Logging the erronous ID but not breaking the LOOP - oracle

Question

Here is my problem. I am looping through some values and some of those values raise an exception. I want to log those values, but the program flow should not break. I mean, if I encounter such a value, I will simply log the error and skip to the next value.
here is the simplified version :
drop table test;
--destination Table
create table test
(
id varchar2(2)
);
-- Error log table
create table test_log
(
id varchar2(10)
);
DECLARE
l_num NUMBER;
BEGIN
FOR c IN 90..102
LOOP
INSERT INTO test
VALUES (c);
l_num:=c;
END LOOP;
EXCEPTION
WHEN OTHERS THEN
dbms_output.put_line(l_num);
INSERT INTO test_log
VALUES (l_num);
COMMIT;
--raise;
END;
/
My problem is, when it's encountering an error, it simply jumps to the exception section and not looping through the later values in the loop.

Answer 1

You can catch the exception in an inner block:
DECLARE
l_num NUMBER;
BEGIN
FOR c IN 90..102
LOOP
l_num:=c;
BEGIN -- inner block
INSERT INTO test
VALUES (c);
EXCEPTION -- in inner block
WHEN OTHERS THEN
dbms_output.put_line(l_num);
INSERT INTO test_log
VALUES (l_num);
END; -- inner block
END LOOP;
END;
/
The loop won't be interrupted if an exception occurs; only that single insert is affected. Note that you don't really want to commit inside the exception handler as that will commit all the successful inserts so far, not just the error. (If you wanted to log the errors but later roll back you could use an autonomous procedure to do the logging, but that's a separate discussion). And you don't want to re-raise the exception as that would still break the loop and the whole outer anonymous block.
Catching 'others' is generally not a good idea; if you have known errors you could encounter - like a bad data or number format - it's better to catch those explicitly. If you have a wider problem like not being able to extend a data file then the insert inside the exception handler would presumably fail anyway though.
You don't really need l_num any more as c is still in-scope for the inner exception handler, so you could simplify slightly to:
BEGIN
FOR c IN 90..102
LOOP
BEGIN -- inner block
INSERT INTO test
VALUES (c);
EXCEPTION -- in inner block
WHEN OTHERS THEN
dbms_output.put_line(c);
INSERT INTO test_log
VALUES (c);
END; -- inner block
END LOOP;
END;
/

Answer 2

You can also use FORALL.
That answer to your request with SAVE EXCEPTION statement because the FORALL loop to continue even if some DML operations fail. And it will be more efficient than a simple LOOP.
Ex :
DECLARE
CURSOR LCUR$VAL IS
SELECT ID
FROM test1;
TYPE LT$TAB IS TABLE OF TEST%ROWTYPE;
LA$TAB_TEST LT$TAB;
dml_errors EXCEPTION;
PRAGMA EXCEPTION_INIT(dml_errors, -24381);
LC$ERRORS NUMBER(11);
LC$ERRNO NUMBER(11);
LC$MSG VARCHAR2(4000 CHAR);
LC$IDX NUMBER(11);
BEGIN
OPEN LCUR$VAL;
LOOP
FETCH LCUR$VAL BULK COLLECT INTO LA$TAB_TEST LIMIT 1000;
BEGIN
FORALL x IN 1 .. LA$TAB_TEST.COUNT SAVE EXCEPTIONS
INSERT INTO test VALUES LA$TAB_TEST(x);
EXIT WHEN LCUR$VAL%NOTFOUND;
EXCEPTION
WHEN DML_ERRORS THEN
LC$ERRORS := sql%BULK_EXCEPTIONS.COUNT;
FOR idx IN 1 .. LC$ERRORS
LOOP
LC$ERRNO := sql%BULK_EXCEPTIONS (idx).ERROR_CODE;
LC$MSG := sqlerrm(-LC$ERRNO);
LC$IDX := sql%BULK_EXCEPTIONS(idx).error_index;
-- here you can log in table : test_log...
END LOOP;
END;
END LOOP;
CLOSE LCUR$VAL;
END;
Hoping that it can help.