Operator "=" not unicode-safe?

PostgreSQL version: 8.0.3
Operating system: Linux (SuSE 9.1)

I have a UNICODE database, trying to compare two unicode strings (Ethiopic
characters). Client encoding is also UNICODE:
===================================================
testdb=> select 'በድሩ ሁሴን'='ሰይፉ ከበደ';
?column?
----------
t
(1 row)

Clearly, it can be seen that they are not equal. The "LIKE" operator also
seems to think so:

testdb=> select 'በድሩ ሁሴን' LIKE 'ሰይፉ ከበደ';
?column?
----------
f
(1 row)
===================================================

What is the problem here?
The behavior is the same with SQL_ASCII databases and the SQL_ASCII client
encoding.

Of course one could always overload the operator or just use LIKE. But
where it really matters is with queries using UNION, EXCEPT or INTERSECT:

==========================

testdb=> select a from a;
a
---------
በድሩ ሁሴን
ሰይፉ ከበደ
(2 rows)

testdb=> select a from b;
a
---------
ሰይፉ ከበደ
(1 row)

testdb=> select a from a union select a from b;
a
---------
በድሩ ሁሴን
(1 row)

testdb=> select a from a except select a from b;
a
---
(0 rows)

testdb=> select a from a intersect select a from b;
a
---------
በድሩ ሁሴን
(1 row)
==========================

What can I do?
With kind regards,

Jörg Haustein