Re: RFE: Please allow unicode ID chars in identifiers

[L A Walsh]

tetsujin@scope-eye.net wrote:
> This is also why I think this should be an optional "encoding marker"
>   
---
   Why?  If it was the current encoding, it wouldn't have high-bits
set.  If it had any high bits set, it's fairly simple to
either presume or validate the script as UTF-8, as it is
self-synchronizing.

   Please note, the RFE was for UTF-8.  That's 1 encoding.
I'm not proposing, suggesting nor wanting a general "locale"
support for arbitrary encodings in bash.  That would be
entirely chaotic and its own "Tower of Babel" incident on
its own.  This suggestion was to allow 1 global/extra character
encoding: UTF-8. 

Regarding what Greg said:
> From: "Greg Wooledge" <wooledg@eeg.ccf.org>
>  The main issue here is that the author's locale may not *exist* on
> the
>  user's machine. There may not be any way for bash to determine which
>  non-ASCII characters constitute "letters" in the author's locale.
>   
---
   If we use UTF-8, that won't be an issue.  It's an internationalized
encoding that can support all languages and locales.  If a script
uses high-bit encoding(s), it can automatically be assumed to be UTF-8.

   In a related way, the program "file" already is able to tell
you what encoding a file is in.