strpos() function ( internal text_postion()) had a bug in unicode database.
dsn=> select id,subject, strpos(subject, ' ') from bd_22 where id = 3927;
id | subject | strpos
------+-------------+--------
3927 | 안녕하세요~ | 0
(1 row)
Time: 1.619 ms
dsn=> select id,subject, strpos(subject, ' ') from bd_22 where id between 3925 and 3927;
id | subject | strpos
------+-----------------------------------------+--------
3925 | 대구에 DB 스터디 관련한...곳..없는가요? | 4
3927 | 안녕하세요~ | 11
(2 rows)
Time: 2.490 ms
----
Sorry, above text is korean language.
strpos returns wrong result.
----- Original Message -----
From: "Tom Lane" <tgl@sss.pgh.pa.us>
To: "Korea PostgreSQL Users' Group" <pgsql-kr@postgresql.or.kr>
Cc: <pgsql-patches@postgresql.org>; "Tatsuo Ishii" <t-ishii@sra.co.jp>
Sent: Saturday, January 31, 2004 12:31 AM
Subject: Re: [PATCHES] v7.4.1 text_position() patch
> Hm. I don't think it can actually fail, because the wchar strings are
> zero-terminated. But it does look like there's a missed speedup here.
> (Tatsuo, do you agree?)
>
> Thanks for the report!
>
> regards, tom lane
>