id,summary,reporter,owner,description,type,status,priority,milestone,component,version,resolution,keywords,cc,os,architecture,failure,difficulty,testcase,blockedby,blocking,related
4080,Use libcharset instead of nl_langinfo(CODESET) if possible.,PHO,igloo,"`nl_langinfo(CODESET)` doesn't always return standardized variations
of encoding names which GNU libiconv understands.

This problem actually affects (at least) NetBSD and OpenBSD:
`GHC.IO.Encoding.Iconv.localeEncoding` suffers from this and then even
`ghc --version` fails. Here is an example:

{{{
/* test1.c */
#include <stdio.h>
#include <locale.h>
#include <langinfo.h>

int main() {
    setlocale(LC_ALL, """");
    printf(""nl_langinfo(CODESET) = \""%s\""\n"", nl_langinfo(CODESET));
    return 0;
}
}}}

{{{
% gcc -o test1 test1.c
% LC_ALL=ja_JP.UTF-8 ./test1
nl_langinfo(CODESET) = ""UTF-8""   // Good.
% iconv -f UTF-8 -t UTF-8 /dev/null && echo ok
ok
% LC_ALL=C ./test1
nl_langinfo(CODESET) = ""646""     // Wtf? You mean ISO 646?
% iconv -f 646 -t UTF-8 /dev/null && echo ok
iconv: conversion from 646 unsupported
iconv: try 'iconv -l' to get the list of supported encodings
% uname -a
NetBSD netbsd 5.99.20 NetBSD 5.99.20 (ADJUSTED) #0: Mon Oct  5 15:05:08 JST 2009
  root@netbsd:/usr/obj/sys/arch/i386/compile/ADJUSTED i386
%
}}}

So we should use libcharset if possible, which is shipped together
with GNU libiconv. See:
http://www.haible.de/bruno/packages-libcharset.html

{{{
/* test2.c */
#include <stdio.h>
#include <locale.h>
#include <libcharset.h>

int main() {
    setlocale(LC_ALL, """");
    printf(""locale_charset() = \""%s\""\n"", locale_charset());
    return 0;
}
}}}

{{{
% gcc -o test2 test2.c -I/usr/pkg/include -L/usr/pkg/lib -lcharset
% LC_ALL=ja_JP.UTF-8 ./test2
locale_charset() = ""UTF-8""    // Good.
% LC_ALL=C ./test2
locale_charset() = ""ASCII""    // Good!
% iconv -f ASCII -t UTF-8 /dev/null && echo ok
ok
%
}}}
",bug,closed,high,7.0.1,libraries/base,6.13,fixed,iconv locale,,Unknown/Multiple,Unknown/Multiple,Runtime crash,,,,,