_id;
mhubadmin.proc_ratesheet_new(var_orgid);
else
select org_child_id into var_orgid from business_relationship where br_id=:old.br_id;
mhubadmin.proc_ratesheet_new(var_orgid);
end if;
end;
CREATE OR REPLACE PROCEDURE proc_ratesheet_new(var_orgid in number)
IS
cursor c3 is select distinct(purs.user_id) from hubuser hu
inner join PVT_USER_RATESHEET purs on hu.USER_ID=purs.USER_ID
               where hu.ORG_ID=var_orgid;
cursor c1(varUser_id number) is select purs.user_id,purs.ratesheet_id from hubuser hu
inner join PVT_USER_RATESHEET purs on hu.USER_ID=purs.USER_ID
               where hu.ORG_ID=var_orgid and purs.USER_ID=varUser_id;
cursor c2(varUser_id number) is select hu.user_id,pbr.ratesheet_id from HUBUSER hu
                         inner join BUSINESS_RELATIONSHIP br on hu.ORG_ID=br.ORG_CHILD_ID
                         inner join PVT_BR_RATESHEET pbr on br.BR_ID=pbr.BR_ID
                              where hu.user_id in(select distinct(purs.USER_ID) from hubuser hu
                                   inner join PVT_USER_RATESHEET purs
                                                                 on hu.USER_ID=purs.USER_ID
                                             where hu.ORG_ID=var_orgid) and hu.user_id=varUser_id;
foundFlag boolean;
incrFound integer;
insertFound integer;
deleteFound integer;
str varchar2(4000);
BEGIN
     incrFound:=0;
     insertFound:=0;
     for c3var in c3 loop
for c2var in c2(c3var.user_id) loop
          insert into test values ('Step3:'||c2var.user_id);
          foundFlag:=false;
          for c1var in c1(c3var.user_id) loop
               if c2var.ratesheet_id=c1var.ratesheet_id then
                    foundFlag:=true;
                         incrFound:=incrFound+1;
                         exit;
                    end if;
               end loop;
               if foundFlag=False then
               --insert into pvt_user_ratesheet (username_ratesheet_id,user_id,ratesheet_id) values (SEQ_USER_RATESHEET.nextval,c3var.user_id,c2var.ratesheet_id);
               dbms_output.put_line('Inserted for user :'||c3var.user_id||' is - ratesheet :'||c2var.ratesheet_id);
               insertFound:=insertFound+1;
               end if;
          end loop;
end loop;
     commit;
     incrFound:=0;
     deleteFound:=0;
     for c3var in c3 loop
          for c1var in c1(c3var.user_id) loop
          foundFlag:=false;
          for c2var in c2(c3var.user_id) loop
                    if c1var.ratesheet_id=c2var.ratesheet_id then
                    foundFlag:=true;
                    incrFound:=incrFound+1;
                    exit;
                    end if;
               end loop;
               if foundFlag=False then
               --delete from pvt_user_ratesheet where user_id=c3var.user_id and ratesheet_id=c1var.ratesheet_id;
                    --dbms_output.put_line('Deleted for userid:'||c3var.user_id||' for ratesheet :'||c1var.ratesheet_id);
                    deleteFound:=deleteFound+1;
               end if;
          end loop;
end loop;
     commit;
EXCEPTION
WHEN OTHERS THEN
DBMS_OUTPUT.put_line (SQLCODE||' '|| SQLERRM);
END;
Regards,
Mathew

Answer 1

In general, I would suggest that you only catch errors that you can handle reasonably. If you know that a SELECT INTO might return 0 rows for example, handle the NO_DATA_FOUND exception. Having a WHEN OTHERS, particularly without re-raising the exception, only serves to hide the source of the problem. I would much rather that my code fail quickly and explicitly than hope that I see an error message in the output and that other code doesn't start failing because the system wasn't left in an appropriate state.
As Mark suggests, you may choose to implement a comprehensive logging framework using autonomous transactions, in which case a WHEN OTHERS would be reasonable, but you would still want to propagate exceptions you cannot handle.
If you see a WHEN OTHERS clause and there is no RAISE, you probably have an error waiting to happen.
Justin