Similar --- but very different. Consider this annotated example. The solution, if your model supports it, is to put a UNIQUE constraint on mst_emp_master(ms_emp_no).

ops$tkyte@8i> create table swa_emp_proc( sw_emp_no char(10) not null );
Table created.

ops$tkyte@8i> create table mst_emp_master( ms_emp_no char(10) not null );
Table created.

ops$tkyte@8i> insert into swa_emp_proc values ( 1 );
ops$tkyte@8i> insert into mst_emp_master values ( 1 );
ops$tkyte@8i> insert into mst_emp_master values ( 1 );
ops$tkyte@8i> commit;
Commit complete.



Here are your views:

ops$tkyte@8i> create or replace view SWA_EMP_PROC_JOIN as select a.* from SWA_EMP_PROC
2 a,mst_emp_master b where a.SW_EMP_NO = b.ms_emp_no
3 /
View created.

ops$tkyte@8i> CREATE OR REPLACE VIEW SWA_EMP_PROC_IN AS SELECT * FROM SWA_EMP_PROC
2 WHERE SW_EMP_NO IN (SELECT MS_EMP_NO FROM MST_EMP_MASTER)
3 /
View created.



ops$tkyte@8i> REM this shows they can be different in the data they return:

ops$tkyte@8i> select * from SWA_EMP_PROC_JOIN;

SW_EMP_NO
----------
1
1

ops$tkyte@8i> select * from SWA_EMP_PROC_IN;

SW_EMP_NO
----------
1

ops$tkyte@8i> REM this shows you cannot delete from the JOIN view since the
ops$tkyte@8i> REM rows in swa_emp_proc are 'multiplied' -- you get a row back
ops$tkyte@8i> REM for each row in mst_emp_master
ops$tkyte@8i> REM It is ambigous how the delete should proceed on this.
ops$tkyte@8i> REM consider "delete form swa_emp_proc_join WHERE ROWNUM = 1"
ops$tkyte@8i> REM that would in effect delete 2 rows -- but it cannot -- its not defined
ops$tkyte@8i> REM so we don't permit it to happen:


ops$tkyte@8i> delete from SWA_EMP_PROC_JOIN;
delete from SWA_EMP_PROC_JOIN
*
ERROR at line 1:
ORA-01752: cannot delete from view without exactly one key-preserved table


but we can delete from the IN since the rows in A are not 'multiplied'

ops$tkyte@8i> delete from SWA_EMP_PROC_IN;
1 row deleted.

ops$tkyte@8i> rollback;
Rollback complete.

Lets get rid of the one row that messes this up:

ops$tkyte@8i> delete from mst_emp_master where rownum = 1;
1 row deleted.

and add a unique constraint:

ops$tkyte@8i> alter table mst_emp_master add constraint ms_emp_no_unique unique(ms_emp_no);
Table altered.

ops$tkyte@8i> REM Now, we have the constraint that makes it so that the rows in
ops$tkyte@8i> REM swa_emp_proc are *not* multiplied by rows in mst_emp_master.
ops$tkyte@8i> REM now we have a key preserved join and can delete from it
ops$tkyte@8i>
ops$tkyte@8i> delete from SWA_EMP_PROC_JOIN;

1 row deleted.

Hope this clears up the confusion.