Re: Selecting tooltip frames considered harmful

[martin rudalics]

> the *semantics* should be "the echo area of the window for
> which the tooltip frame is transient", I think (and if tip_last_frame
> gives us that, then that can be used in the implementation).

tip_last_frame should give us that.

>> We currently refuse to delete a frame when its minibuffer window
>> serves as the minibuffer window of another frame.
>
> I think this is an accident of implementation, not a feature.
> As a user I've found it annoying on a bunch of occasions.
> So we don't have to reproduce that misfeature here.
>
> A more useful behavior would be to allow deleting that frame, and then
> next time the other frame needs a miniwindow, you just go "oops, our
> miniwindow has been deleted, let's look for another one".
>
> We could still have a check when deleting a frame to make sure there
> remains at least 1 miniwindow somewhere, of course.

Using a mechanism similar to that for finding the new
'default-minibuffer-frame' I presume.

martin