Обсуждение: Re: French Characters

At 18:11 26.05.2003, Al-Karim Bhamani (LCL) said:
--------------------[snip]--------------------
>I created a database using UNICODE Characterset.
>
>Created a table
>
>CREATE TABLE a (
>
>  a char(30)
>
>);
>
>Tried to insert a row through PG_ADMIN
>insert into a values('qualité')
>
>This resulted in the following error
>
>Number: -2147467259
>Description: ERORR: Invalid UNICODE character sequence found (0xe90000)
>
>I believe It's because of the French character "é" .
>
>We need to insert French characters in the database.
>
>The DATABASE is UNICODE
>
>Is there anything which I am missing.
--------------------[snip]--------------------

You need to set the client to a matching character set if you're using
UNICODE as database character set.

E.g., using psql, upon connecting to a UNICODE database, psql automatically
uses the unicode caracter set - there's no 'é' character in UNICODE, as
this will be encoded using 2 bytes.

Example PSQL session:

-- create and connect to UNICODE database
template1=# create database "dem" with ENCODING='utf8';
CREATE DATABASE
template1=# \c demo
You are now connected to database demo.
-- check the current client encoding
demo=# \encoding
UNICODE
-- switch to an encoding we can use via keyboard
demo=# \encoding latin1
demo=# \encoding
LATIN1

-- create a test table
demo=# create table test (text varchar);
CREATE
demo=# insert into test values ('é');
INSERT 17033860 1
demo=# select * from test;
 text
------
 é
(1 row)

-- now check this with UNICODE encoding
demo=# \encoding unicode
demo=# select * from test;
 text
------
 é
(1 row)

demo=#

PostgreSQL will automatically convert between different client/server
character sets, provided the client side is correctly initialized using an
encoding it supports.

These conversions from/to UNICODE are available as to the docs:
EUC_JP, SJIS, EUC_KR, UHC, JOHAB, EUC_CN, GBK, EUC_TW, BIG5, LATIN1 to
LATIN10, ISO_8859_5, ISO_8859_6, ISO_8859_7, ISO_8859_8, WIN, ALT, KOI8,
WIN1256, TCVN, WIN874, GB18030, WIN1250

Hope this helps,

--
   >O     Ernest E. Vogelsinger
   (\)    ICQ #13394035
    ^     http://www.vogelsinger.at/

On Mon, 26 May 2003, Ernest E Vogelsinger wrote:

> You need to set the client to a matching character set if you're using
> UNICODE as database character set.
>
> E.g., using psql, upon connecting to a UNICODE database, psql automatically
> uses the unicode caracter set - there's no 'é' character in UNICODE, as
> this will be encoded using 2 bytes.

There's something I'm not understanding.  Unicode does contain the
accented e character - it's character 0x00e9.  I thought the whole point
of unicode was to avoid having to specify character sets and have the
whole thing "just work" but it would seem that's not the case.  If a app
is created and an installable ISO containing that app is built and
downloaded by users in China, Zimbabwe, Egypt, Brazil and France, each
will require one of these client encodings be selected.  How then, can the
app 'know' where it is and select the correct encoding?

Cheers!
Jon