Thread: pgAdmin2 Japanese display

Dear all,

Do you think it would be possible to display Japanese data in pgAdmin2 if we
could recode UFT-8 (Server-side) into some Windows-compliant encoding
(client-side) using SET CLIENT_ENCODING = WIN_JAP

Are there some (undocumented) Windows Japanese encodings available in
PostgreSQL? Would it work if some WIN_JAP encoding was added to PostgreSQL?

Cheers,
Jean-Michel POURE

> -----Original Message-----
> From: Jean-Michel POURE [mailto:jm.poure@freesurf.fr]
>
> Dear all,
>
> Do you think it would be possible to display Japanese data in
> pgAdmin2 if we
> could recode UFT-8 (Server-side) into some Windows-compliant encoding
> (client-side) using SET CLIENT_ENCODING = WIN_JAP
>
> Are there some (undocumented) Windows Japanese encodings available in
> PostgreSQL?

SET CLIENT_ENCODING to 'SJIS'
Strictly speaking, you need a multibyte psqlodbc driver
to handle SJIS(Shift Jis) properly.

regards,
Hiroshi Inoue

Le Lundi 25 Février 2002 17:19, Hiroshi Inoue a écrit :
> SET CLIENT_ENCODING to 'SJIS'
> Strictly speaking, you need a multibyte psqlodbc driver
> to handle SJIS(Shift Jis) properly.

Dear Hiroshi,

Thank you for your answer. Could you tell us more about SJIS:
1) Is it Microsoft encoding standard for Japanese?
2) Is the multibyte psqlodbc driver a separate driver? What are the
settings to switch odbc to SJIS?

Do you think it would be usefull to convert UTF-8 into UCS-2 using:
SET CLIENT_ENCODING to 'UCS-2'.

This would allow us to test pgAdmin2 with full Unicode support without
changing the odbc drivers. Right?

Best regards,
Jean-Michel POURE

Jean-Michel POURE allegedly said:
> Le Lundi 25 Février 2002 17:19, Hiroshi Inoue a écrit :
>> SET CLIENT_ENCODING to 'SJIS'
>> Strictly speaking, you need a multibyte psqlodbc driver
>> to handle SJIS(Shift Jis) properly.
>
> 2) Is the multibyte psqlodbc driver a separate driver? What are the
> settings to switch odbc to SJIS?

Yes it is - it's not the one we distribute from odbc.postgresql.org or
with pgAdmin.

Hiroshi, will the multibyte version work OK with Latin encodings? I
suppose what I'm asking is, should we distribute the multibyte version by
default or would it break things for people that are not using multibyte
encodings?

Regards, Dave.

Jean-Michel POURE wrote:
>
> Le Lundi 25 F騅rier 2002 17:19, Hiroshi Inoue a 馗rit :
> > SET CLIENT_ENCODING to 'SJIS'
> > Strictly speaking, you need a multibyte psqlodbc driver
> > to handle SJIS(Shift Jis) properly.
>
> Dear Hiroshi,
>
> Thank you for your answer. Could you tell us more about SJIS:
> 1) Is it Microsoft encoding standard for Japanese?

Yes. You can see SJIS Japanese strings if you set the locale
*Japanese*.

> 2) Is the multibyte psqlodbc driver a separate driver?

Yes and no. You could build the multibyte driver with
the same source but the different compile option.

> What are the
> settings to switch odbc to SJIS?

Please set the connect_settings DSN option as
 set client_encoding to 'SJIS'.
The commands are issued immediately after the connection
was established and you can see SJIS(converted from EUC_JP,
UTF-8 etc) Japanese strings. Note that you can see the strings
properly in most cases even though you don't use the multibyte
driver. However SJIS could have ASCII(e.g. \, { ) byte as
the second byte of some multibyte characters and they sometimes
confuses non-multibyte psqlodbc driver. It occurs with neither
EUC_XX nor UTF-8 because every byte of their multibyte characters
is non-ASCII.

> Do you think it would be usefull to convert UTF-8 into UCS-2 using:
> SET CLIENT_ENCODING to 'UCS-2'.

The *SET CLIENT_ENCODING .. * command is to request the
server to convert every server side text string(server encoding)
to the specified client_encoding when retrieving the text strings
at the client side. However currently(maybe in the future also)
PostgreSQL doesn't support any UCS-2 <-> .. conversion.
The main reason is that all currently supported encodings contain
ASCII as a part of themselves and they are always teminated by
NULL('\0') but UCS-2 satisfies neither.
To support Unicode(UCS-2) in our ODBC driver we would
have to implement the APIs for Unicode like
 SQLColAttributesW, SQLColumnPrivilegesW, SQLColumnsW,
 SQLConnectW, SQLDataSourcesW, SQLDescribeColW,
 SQLDriverConnectW, SQLErrorW, SQLExecDirectW,
 SQLForeignKeysW, SQLGetConnectOptionW,
 SQLGetCursorNameW, SQLGetInfoW, SQLNativeSqlW,
 SQLPrepareW, SQLPrimaryKeysW, SQLProcedureColumnsW,
 SQLProceduresW, SQLSetConnectOptionW, SQLSetCursorNameW,
 SQLSpecialColumnsW, SQLStatisticsW, SQLTablesW,
 SQLTablePrivilegesW
and handle SQL_WCHAR and SQL_C_WCHAR at various(not
clear now) places using UCS-2 <-> UTF-8 conversion.

regards,
Hiroshi Inoue

Dave Page wrote:
>
> Jean-Michel POURE allegedly said:
> > Le Lundi 25 F騅rier 2002 17:19, Hiroshi Inoue a 馗rit :
> >> SET CLIENT_ENCODING to 'SJIS'
> >> Strictly speaking, you need a multibyte psqlodbc driver
> >> to handle SJIS(Shift Jis) properly.
> >
> > 2) Is the multibyte psqlodbc driver a separate driver? What are the
> > settings to switch odbc to SJIS?
>
> Yes it is - it's not the one we distribute from odbc.postgresql.org or
> with pgAdmin.
>
> Hiroshi, will the multibyte version work OK with Latin encodings? I
> suppose what I'm asking is, should we distribute the multibyte version by
> default or would it break things for people that are not using multibyte
> encodings?

According to Eiji Tokuya who introduced the multibyte support,
the multibyte version has no problem with single byte encodings.

regards,
Hiroshi Inoue