Bug#345639: nl_langinfo(YESEXPR) ignores LANGUAGE, no apparent workaround

[Tollef Fog Heen <tfheen@debian.org>]

* Peter Samuelson 

| [Tollef Fog Heen]
| > Either that, or make sure YESSTR/NOSTR (from I18N::Langinfo) are
| > usefully populated and use those.  (They seemed generally not to be
| > based on the discussion on #debian-devel yesterday).
|
| It's not clear to me whether YESSTR/NOSTR are supposed to be
| single-letter abbreviations, or whole words, since the locales I have
| installed don't include them.  So if you want to print a localised
| version of [y/N] or [Y/n], you might have to truncate those strings, as
| well as convert to uppercase/lowercase.

I don't think that is very important, since it seems like yesexpr
should match yesstr.

| Both trunctation and case-conversion are quite non-trivial, unless
| nl_langinfo(CODESET) happens to be a single-byte, stateless character
| set (for which the 'char' datatype and the <ctype.h> functions can be
| used), like the ISO-8859 family.

True.

[...]

| This, too, makes assumptions about the character set.  It won't work
| too well for Korean or Russian, I expect.

I have no idea if you have something like a part of a word which is
enough to mean yes or no in CJKV languages (like «y» is in English).
If you don't have those, you'd need to write out the equivalent of
«yes» and not just «y».

-- 
Tollef Fog Heen                                                        ,''`.
UNIX is user friendly, it's just picky about who its friends are      : :' :
                                                                      `. `' 
                                                                        `-