bug#46469: [External] : bug#46469: 27.1; `isearch-del-char' should move point further back

[Eli Zaretskii <eliz@gnu.org>]

> From: Augusto Stoffel <arstoffel@gmail.com>
> Cc: Drew Adams <drew.adams@oracle.com>,  Juri Linkov <juri@linkov.net>,
>  46469@debbugs.gnu.org
> Date: Sun, 14 Feb 2021 08:18:23 +0100
> 
> Re the idea that we can't change the behavior of an old command: if
> taken too seriously, this principle would imply that the standard Emacs
> UI can never improve

I didn't say that behavior of old commands cannot change, I said it
cannot change _unconditionally_.  That is, users should have a way to
revert to the old behavior.  It is generally preferable to make any
change in behavior opt-in; but even if we decide to make it the
default, there should be a way to go back.  That is because Emacs is a
very old and stable program, with users who are used to its present
behavior, and we try very hard not to break things that were working
for many years, nor change them in incompatible ways, since such
changes tend rightfully annoy veteran users.

In this particular case, whatever new aspects of the behavior you
suggest to add, they should be conditioned on a user option, and users
should be told in NEWS about the new behavior and the way to go back
to the old one (if the new one is made the default).

> Also, `isearch-del-char' changed from one obscure key to another
> obscure key in Emacs 27.  So clearly things can change.

It is IMO unfortunate that so many changes are happening lately in the
Isearch area regarding key bindings.  But that doesn't mean we should
make more changes there too lightly, let alone make them
unconditional.  At least keybindings can be easily undone on the user
level.

> Re this being a personal preference: I wouldn't bother to send a patch
> if I thought so.

"Personal preference" should not be interpreted to mean you are the
only person to prefer that.  Rather, it means that some people may
want that and some won't.  At least two people in this thread opined
that the existing behavior is fine, so clearly at least some people
would like to have the existing behavior.  Which doesn't mean what you
suggest doesn't have its place, it just means it isn't the only
opinion among Emacs users.

Where some users prefer one behavior and some prefer another, a user
option to control which behavior actually happens is a good way to
allow everyone to have what they want at a price of flipping a
variable.

> Is there a case where the current behavior is much more convenient
> and/or takes the search to a state that can't be easily reproduced by
> the patched version?

It is more convenient to me because I'm used to it: when I want to
remove the last character from the search string, I type DEL twice
without even looking.  I'm guessing there are others who have the same
habits.