Hello Tom,
Thanks for the proposed patch. However, it is probably incomplete:
need_charset_alias=false is only possible if the libc has nl_langinfo(CODESET)
*and* it always returns the codesets normalized in the GNU way.
It is possible that some newer version of FreeBSD has nl_langinfo(CODESET).
Can you look up which was the first such version?
Then, a mapping from FreeBSD codeset names to GNU codeset names needs
to be installed, like it has been done in config.charset lines 387..412 for
NetBSD
and OpenBSD. To this effect
- check out the libiconv sources (from the git repository of project
'libiconv'
on savannah.gnu.org),
- read the README of the subdirectory libcharset/tools/,
- execute the "all-charsets" script in that directory,
- report the output here.
I think your patch was right for locales where no mapping is needed (probably
this includes the UTF-8 locales). However, while at it, let's do it right for
all locales.
Bruno
