bug#28443: 26.0.50; Pause when fullscreening when ns-use-native-fullscreen is nil on macOS

[Alan Third]

On Wed, Sep 13, 2017 at 12:33:48PM -0700, Aaron Jensen wrote:
> On Wed, Sep 13, 2017 at 12:27 PM, Alan Third <alan@idiocy.org> wrote:
> > On Wed, Sep 13, 2017 at 05:15:49AM -0700, Aaron Jensen wrote:
> >>
> >> When ns-use-native-fullscreen is nil and a frame is set to fullscreen,
> >> there is a slight pause before the frame goes fullscreen. I believe the
> >> pause is about the same length as the animation when native fullscreen
> >> is used, so I'm wondering if there's some timing code in there that can
> >> be disabled when non-native fullscreen is being used.
> >
> > Found it. In ns_fullscreen_hook in nsterm.m there is a timer:
> >
> >       /* Old style fs don't initiate correctly if created from
> >          init/default-frame alist, so use a timer (not nice...).
> >       */
> >       [NSTimer scheduledTimerWithTimeInterval: 0.5 target: view
> >                                      selector: @selector (handleFS)
> >                                      userInfo: nil repeats: NO];
> >
> > if I replace the timer with
> >
> >     [view handleFS]
> >
> > it works absolutely fine, but I have to assume there are situations
> > where it doesn’t.
> 
> My guess is that the timer was there to sync w/ the animation. If the
> animation is disabled, there shouldn't be a timer, but if it is, there
> should be. This is just a guess.

I should have read it a bit more closely. The comment’s saying there’s
a problem when

    (add-to-list ’frame-default-alist ’(fullscreen . fullboth))

is used, and sure enough, when I set it and create a new frame, it
draws incorrectly.

I notice that changing back from fullscreen results in the top of the
emacs window being hidden behind the toolbar. Clearly non‐native
fullscreen has some resizing issues.

It can be (correctly) emulated with:

    (set-frame-parameter nil 'undecorated t)
    (setq ns-auto-hide-menu-bar 1)
    (tool-bar-mode -1)
    (toggle-frame-maximized)

which makes me wonder if we even need this code and could move it to
lisp?
-- 
Alan Third