bug#56662: 29.0.50; Funny region highlights when highlight-nonselected-windows is t

[Juri Linkov]

>> In any case, I don't think there's any reasonable action we can take.
>> "Active region" is already bending Emacs quite a bit, the current
>> scenario seems to break the camel's hump in unexpected ways (altho
>> there's always a reason to explain the behaviour).  Closing the bug
>> report.
>
> That's fair.  My argument for keeping it open would be that people who
> would be able and willing to scratch that itch in the future might not
> think to search for closed reports; but then again I suppose there's not
> a lot of information in our present discussion that they wouldn't be
> able to rediscover on their own.

It's very easy to make region highlighting window-local with just

diff --git a/lisp/simple.el b/lisp/simple.el
index 5443d961e1..04002073bd 100644
--- a/lisp/simple.el
+++ b/lisp/simple.el
@@ -6944,7 +6944,7 @@ redisplay--update-region-highlight
                            (eq window (minibuffer-selected-window))))))
         (funcall redisplay-unhighlight-region-function rol)
       (let* ((pt (window-point window))
-             (mark (mark))
+             (mark (or (window-parameter window 'mark) (mark)))
              (start (min pt mark))
              (end   (max pt mark))
              (new

Then the main task would be to find the right logic of setting
the window-local mark.  Since this feature request is closed,
here is a possible starting point for anyone who wants to experiment
with the prototype trying different use cases:

  (add-hook 'activate-mark-hook
            (lambda () (set-window-parameter nil 'mark (mark))))
  (add-hook 'deactivate-mark-hook
            (lambda () (set-window-parameter nil 'mark nil)))
  (add-hook 'window-selection-change-functions
            (lambda (window)
              (when (eq window (selected-window))
                (set-marker (mark-marker) (window-parameter window 'mark)))))