Re: Documenting composite LC_ALL locale, anyone?

On Sun, Feb 01, 2004 at 02:49:13PM +0200, Ruslan Ermilov wrote:
> One option to deal with this complex case (LC_ALL, composite locale,
> and when some of the components start with a `/') would be to replace
> each single slash with two adjacent slashes, so an argument of
> `/pathto/localedef/output' would become `//pathto//localedef//output'.
> Then double slashes in the input would represent a single slash, and
> single shash would represent a component separator.

//xxx///yyy//zzz is /xxx /yyy/zzz or /xxx/ yyy/zzz ?

> Unfortunately, POSIX-200x is silent on this case while requiring that
> the output of ``setlocale(LC_ALL, NULL)'' should be such that when
> it's fed back to ``setlocale(LC_ALL, ...)'' it should restore all of
> the locale components.  Perhaps we should seek for some clarifications
> in the Austin Group?  ;)

This is implementation-defined. It does not matter to standartize it 
because it can't be modified, only saved/restored.

-- 
Andrey Chernov | http://ache.pp.ru/

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (FreeBSD)

iQCVAwUBQBz5HeJgpPLZnQjrAQFI0gQA5MlqmflcNheTplLl48v+pvmwKM42Zz3y
mLfanuqp6u+18Iqw8bgunGTXSZc10f9kTb1bOuAwLo4xFh6Kz0nSIJJbaeSIpWZf
KWkwjTJhh5+pJbuHJHSTrBKE9yLLHspbR9g3Jn7H+NA0cID9Pcqjoc7Sn/6GnPJ/
EGppU+oDIy0=
=hHKA
-----END PGP SIGNATURE-----