bug#66394: 29.1; Make register-read-with-preview more useful

[Eshel Yaron]

Thierry Volpiatto <thievol@posteo.net> writes:

> Eshel Yaron <me@eshelyaron.com> writes:
>
>> So it'd be great to have the previous behavior available in Emacs 30.
>
> Sorry but I wont write this, it is not complicated to write but needs
> works and attention and I spent enough time on this.
>
> The only thing you mentionned I agree with is the necessity now to use C-q
> to insert key sequence (note that C-n/p will insert it alone), but it is
> not a big annoyance right? (most people don't use this, I don't for
> one).  For the preview buffer not visible, note that you can pop to it
> at any moment with C-h.

What about the fact that `C-x r s` and friends by default no longer work
in the minibuffer?

> Perhaps you will get used to the new behavior after some time,

Why can't I, and other users, have the previous behavior, though?  It's
great to innovate with new alternatives, but why should we break user
workflows in the process, without as much as a NEWS entry to warn them?

> otherwise it is easy to revert completely my commits (it is the
> development branch of emacs after all).

Seeing as you are not willing to make this change backward compatible, I
think that would make sense.  I don't have commit rights to emacs.git,
so I can't do that myself, though.

I do think it shouldn't be that hard to extend the previous
implementation to _optionally_ ask for confirmation before overwriting
register contents, without using the minibuffer.  That way we'd have the
new behavior that you want to introduce without the added breakage.
Would you be willing to test such a patch if I write it?


Thanks,

Eshel