Call for testing: OpenSSH 7.8

[Ingo Schwarze]

Hi Michael,

Michael wrote on Mon, Aug 20, 2018 at 08:32:03PM +0200:

> p.s. aka FYI: AIX also has the program locale(1), but that gives
> a lot less information, imho.

Sure it has, it's POSIX, and sure it provides less info than you
have shown from the non-standard tool.  But by definition, it is
unclear what the output of a non-standard tool may mean.

> And, maybe also a bit misleading - at least for the ignorant. I would
> have expected UTF-8, but instead, it uses EN_US and a symbolic link.
> 
> root at x064:[/data/prj/openbsd/mindrot/openssh-7.8.0.20]locale
> LANG=EN_US
> LC_COLLATE="EN_US"
> LC_CTYPE="EN_US"
> LC_MONETARY="EN_US"
> LC_NUMERIC="EN_US"
> LC_TIME="EN_US"
> LC_MESSAGES="EN_US"
> LC_ALL=
> 
> While the default language locale(1) is:
> 
> root at x066:[/data/prj/python/python3-3.8]locale
> LANG=en_US
> LC_COLLATE="en_US"
> LC_CTYPE="en_US"
> LC_MONETARY="en_US"
> LC_NUMERIC="en_US"
> LC_TIME="en_US"
> LC_MESSAGES="en_US"
> LC_ALL=

Seems quite unusual indeed and hard to understand - but AIX is free
to call their locales however they want.  Neither locale names nor
the return value from nl_langinfo(CODESET) are standardized.
According to POSIX, they need contain elements relating to languages,
territories, encodings, or even filenames, and the return value
from nl_langinfo(CODESET) need not match the LC_CTYPE variable.
AIX would be free to ask their users to specify "apples" or "oranges"
in LC_CTYPE and yet report back "bananas" from nl_langinfo(CODESET).
;-)

Yours,
  Ingo