Re: iconv: WCHAR_T codeset

[Joerg Sonnenberger <joerg%bec.de@localhost>]

On Thu, Jul 21, 2016 at 06:57:52PM +0200, Thomas Klausner wrote:
> Hi!
> 
> GNU libiconv supports a codeset called "WCHAR_T".
> Some software seems to use it.
> 
> https://forum.filezilla-project.org/viewtopic.php?f=2&t=41458&p=147840#p147840
> 
> NetBSD and Solaris do not support this. Is this something we should support?
> 
> If not, is there a portable recommendation we can give such software?

It is something that is fundamentally impossible to implement without
enforcing a single internal encoding. In the Citrus model, the wchar_t
encoding is a property of the locale. GNU libc on the other hand always
uses UCS32. As such, it *can't* be correctly supported and software
should not use it. IMO it is quite pointless as well, the same effect
can be obtained by using iconv to convert to the charset of the desired
locale (see nl_langinfo) and then using mbstowcs or friends.

Joerg