Обсуждение: There is a doubt of fatal bug on PostgreSQL 8.0.4.

There is a doubt of fatal bug on PostgreSQL 8.0.4.

I found this problem when I developed a system 3 days ago.
I examine some version.
It has no problem on version 8.0.3 and on 8.1.0.
It occured only on 8.0.4 and 2 byte character.

Our development environment is
Server OS: Red Hat Enterprise Linux ES 3
DB : PostgreSQL 8.0.4
both encodes are EUC_JP

This is a test order.

1. Create the table for test.
CREATE TABLE TEST (
    ID    VARCHAR(1)
   ,NAME  VARCHAR(40)
   ,PRIMARY KEY (ID)
)

2. Make some records on the table.
INSERT INTO TEST VALUES ('1', '$B$"$"$"(B');
INSERT INTO TEST VALUES ('2', '$B$$$$$$(B');
INSERT INTO TEST VALUES ('3', '$B$"(B');

* Please make at least two records of the same number of characters and two
byte character.

3. Query this table by name.
SELECT * FROM TEST
WHERE NAME = '$B$"$"$"(B'

The results are '$B$"$"$"(B' and '$B$$$$$$(B'.
It should be only '$B$"$"$"(B'.

When you also query "WHERE NAME = '$B$&$&$&(B'", it'll reply the same result.
It should be no result.

Please examine and fix it.
I'm waiting for the version 8.0.5. Or I'll change version 8.1.0.

Sincerely.

_________________________________________________________________
MSN Hotmail $B$,B#$k(B Hot Xmas $B!_(B VOGUE
http://promotion.msn.co.jp/hotmail/xmas/index.html $BAGE($J%/%j%9%^%9%3%U%l$r(B
$B%W%l%<%s%H(B

"Kasahara Kenichi" <kenjpm90@hotmail.com> writes:
> Our development environment is
> Server OS: Red Hat Enterprise Linux ES 3
> DB : PostgreSQL 8.0.4
> both encodes are EUC_JP

What locale is the database using?  Strange comparison behavior is
fairly common if the locale expects a different encoding than you
are using...

            regards, tom lane

> What locale is the database using?  Strange comparison behavior is
> fairly common if the locale expects a different encoding than you
> are using...

locale is 'C'.
client_encoding is 'SJIS'.

Locale 'C' is the same settings when I used version 8.0.3.
I think this is right..Or should I appoint "no locale" when I initialize
database?

regards, Kenichi Kasahara

_________________________________________________________________
$BL5NA$G(B250MB$B$N<u?.%\%C%/%9$,;H$($k!V(BMSN Hotmail$B!W(B
http://promotion.msn.co.jp/hotmail/

"Kasahara Kenichi" <kenjpm90@hotmail.com> writes:
>> What locale is the database using?  Strange comparison behavior is
>> fairly common if the locale expects a different encoding than you
>> are using...

> locale is 'C'.

Hm, it should work OK with C locale.  I get apparently-correct output
on my 8.0 installation:

CREATE TABLE
INSERT 155246 1
INSERT 155247 1
INSERT 155248 1
 id |     name
----+--------------
 1  | $B$"$"$"(B
(1 row)

 id | name
----+------
(0 rows)

> client_encoding is 'SJIS'.

Um ... you said EUC_JP before.  But I get the same result either way.

            regards, tom lane