Martijn van Oosterhout wrote:
> On Fri, Oct 06, 2006 at 12:44:43PM +0300, Martins Mihailovs wrote:
>> I would be a glad to hear your solutions, experience in web application
>> with multi languages (searching with indexing, sorting and others
>> problems with multi byte encoding).
>>
>> For developers: what are your future plans about UTF-8 in Postgres?
>
> Currnetly postgresql using the OS to do sorting, upper/lower
> conversions, etc. Which means you stuck with whatever your OS does,
> which may or may not be any good.
>
> The fix is easy, make postgresql use some kind of system independant
> locale database to do these transformations. We just need someone to
> write it....
>
There are some misunderstood. Im using Linux 2.6.16.4, postgresql 8.1.4,
(there are one of locale: lv_LV.utf8, for Latvian language). But if I
want do "lower", then with standard latin symbols all is ok, but with
others special symbols (like umlaut in Germany) there is problems, and
sorting is going not like alphabet but like latin alphabet and specials
symbols after. :(
For sorting there are solution function nls_to_string(), witch is using
C library and system locales for converting strings to binary. But it
dont work for search (like, tsearch2, e.c.)
I dont know what to do, but function lower() is not working for special
symbols (LATIN 7).
And I tried in windows - the same... (as I know windows knows how to
converts latvian symbols lower, upper... in UTF-8)...