Hi,
I'm experiencing some problems while ordering Turkish characters.
Here is that I mean:
Let's say we have two records:
Onder <- There are two dots on O. (ASCII CODE: 153)
Ozan
In Turkish alphabet, O (dotless) comes before O (with dots). So, Ozan
should be listed before Onder, but it's listed after Onder; since
PostgreSQL thinks that those letters are the same...
To begin with:
operdb=# SELECT version(); version
---------------------------------------------------------------------------------------------------------PostgreSQL
7.3.2on i686-pc-linux-gnu, compiled by GCC gcc (GCC) 3.2
20020903 (Red Hat Linux 8.0 3.2-7)
(installed from rpm)
Here is an simple example: Lines 1, 10,11,12 have O with dots as the first
letter.
==========SELECT oper_uzun_adi FROM operler ORDER BY oper_uzun_adi;
...
Ömer KORKMAZOner BINBASOnur BAŞOnur TURHANOsman Selçuk SARIOĞLUOsman TOPANOzan GÜLENOzan SARIOzge YAZICIOGLUÖzgür
ÇAKICIÖzgürÖZDEMİRÖzlem BAL
...
==========
But if I apply a select query and use LIKE, PostgreSQL gives me correct
solutions:
operdb=# SELECT oper_uzun_adi FROM operler WHERE oper_uzun_adi ILIKE 'O%'; oper_uzun_adi
------------------------------------------Osman TOPANOner BINBASOzge YAZICIOGLUOnur BAŞOnur TURHANOzan GÜLENOsman
SelçukSARIOĞLUOzan SARI
(8 rows)
Now for O with dots:
operdb=# SELECT oper_uzun_adi FROM operler WHERE oper_uzun_adi ILIKE 'Ö%'; oper_uzun_adi
------------------------------------------Özgür ÇAKICIÖzlem BALÖmer KORKMAZÖzgür ÖZDEMİR
So, LIKE understands that they are different chars; but if I simply order
them; I do not get the correct result.
Here is the locale setting:
# Locale settings
#
# (initialized by initdb -- may be changed)
LC_MESSAGES = 'en_US'
LC_MONETARY = 'en_US'
LC_NUMERIC = 'en_US'
LC_TIME = 'en_US'
If I change it to tr_TR, nothing changes.
Could anyone help me?
Best regards,
--
Devrim GUNDUZ
devrim@gunduz.org devrim.gunduz@linux.org.tr http://www.gunduz.org
