3 - surface.set_grab_mode(GRAB_OWNER_EVENTS vs GRAB_SURFACE_EVENTS), to
4 make menus work right: click and drag in a menubar grabs the
5 pointer to the menubar (which we need for detecting motion into
6 another menu item), but we need events for the popup menu surface
9 - input_device.attach() should use a timestamp so the server can
10 discard late requests (sending a request to set the pointer image
11 in response to a motion event, the pointer leaves and then reenters
12 the surface, before the server receives the reqest -> the server
15 - The message format has to include information about number of fds
16 in the message so we can skip a message correctly. Or we should
17 just give up on trying to recover from unknown messages.
19 - generate pointer_focus (and drag focus) on raise/lower, move
20 windows, all kinds of changes in surface stacking.
26 Root window must send NULL type (to decline drop) or
27 x-wayland/root-something type if the source offers that. But the
28 target deletes the drag_offer object when drag.pointer_focus leaves
31 How do we animate the drag icon back to the drag origin in case of
34 How to handle surfaces from clients that don't know about dnd or
35 don't care? Maybe the dnd object should have a
36 dnd.register_surface() method so clients can opt-in the surfaces
37 that will participate in dnd. Or just assume client is not
38 participating until we receive an accept request.
40 - Pointer image issue:
42 - A touch input device doesn't have a pointer; indicate that
45 - Cursor themes, tie in with glyph/image cache.
47 - copy-n-paste, store data in server (only one mime-type available)
48 or do X style (content mime-type negotiation, but data goes away
51 - Discard buffer, as in "wayland discarded your buffer, it's no
52 longer visible, you can stop updating it now.", reattach, as in "oh
53 hey, I'm about to show your buffer that I threw away, what was it
54 again?". for wayland system compositor vt switcing, for example,
55 to be able to throw away the surfaces in the session we're
56 switching away from. for minimized windows that we don't want live
59 - Initial placement of surfaces. Guess we can do, 1)
60 surface-relative (menus), 2) pointer-relative (tooltips and
61 right-click menus) or 3) server-decides (all other top-levels).
63 - When a surface is the size of the screen and on top, we can set the
64 scanout buffer to that surface directly. Like compiz unredirect
65 top-level window feature. Except it won't have any protocol state
66 side-effects and the client that owns the surface won't know. We
67 lose control of updates. Should work well for X server root window
68 under wayland. Should be possible for yuv overlays as well.
70 - what about cursors then? maybe use hw cursors if the cursor
71 satisfies hw limitations (64x64, only one cursor), switch to
72 composited cursors if not.
74 - clients needs to allocate the surface to be suitable for
75 scanout, which they can do whenever they go fullscreen.
77 - multihead, screen geometry and crtc layout protocol, hotplug
79 - input device discovery, hotplug
81 - Advertise axes as part of the discovery, use something like
82 "org.wayland.input.x" to identify the axes.
84 - keyboard state, layout events at connect time and when it
85 changes, keyboard leds
91 - synaptics, 3-button emulation, scim
93 - auth; We need to generate a random socket name and advertise that
94 on dbus along with a connection cookie. Something like a method
95 that returns the socket name and a connection cookie. The
96 connection cookie is just another random string that the client
97 must pass to the wayland server to become authenticated. The
98 Wayland server generates the cookie on demand when the dbus method
99 is called and expires it after 5s or so.
101 - or just pass the fd over dbus
103 - drm bo access control, authentication, flink_to
105 - Range protocol may not be sufficient... if a server cycles through
106 2^32 object IDs we don't have a way to handle wrapping. And since
107 we hand out a range of 256 IDs to each new clients, we're just
108 talking about 2^24 clients. That's 31 years with a new client
109 every minute... Maybe just use bigger ranges, then it's feasible
110 to track and garbage collect them when a client dies.
112 - Add protocol to let applications specify the effective/logical
113 surface rectangle, that is, the edge of the window, ignoring drop
114 shadows and other padding. The compositor needs this for snapping
115 and constraining window motion. Also, maybe communicate the opaque
116 region of the window (or just a conservative, simple estimate), to
117 let the compositor reduce overdraw.
119 - multi gpu, needs queue and seqno to wait on in requests
125 - draw window decorations in gtkwindow.c
127 - Details about pointer grabs. wayland doesn't have active grabs,
128 menus will behave subtly different. Under X, clicking a menu
129 open grabs the pointer and clicking outside the window pops down
130 the menu and swallows the click. without active grabs we can't
131 swallow the click. I'm sure there much more...
133 - Port Qt? There's already talk about this on the list.
137 - move most of the code from xf86-video-intel into a Xorg wayland
140 - don't ask KMS for available output and modes, use the info from
141 the wayland server. then stop mooching off of drmmode.c.
143 - map multiple wayland input devices to MPX in Xorg.
145 - rootless; avoid allocating and setting the front buffer, draw
146 window decorations in the X server (!), how to map input?
148 - gnome-shell as a wayland session compositor
150 - runs as a client of the wayland session compositor, uses
151 clutter+egl on wayland
153 - talks to an Xorg server as the compositing and window manager
154 for that server and renders the output to a wayland surface.
155 the Xorg server should be modified to take input from the system
156 compositor through gnome-shell, but not allocate a front buffer.
158 - make gnome-shell itself a nested wayland server and allow native
159 wayland clients to connect and can native wayland windows with
160 the windows from the X server.
162 - qemu as a wayland client; session surface as X case
164 - qemu has too simple acceleration, so a Wayland backend like the
165 SDL/VNC ones it has now is trivial.
167 - paravirt: forward wayland screen info as mmio, expose gem ioctls as mmio
169 - mapping vmem is tricky, should try to only use ioctl (pwrite+pread)
171 - not useful for Windows without a windows paravirt driver.
173 - two approaches: 1) do a toplevel qemu window, or 2) expose a
174 wayland server in the guest that forwards to the host wayland
175 server, ie a "remote" compositor, but with the gem buffers
176 shared. could do a wl_connection directly on mmio memory, with
177 head and tail pointers. use an alloc_head register to indicate
178 desired data to write, if it overwrites tail, block guest. just
179 a socket would be easier.
181 - moblin as a wayland compositor
183 - clutter as a wayland compositors