Query engine fails for DISTINCT if NLS_CHARACTERSET AL32UTF8 on 11.1.0.6.0

Question

The issue :
DISTINCT doesn't work for a parent child relationship if the total amount of chars, over which a distinct needs to be executed, exceeds 2000 chars.
Currently I have this issue on a Oracle 11.1.0.6.0 standard edition server.
Note this issue seems to be specific for NLS_CHARACTERSET AL32UTF8, the issue can't be reproduced if NLS_CHARACTERSET WE8MSWIN1252
The current settings :
select * from nls_database_parameters
PARAMETER VALUE
NLS_LANGUAGE AMERICAN
NLS_TERRITORY AMERICA
NLS_CURRENCY $
NLS_ISO_CURRENCY AMERICA
NLS_NUMERIC_CHARACTERS .,
NLS_CHARACTERSET AL32UTF8
NLS_CALENDAR GREGORIAN
NLS_DATE_FORMAT DD-MON-RR
NLS_DATE_LANGUAGE AMERICAN
NLS_SORT BINARY
NLS_TIME_FORMAT HH.MI.SSXFF AM
NLS_TIMESTAMP_FORMAT DD-MON-RR HH.MI.SSXFF AM
NLS_TIME_TZ_FORMAT HH.MI.SSXFF AM TZR
NLS_TIMESTAMP_TZ_FORMAT DD-MON-RR HH.MI.SSXFF AM TZR
NLS_DUAL_CURRENCY $
NLS_COMP BINARY
NLS_LENGTH_SEMANTICS BYTE
NLS_NCHAR_CONV_EXCP FALSE
NLS_NCHAR_CHARACTERSET AL16UTF16
NLS_RDBMS_VERSION 11.1.0.6.0
select * from v$version
BANNER
Oracle Database 11g Release 11.1.0.6.0 - Production
PL/SQL Release 11.1.0.6.0 - Production
CORE     11.1.0.6.0     Production
TNS for Linux: Version 11.1.0.6.0 - Production
NLSRTL Version 11.1.0.6.0 - Production
This is how I can reproduce the issue :
DROP TABLE aparent CASCADE CONSTRAINTS;
DROP TABLE achild CASCADE CONSTRAINTS;
CREATE TABLE aparent
id NUMBER(19) NOT NULL,
cola VARCHAR2(1000 CHAR),
colb VARCHAR2(500 CHAR),
colc VARCHAR2(50 CHAR)
CREATE TABLE achild
id NUMBER(19) NOT NULL,
aparent number
insert into aparent (id,cola,colb,colc) values ( 1,'P1','A','A');
insert into achild (id,aparent) values ( 1,1);
insert into achild (id,aparent) values ( 2,1);
insert into achild (id,aparent) values ( 3,1);
insert into aparent (id,cola,colb,colc) values ( 2,'P2','A','A');
insert into achild (id,aparent) values ( 4,2);
insert into achild (id,aparent) values ( 5,2);
insert into achild (id,aparent) values ( 6,2);
commit;
select distinct
p.cola, -- 1000
p.colb, -- 500
p.colc, -- 50
p.colc, -- 50 1600 char
p.colc, -- 50
p.colc, -- 50 1700 char
p.colc, -- 50
p.colc, -- 50 1800 char
p.colc, -- 50
p.colc, -- 50 1900 char
p.colc, -- 50
p.colc, -- 50 2000 char
p.colc -- 50 *2050 char*
from aparent p
, achild c
where c.aparent = p.id
and p.cola = 'P1'
This will result in 3 rows
select distinct
p.cola, -- 1000
p.colb, -- 500
p.colc, -- 50
p.colc, -- 50 1600 char
p.colc, -- 50
p.colc, -- 50 1700 char
p.colc, -- 50
p.colc, -- 50 1800 char
p.colc, -- 50
p.colc, -- 50 1900 char
p.colc, -- 50
p.colc, -- 50 *2000 char*
from aparent p
, achild c
where c.aparent = p.id
and p.cola = 'P1'
This will result in 1 row
The difference between both queries is that total amount of chars, over which a distinct needs to be executed, exceeds 2000 chars in the first situation.
Any idea on what the cause of this issue could be?

Answer 1

Hi,
From 11g docs:
Restrictions on DISTINCT and UNIQUE Queries ...
      When you specify DISTINCT or UNIQUE,
      the total number of bytes in all select list expressions is limited to the size of a data block minus some overhead.
      This size is specified by the initialization parameter DB_BLOCK_SIZE.So, the restriction is concerned with number of bytes.
AL32UTF8 is a multi-byte characterset, WE8MSWIN1252 is a single-byte characterset.
This must be the difference(?), probably the same thing in other versions as well.
Regards
Peter