Check that value for GTK_HIDE_TITLEBAR_WHEN_MAXIMIZED has actually
changed before reconfiguring the window.

Bug: 14511

GTK+ will update its GTK_FRAME_EXTENTS property continuously even when
the value hasn't actually changed, which causes the maximized state to
be cleared when transitioning to/from fullscreen to maximized.

Check that he extents have changed and reconfigure the client's window
only when the value has changed.

Bug 14514

xfwm4 would fail to add maximized states using extended window manager
hints mechanisms.

Bug: 14475

Skip ARGB10 GLX configs as it won't work well.

For debugging purpose.

Just a small attempt at optimizing things a bit.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

Bug: 14323

On unmap/destroy, the damage may already be destroyed along with the
attached resource, make sure we catch those errors so it doesn't bother
gtk+.

This adds:
  myDisplayErrorTrapPush()
  myDisplayErrorTrapPop()
  myDisplayErrorTrapPopIgnored()
  myDisplayBeep()
  myDisplayGetKeymap()

Se we don't have to check for gtk+ version all over the place.

That would prevent building with --enable-debug=full

Make traces more consistent.

Andre Miranda authored 7 years ago and Olivier Fourdan committed 6 years ago

Bug: 14240

Adam Purkrt authored 7 years ago and Olivier Fourdan committed 6 years ago

Bug: 14181

The comparison is corrected, closing a one pixel gap between
TILE_LEFT and TILE_UP_LEFT and between TILE_RIGHT and TILE_UP_RIGHT.

Bug 14164

Still need to raise the window on activate when using focus follow mode
otherwise.

Adam Purkrt authored 7 years ago and Olivier Fourdan committed 7 years ago

Bug: 14174

To keep the distance of the grip from the edges of the window
in proportion when unmaximizing/untiling while moving the window,
the grip position is recalculated using pxratio and pyratio.

Bug 14164

This is a regression introduced by 3877bf45, where we would differ
raising the window until it's focused.

When using click to focus, activating a transient window would raise
only that transient and not its ancestor, if any.

Change the logic to raise the ancestor if it's different from the
requested window itself.

Would break --enable-debug=full as this induces -Werror.

No need to have that public, so we can please the compiler.

Adam Purkrt authored 7 years ago and Olivier Fourdan committed 7 years ago

Bug: 11960

This patch fixes the erratic behaviour when unmaximizing a window by
dragging it from the absolute top.

clientToggleMaximized is now called with each motion event if
a window is maximized, hence it is always unmaximized before possibly
toggling maximization again in clientTile further down.

When the xrandr layout is changed, we recompute the screen size on the
actual monitors size and layout.

Use this computed value instead of the size reported by Xlib as it's not
yet updated by the time we reconfigure our display. This is particularly
bad when the compositor is enabled as the overlay window end up being of
the wrong size.

David Siroky authored 7 years ago and Olivier Fourdan committed 7 years ago

Bug: 2701

Optionally raise the window being considered while cycling.

Bug: 13954

When a client issues a ConfigureRequest() on a toplevel window, that
window may transition from maximized to un-maximized state.

Make sure we repaint and relocate the client window within the frame.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

When using xpresent as vblank method, the clipping region would not
match the zoom level causing artifacts.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

vblank method can be selected using a command line option.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

When running xfwm4 from a nested X11 server (Xnest, Xephyr) from within
a Wayland session, GDK will selects in priority the Wayland backend,
which will lead to a crash because xfwm4 is an X11 window manager and
not a Wayland client.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

Bug: 13954

Commit aee8b25a:

    Do not prevent ALT+Mouse resizing for borderless maximized windows.

Introduced a regression with windows using size increment (like most
terminal emulators).

Fix the size increment issue while preserving the possibility to resize
maximized window.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

Andrey Alekseenkov authored 7 years ago and Olivier Fourdan committed 7 years ago

Bug: 13964

Bug: 13970

Not that it matters much in this case though.

Signed-off-by: Olivier Fourdan <fourdan@xfce.org>

P. Pronk authored 8 years ago and Olivier Fourdan committed 7 years ago

Bug: 13284

When dragging a window from a large monitor to the top of a smaller monitor to
maximize the window, it is possible that the window gets maximised on a
different monitor then the monitor your mouse is.

This happens because clientToggleMaximized uses the center point of the window
to calculate on which monitor the window should be maximised.

To fix this behaviour this commits adds a clientToggleMaximizedAtPoint method
so we can tell on which monitor the window should be maximised.

Xfwm error handler will cover all errors itself, there is no
necessity to use error traps for client.

GDK X error traps were changed in GDK 3 so using these traps
caused unwanted SIGTRAPs within GDK X error handler.