encoding again - Mailing list pgsql-general

Hi, sorry that this email is a little bit long, but it is actully not :-))


**** I have a database 'unidb' created with -E UNICODE.

$ psql -l
        List of databases
   Name    |  Owner  | Encoding
-----------+---------+-----------
 unidb     | kathy   | UNICODE


**** I input Chinese data in unicode form. E.g.
logging-threshold=\u65e5\u5fd7\u9608\u503c
polling_setting_error=\u8bbe\u7f6e\u8f6e\u8be2\u95f4\u9694\u65f6\u51fa\u9519

unidb=# show client_encoding;
NOTICE:  Current client encoding is 'UNICODE'
SHOW VARIABLE

unidb=# select * from testbytes;
          name           |          value
-------------------------+-------------------------
 logging_setting_error   | 设置æ¥å¿éå¼æ¶åºé
 polling_setting_error   | 设置轮询é´éæ¶åºé


**** When I retrieve data, I did

unidb=# set client_encoding to 'EUC_CN';
unidb=# show client_encoding;
NOTICE:  Current client encoding is 'EUC_CN'
SHOW VARIABLE

unidb=# select * from testbytes order by value;
          name           |          value
-------------------------+-------------------------
 logging_setting_error   | ־ֵʱ
 polling_setting_error   | ѯʱ


Three problems here:
1) the sorting is based on unicode value, not EUC_CN encoding value.
2) I wrote the ResultSet to a file by using OutputStreamWriter(file, "EUC_CN"). The
file is not readable from the browser with any charset setting.
3) Changing client_encoding from UNICODE to EUC_CN actually alter/loose the data if
you compare the above "select *" statements.

I wonder why this happens ?? According to the doc, automatic encoding coversion
between UNICODE and EIC_CN is supported.

Any help is highly appreciated.
thanks,
kathy