Database Software

Hi!

My environment is like the following:
- oracle 11.2.0.2 EE
- CHARACTERSET: AL32UTF8

TESTCASE PREPARE:

CREATE TABLE THU
(
C1 NUMBER,
C2 VARCHAR2(5 CHAR),
C3 VARCHAR2(10 CHAR),
C4 NUMBER
);

REM INSERTING into THU
SET DEFINE OFF;
Insert into THU (C1,C2,C3,C4) values ('1','qq','qq test qq','1');
Insert into THU (C1,C2,C3,C4) values ('2','éáő','qq éáő qq','2');
Insert into THU (C1,C2,C3,C4) values ('3','őúó','qq ÓÚŰÁ qq','3');
Insert into THU (C1,C2,C3,C4) values ('4','őóüö','qq ŐÚŰ qq','4');
Insert into THU (C1,C2,C3,C4) values ('5','ÜÖü','qq ŰŰŰŰ qq','5');
commit;


TESTCASE 1:
========

export NLS_LANG=HUNGARIAN_HUNGARY.EE8ISO8859P2

sqlplus hu/hu

set linesize 2000

select * from thu;

SQL> r
1* select * from thu

C1 C2 C3 C4
---------- ----- ---------- ----------
1 qq qq test qq 1
2 éáő qq éáő qq 2
3 őúó qq ÓÚŰÁ qq 3
4 őóüö qq ŐÚŰ qq 4
5 ÜÖü qq ŰŰŰŰ qq 5


TESTCASE 2:
========

export NLS_LANG=HUNGARIAN_HUNGARY.AL32UTF8

sqlplus hu/hu

set linesize 2000

select * from thu;

SQL> select * from thu;

C1 C2 C3 C4
---------- -------------------- ---------------------------------------- ----------
1 qq qq test qq 1
2 éáő qq éáő qq 2
3 őúó qq ÓÚŰÁ qq 3
4 őóüö qq ŐÚŰ qq 4
5 ÜÖü qq ŰŰŰŰ qq 5


As you can see sqlplus fills up each character with 4 bytes in second testcase. This is a problem with spooled flat files which get processed for example with sql loader (fixed positioned). So is the only way for dealing with flat files to set NLS_LANG to an one byte setting? Generally we've the recommendation to use always AL32UTF8 for NLS...

Has anyone any input for my problem?

Thanks
Markus