Re: The netsec thread

[Eli Zaretskii <eliz@gnu.org>]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: Jimmy Yuen Ho Wong <wyuenho@gmail.com>,  emacs-devel@gnu.org
> Date: Mon, 23 Jul 2018 14:46:44 +0200
> 
> > lisp/international/mule-cmds.el, look at
> > select-safe-coding-system-interactively, a subroutine of
> > select-safe-coding-system.  In a nutshell, it prepares and pops up a
> > buffer with special content, then prompts the user to select one of
> > several alternatives.
> 
> Hm...  I ran:
> 
> (select-safe-coding-system-interactively 1 3 (find-coding-systems-region 1 3) nil)
> 
> and it just popped up a buffer with some text and then called
> `read-coding-system', which is kinda close to what the NSM prompt is
> doing already.

That's not the best demonstration of what it does.  Try evaluating the
below in *scratch*, and then play with the links it creates to point
out characters that cannot be encoded:

  (find-file "try-select-safe-coding-system")
  (set-buffer-file-coding-system 'latin-1)
  (prefer-coding-system 'latin-1)
  (insert "text with smileys 😺 😇 😎")
  (save-buffer)

I didn't mean to say you must copy what it does, just to point out an
example where we ask a question, but let the user click or type RET on
active parts of the buffer to see additional information.  I also
don't see any particular reason to avoid modal questions, as the user
can always return to the minibuffer after looking at the additional
info, as in the above demonstration.

HTH