Usage of goto

Question

OK, I have a situation where inside a loop, if I do not find data for a particular record, skip that and go to next record in the loop, so i tried goto
create table example(id number, val number);
insert into example values (1,4);
insert into example values (2,null);
insert into example values (3,5);
commit;
create or replace procedure p_example
is
type t_ref_cur is ref cursor;
type tn is record(id number, val number);
post_rec tn;
post_cur t_ref_cur;
v_sql varchar2(4000);
v_res number;
begin
v_sql := 'select id, val from example order by id';
open post_cur for v_sql;
loop
   fetch post_cur into post_rec;
   exit when post_cur%notfound;
   begin
   select val
   into v_res
   from example
   where id = post_rec.id;
   exception
     when no_data_found then
       dbms_output.put_line('in nodatafound');
       goto myend;
   end;
    dbms_output.put_line('post_rec.id: '||post_rec.id||' post_rec.val: '||post_rec.val);
<<myend>>
null;
end loop;
end p_example;Idea is, we get a no_data_found when it comes to id 2, and so we have to skip 2 from going any further
and go to next id (that is 3),
but looks like my goto is not working here, 2 is not skipped, i can print 2 also in the print statement

Answer 1

As already mentioned, goto is a very easy construct to abuse. Personally I don't like it however it does occasionally have a place. Essentially it provides the same type of jumping capability as an exception block. However I think an exception block is clearer, more recognisable and more explicit - very few people use gotos in PL/SQL in my experience, but exception blocks are much more common place. So one alternative would be to raise a custom exception and capture that i.e.
create or replace procedure p_example
is
type t_ref_cur is ref cursor;
type tn is record(id number, val number);
post_rec tn;
post_cur t_ref_cur;
v_sql varchar2(4000);
v_res number;
    lex_MyNoData        EXCEPTION;
    PRAGMA EXCEPTION_INIT(lex_MyNoData, -20001);
begin
v_sql := '  select
                e1.id,
                e1.val
            from
                example e1
            left outer join
                example e2
            on 
                ( e1.id ';
open post_cur for v_sql;
loop
    BEGIN
       fetch post_cur into post_rec;
       exit when post_cur%notfound;
       begin
       select val
       into v_res
       from example
       where id = post_rec.id;
       if v_res is null then
          goto myend;
       end if;
       exception
         when no_data_found then
           RAISE_APPLICATION_ERROR(-20001,'Res is null);
       end;
        dbms_output.put_line('post_rec.id: '||post_rec.id||' post_rec.val: '||post_rec.val);
    EXCEPTION
        WHEN lex_MyNoData THEN
            NULL;
    END;
end loop;
end p_example;However, I've not yet come across a circumstance where it was not possible to reorganise the logic of a unit of code to avoid having gotos or something similar. If you've got an existing code base that you're maintaining, you might not have the choice of refactoring it to the extent necessary, but in those case I would make what you're doing as explicit as possible.
David