Re: Systematizing back navigation

[Dmitry Gutov <dmitry@gutov.dev>]

On 21/11/2023 08:57, Eli Zaretskii wrote:
> On November 21, 2023 8:35:26 AM GMT+02:00, Po Lu<luangruo@yahoo.com>  wrote:
>> Eli Zaretskii<eliz@gnu.org>  writes:
>>
>>> Perhaps you bumped into some
>>> specific situation, and then tried to generalize it too much.  In that
>>> case, I suggest to describe the original issue, not its
>>> generalizations.
>> I did describe three specific scenarios that I come across frequently:
>> Info, Help and Gnus Article buffers.  Besides those, there are also
>> pop-ups displayed by the likes of VC,
>> dired-mouse-find-file-other-window, and virtually every other caller of
>> display-buffer.  With these, I'd rather Back dismissed the new window
>> altogether than switch to whatever window previous-buffer selects.
> 
> IMO, it is entirely reasonable to have Back bound to a command that "goes back" where that makes sense.

Would it make sense to bind it to xref-go-back globally? With 
mode-specific overrides in place.

> Dismissing a window with Back is basically a smartphone thing, so we could bind Back to the function bound to ESC or C-g, but only on Android; doing that on other platforms would be inconsistent with the general behavior of those platforms, and should be avoided, IMO.

Dismissing the window on 'M-,' if it was just created by 
xref-find-definitions-other-window kind of makes sense to me. Though it 
might mess up the goals the user had set it up for.

OTOH, 'C-x 4 . M-,' currently creates two windows showing the same 
buffer side-by-side, and that's much more easily reached with 'C-x 3'.