Re: ielm automatic saving of history -- bug 67000

[Eli Zaretskii <eliz@gnu.org>]

> Date: Thu, 31 Oct 2024 16:12:05 +0530 (IST)
> Cc: emacs-devel@gnu.org
> From: Madhu <enometh@meer.net>
> 
> *  Eli Zaretskii  <86jze7q4io.fsf@gnu.org>
> Wrote on Wed, 16 Oct 2024 21:36:15 +0300
> >> the ielm-history file from which the comint-input-ring is initalized
> >
> > That's not what bothers me.  What bothers me is the way we save the
> > history of the ielm command the first time, when they are not taken
> > from a file, but from the user input.
> >
> >> Here I am prompted with the coding system warning . If i go ahead
> >> and qchoose utf-8-unix, it goes ahead and saves it. I can compare
> >> 2.txt.binary and 2.txt.out to see that utf-8-unix has not resulted
> >> in any data loss.  Hence all the claims in my previous message.
> >
> > The data loss I had in mind is not in this case.
> 
> I still think there is some fishy behaviour when raw-text-unix is used
> by as the initial (unspecified) encoding.  The message which the
> prompt presents (that utf-8-unix cannot be used to save etc.) is
> surely incorrect, since it goes ahead and saves it anyway as
> utf-8-unix.

The explanation presented when Emacs requests a safe encoding was
worded in a way that could be interpreted to mean the text cannot
possibly be encoded bu utf-8.  The reality is subtly different: utf-8
cannot encode that _safely_, but it can encode it with some data loss
if forced to do so.  I have modified the wording to say "safely
encode", to make that part more explicit.

> BTW did you have a chance to consider the patch I mailed on
> oct-17-2024? -- Regards, madhu

No, I hoped Simen will get to that at some point.