Why does this use of accents in psql fail for a Unicode database?

[Frank van Vugt]

Hi,

I noticed that queries from one of my php-clients were failing when they
contained certain accented characters, but upon trying the same thing in
psql, it showed failure there as well. Strangely enough, at least one
character is working o.k.

Could someone enlighten my why....?

(using SuSE v9.0)

db=# select version();
version
------------------------------------------------------------------------
PostgreSQL 7.4.1 on i586-pc-linux-gnu, compiled by GCC gcc (GCC) 3.2.3
(1 row)

db=# \l
List of databases
Name | Owner | Encoding
---------------+----------+----------
db | vugtf | UNICODE
<cut>
(6 rows)

This database was created using '--no-locale' as initdb-parameter
The postgres.conf contains

db=# \encoding
UNICODE

db=#create table f_test(value text);
CREATE TABLE

megafox=# insert into f_test values ('¤');
INSERT 2518547 1

db=# insert into f_test values ('ç');
db'#
<ctrl>+<c>

db=# insert into f_test values ('õ');
db'#
<ctrl>+<c>

db=# insert into f_test values ('é');
db'#
<ctrl>+<c>

megafox=# insert into f_test values ('één');
ERROR: invalid byte sequence for encoding "UNICODE": 0xe9e96e
=> the insert that succeeds is that of a Euro-sign

=> all accented characters including the Euro are created using a composekey

=> I guess in the last insert, the two hidden '-characters work as a pair....
Any hints on what's causing this behaviour are welcomed..... ;-)


--
Best,

Frank.
---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster