Re: Locales with legacy codepages

[Рысь <lynx@sibserver.ru>]

17 сентября 2014 г. 0:43:46 KRAT, Rich Felker <dalias@libc.org> пишет:
>On Tue, Sep 16, 2014 at 11:23:35AM +0800, Рысь wrote:
>> Hello,
>> 
>> I am running a server which needs to speak with connected clients on
>> a legacy codepage (in particular, cp1251, widely used by Russian
>> people on Windows), and it has built-in gettext translation which
>> can be activated only by setting a proper locale.
>> However I don't know how musl supports locales with legacy
>> codepages, if even. I don't fully understand how such translations
>> work, so if there any ways to make legacy codepage locales work?
>> 
>> On previous glibc system I started such server with
>> LC_ALL=ru_RU.CP1251 environment variable and appropriate locale was
>> defined with localedef.
>> 
>> I have an option to convert all outgoing locale messages with iconv
>> if musl does not have such support.
>
>Using iconv is the the way to do this with musl. The only encoding
>supported as the native locale encoding is UTF-8 and there are no
>plans to change this. Since musl's iconv is not optimized for
>conversion _to_ legacy codesets, if you find this is a performance
>bottleneck, you might do better just to skip iconv and use your own
>conversion function that's specific to cp1251.
>
>Rich

Ok, I understand that my case is from those that should not happen in modern environments like systems based on musl libc. I will try iconv then.
Thanks!