pillule <pillule@riseup.net> writes:

>>> What's wrong with putting the first disjunct into the 
>>> conditional as in
>>> the below?  In general, always try to avoid larger indentation 
>>> changes -
>>> they can make forensics cumbersome while bisecting.
>>>
>>>      (cond
>>>       ;; First try to delete dedicated windows that are not 
>>>       side       windows
>>>       ((and dedicated (not (eq dedicated 'side))
>>>             (window--delete window 'dedicated (eq bury-or-kill 
>>>             'kill))))
>>>       ((and (not prev-buffer)
>>>              (eq (nth 1 quit-restore) 'tab)
>>>              (eq (nth 3 quit-restore) buffer))
>>
>> The difference is a window dedicated with flag t may not be 
>> deletable, and in this case, we want
>> it
>> to pass through the others conditionals branch of 
>> quit-restore-window, so it can try to use the
>> 'quit-restore parameter, close the tab or to fallback in t, 
>> etc.
>> Explaining it makes me thing I could use 'window-deletable-p' 
>> in its conditional and ...
>> I guess, problem solved
>>
>
> I read it again and think you were right,
> when  (window--delete window 'dedicated (eq bury-or-kill 'kill))
> is part of the conditional, it indeed already fail if the window 
> is not deletable ;
>
> I will correct that in the next revision.

hm. here again minors corrections. sorry for the noise.