What is the difference between BEGIN followed by IF and IF followed by BEGIN - oracle

Question

I am new to PL/SQL. What is the difference between the below two flows?
BEGIN followed by IF
BEGIN
IF flag= 1 THEN
FOR SOME_REC(id)
LOOP
END LOOP;
END IF;
END
IF followed by BEGIN
IF flag= 1 THEN
BEGIN
FOR SOME_REC(id)
LOOP
END LOOP;
END
END IF;

Answer 1

By its design PL/SQL is a block structured language with each block consisting of 4 components:
Declaration Section - optional:; DECLARE or Named Block (Function, Procedure, ...)
Execution Section - required: BEGIN
Exception Section - optional: EXCEPTION
End Statement - required: END
As #Mr.AF indicates each block defines the SCOPE of variables. Further these blocks can be nested. So a variable declared in an outer block is visible within that block and all blocks nested within. Subject to the rule that closest declaration wins and items declared in an nested bloc are not visible in outer blocks. See GURU99 for fuller description. Try the following.
declare
var varchar2(5) := '12345';
begin
dbms_output.put_line ( 'Outer block has variable ''var'' and has a length of '
|| to_char(length(var)) || ' and value of ' || var);
declare -- start nested block
var varchar2(8);
inner_only varchar2(8) := 'ShotTerm';
begin
var := '12345678';
dbms_output.put_line ( 'Inner block has variable ''var'' and has a length of '
|| to_char(length(var)) || ' and value of ' || var
|| ' and also variable inner_only with value ' || inner_only);
end ; -- end nested block
dbms_output.put_line ( 'Outer block variable ''var'' still has a length of '
|| to_char(length(var)) || ' and value of ' || var);
-- uncomment below to see difference
-- dbms_output_put_line('A reference to ' || inner_only || ' does not work; gives compile error');
end ; --outer block
For your specific scenarios #1 the IF and LOOP are being run in an outer (or only) block. While #2 the IF is run in the outer block while the loop runs in an inner block. Note: #2 must have a prior BEGIN and the END IF must be followed by an END - without them you will get a compile syntax error.

Answer 2

BEGIN ... END keywords implies starting point and end point of a code block like { } and compiler will validate local references (locality of references) within the block.otherwise, unexpected results may be happened.

Answer 3

The way you put it, without any further explanation of what problem you are trying to solve (if any), then there's no much difference in those two pieces of code. Both will do what they are supposed to do.
However, I'd like to say something about the 3rd case you never mentioned: BEGIN-EXCEPTION-END block within the loop, which does make the difference.
IF flag = 1 THEN
FOR SOME_REC(id) LOOP
BEGIN
do_something_here;
EXCEPTION
when some exception then
handle it
END;
END LOOP;
END IF;
What would be its purpose? In both of your examples, if any exception occurs within the loop, processing will be stopped and the procedure will terminate.
But, if there's an inner begin-exception-end block, it lets you capture an error within the loop, handle it the way you find the most appropriate (if there's nothing better, just log the error) and move on to the next loop cycle.
Doing so, you'll be able to process all records returned by the cursor. Some of them might fail, that's true, but you'll be able to proceed with the execution.
Here's an example based on Scott's EMP table; I'll try to divide SAL and NVL(COMM, 0) which will result in divide by zero error in some cases.
SQL> select ename, sal, nvl(comm, 0) comm from emp where deptno = 30 order by ename;
ENAME SAL COMM
---------- ---------- ----------
ALLEN 1600 300
BLAKE 2850 0 --> 2850/0 --> error
JAMES 950 0 --> error
MARTIN 1250 1400
TURNER 1500 0 --> error
WARD 1250 500
6 rows selected.
Code, as you wrote it:
SQL> set serveroutput on
SQL> declare
2 l_divided number;
3 begin
4 for cur_r in (select ename, sal, nvl(comm, 0) comm
5 from emp
6 where deptno = 30
7 order by ename
8 )
9 loop
10 l_divided := round(cur_r.sal / cur_r.comm);
11 dbms_output.put_line(cur_r.ename ||': ' || l_divided);
12 end loop;
13 end;
14 /
ALLEN: 5
declare
*
ERROR at line 1:
ORA-01476: divisor is equal to zero
ORA-06512: at line 10
SQL>
See? Only the first one, Allen, had valid data. The rest of them weren't even calculated, as soon as the first error appeared.
But, if you rewrite it as I suggested, then all of them are processed:
SQL> declare
2 l_divided number;
3 begin
4 for cur_r in (select ename, sal, nvl(comm, 0) comm
5 from emp
6 where deptno = 30
7 order by ename
8 )
9 loop
10 begin
11 l_divided := round(cur_r.sal / cur_r.comm);
12 dbms_output.put_line(cur_r.ename ||': ' || l_divided);
13 exception
14 when zero_divide then
15 dbms_output.put_line(cur_r.ename ||': COMM is equal to zero');
16 end;
17 end loop;
18 end;
19 /
ALLEN: 5
BLAKE: COMM is equal to zero
JAMES: COMM is equal to zero
MARTIN: 1
TURNER: COMM is equal to zero
WARD: 3
PL/SQL procedure successfully completed.
SQL>
See if the above makes any sense in what you asked.