bug#52491: 28.0.90; Regression in window deletion with minibuffer

[martin rudalics]

> It's very strange that evaluating in the *Completions* buffer with
>
>    M-: (selected-window)
>
> shows that the selected window is the minibuffer window.
>
> Do you have an idea what is going on?

Not yet.  I suspect select_mini_window_flag is somehow involved here but
I cannot see how it could possibly affect 'selected-window'.

>> ... the attached patch should cure it (I suppose this got broken in
>> commit d0c7d8bc22a935f2a79747a96b4043f0b449a212 but didn't check).
>
> Thanks, then this patch should be pushed to master,
> and another shorter patch to the release branch?

No.  This patch might get the "t" case wrong - there we do not
necessarily want to select the previous window.

martin