Re: [bug-gnu-libiconv] Solaris libiconv vs. GNU libiconv in terms of 646

[Bruno Haible]

Dagobert Michelsen wrote:
> But why has this work been done for aix and osf1 then?

It has not. What you see in encodings_aix.def and encodings_osf1.def are the
support for encodings that exist only on these platforms, not for aliases that
are platform specific.

> However, patching dozens of softwares seems more work to
> me than providing an encoding_solaris.def. Would you
> accept a patch providing the solaris encodings?

Hmm. Looking at the source code of zsh, it appears that it uses the
result of nl_langinfo(CODESET) only to pass it to iconv_open. (They don't
store it in configuration files, pass it to X11 or over the network, use it
in GUI menus etc.) This is correct.

Also you are right about the "dozens of software". Googling for "libiconv 646"
there are many hits relating to Solaris, NetBSD, OpenBSD, Samba. NetBSD and
OpenBSD should fix their systems. Samba should not make platform specific
encodings visible across systems. So the remaining problematic case are
these OSes: AIX, HP-UX, IRIX, OSF/1, Solaris.

I'm adding support to libiconv to recognize these platform specific aliases
*on these platforms only*.

Bruno