Gtk4 and Wayland

[Thiago Milczarek Sayão]

Hi,

It seems it would be very hard to have gtk4 and/or Wayland on JavaFX,
unless we stop doing window manager related things.
Resizing the window would be possible, but not stacking or moving it around.

Wayland simply does not have this functionality and gtk4 removed it as
well, as pointed out here

https://gnome.pages.gitlab.gnome.org/gtk/gtk4/migrating-3to4.html#adapt-to-gdkwindow-api-changes

and here:

https://discourse.gnome.org/t/gtk4-migration-window-management-functions-gone/7542/4

Even on gtk3 it's a hard thing to do (deal with stacking and positioning).

So I agree with the Gtk developers position to simply not deal with window
management at this level, but it's somehow unfortunate.

On the "experimenting with gtk4" side, it seems like I hit a dead end
because gtk4 does not provide the needed functionality nor Wayland (to deal
with it directly).

Sticking with X11 also seems like not optimal for the long term, it will be
replaced eventually.

Maybe JavaFX should move away from window management as well?

-- Thiago.



Em qua., 1 de set. de 2021 às 08:42, Kevin Rushforth <
kevin.rushforth at oracle.com> escreveu:

> This seems reasonable to me as well, at least for JavaFX since we
> already rely on Gtk for most of the windowing toolkit on Linux.
>
> -- Kevin
>
> On 9/1/2021 4:26 AM, Mario Torre wrote:
> > On Wed, Sep 1, 2021 at 1:18 PM Johan Vos <johan.vos at gluonhq.com> wrote:
> >> Hi Thiago,
> >>
> >> I was thinking (and experimenting) in the same direction, and so far
> that
> >> is working good. I don't see disadvantages, but it would be good to find
> >> out about it before we move forward.
> >> Maybe the main issue to me is that GTK comes with lots of dependencies.
> We
> >> already have that situation today, so it is not going to be worse. But
> if
> >> we would use the Wayland protocol on a lower level (with a
> Wayland-specific
> >> glass platform, instead of GTK), we could probably reduce the
> dependencies.
> >> However, this comes at the price of creating *and maintaining* more
> >> low-level code.
> >>
> >> I've been running GTK3 on Wayland and that works fine too. However, it
> >> might be better to focus Wayland support for GTK4. I think there won't
> be
> >> many installs that have no X11, but only Wayland, and that have GTK3
> and no
> >> GTK4.
> > I tend to agree with you, I think we should consider this for Wakefield
> too.
> >
> > Cheers,
> > Mario
> >
> >> - Johan
> >> - Johan
> >>
> >> On Tue, Aug 31, 2021 at 11:03 PM Thiago Milczarek Sayão <
> >> thiago.sayao at gmail.com> wrote:
> >>
> >>> Hi,
> >>>
> >>> I did some investigation on gtk4 and wayland.
> >>>
> >>> After some research I ended up with the conclusion that the best way
> is to
> >>> do a separate gtk4 backend, that would support X11 and Wayland.
> >>>
> >>> This would be a good start:
> >>> https://github.com/openjdk/jfx/pull/77/files
> >>>
> >>> Why?
> >>>
> >>> Gtk4 moves the decoration to the client side, which is GREAT since
> knowing
> >>> the window size with decoration was a real pain.
> >>>
> >>> We probably won't want to do all the decoration work, Gtk does that,
> but on
> >>> GtkWindow level, not GdkSurface (which replaces GdkWindow).
> >>>
> >>> Thus the move to use "more Gtk" (hence "less Gdk") which is exactly
> what
> >>> the PR does. It also removes Applet code
> >>>
> >>> This is also a good starting point:
> >>> https://gnome.pages.gitlab.gnome.org/gtk/gtk4/migrating-3to4.html
> >>>
> >>> --Thiago.
> >>>
> >
>
>