bug#7925: 23.2.91; report-emacs-bug doesn't like curly quotes

[Eli Zaretskii <eliz@gnu.org>]

> From: Chong Yidong <cyd@stupidchicken.com>
> Date: Sat, 29 Jan 2011 18:34:19 -0500
> Cc: 7925@debbugs.gnu.org
> 
> Reuben Thomas <rrt@sc3d.org> writes:
> 
> > Pretty much every time I use report-emacs-bug, I get the message
> > “Convert non-ASCII letters to hexadecimal?” There are two problems here:
> >
> > 1. From looking at the code in report-emacs-bug-hook, it really means
> > “non-ASCII characters”, as it looks for any non-7-bit character, not
> > just non-7-bit letters.
> >
> > 2. In this age of UTF-8, and, for example, my use of smart-quotes-mode,
> > what do you have against non-ASCII characters? Perhaps you’d like to try
> > to make sure that mail to report-emacs-bug is in English, but for that
> > you should be using some form of linguistic analysis.
> 
> Maybe this question dates from the days when email was ASCII only.

Actually, this was added in Emacs 20 or 21, when Emacs got support for
multi-lingual editing.  the reason was to make sure bug reports are
written in English and not in some other languages that maintainers
could not command.  Also, various MUAs, even in Emacs, might convert
non-ASCII characters into illegible mess.

> At the least, this check ought to only be applied to the mail
> headers (MIME is needed for non-ASCII in headers).

Actually, I think the body is the more important part.

If we are willing to allow non-ASCII characters, we should at least
limit that to some subset that does not affect the text itself.
Linguistic analysis seems to be a bit of overkill (this command should
be reliable even in a partially sick Emacs session, remember?).  Maybe
some kind of list of allowed non-ASCII characters will be good enough.

(Not that I understand why it is a problem to use "..." or some such,
when one reports an Emacs bug.  But that's me.)