Good practice question

Question

there is a procedure p(i in number, o out number) that checks some data and finishes this way if some criteria are met:
procedure p(i in number, o out number) is
select count(*) into v_counter
from ... ;
if v_counter > 0 then
  o := 123;
  return;
end if;
exception
  when others then
  o := 99999;
end;is this a good practice to finish this way (ie. return) or some other ways would be better in terms of "good practice" (like raising custom exception and handling it in exception block where out variable could be set up)?
I would appreciate your suggestions about good practices in scenario above.
thank you

Answer 1

943276 wrote:
there is a procedure p(i in number, o out number) that checks some data and finishes this way if some criteria are met:
..snipped..
is this a good practice to finish this way (ie. return) or some other ways would be better in terms of "good practice" (like raising custom exception and handling it in exception block where out variable could be set up)?I would call using out params as exception/error codes wrong. Never mind good or acceptable practise. It is just plain wrong. As wrong in PL/SQL as it would be wrong in Java, C#, Pascal or another language that implements exceptions.
Why? Because in such a language an exception is raised to indicate an error condition - an exception to the norm has occurred. If code suppresses the exception, that code is now saying that there is no error condition. It explicitly tells the caller that processing was successful and no exceptions to the norm have occurred.
Returning the error code as an out parameter is not acceptable - as the "contract" in such a language, between the caller and that code, states that errors are handled as exceptions. That is how the language is designed to be used. That is how the run-time of the language works.
Violating this fundamental concept in such a language makes absolutely no sense.