3 - how does clients move their surfaces? set a full tri-mesh every
4 time? probably just go back to x and y position, let the compositor
5 do the fancy stuff. How does the server apply transformations to a
6 surface behind the clients back? (wobbly, minimize, zoom) Maybe
9 - switch scanout when top surface is full screen
11 - what about cursors then? maybe use hw cursors if the cursor
12 satisfies hw limitations (64x64, only one cursor), switch to
13 composited cursors if not.
15 - multihead, screen geometry and crtc layout protocol, hotplug
17 - input device discovery, hotplug
19 - Advertise axes as part of the discovery, use something like
20 "org.wayland.input.x" to identify the axes.
22 - keyboard state, layout events at connect time and when it
27 - input handling - keyboard focus, multiple input devices,
28 multiple pointers, multi touch.
30 - wayland-system-compositor
32 - device kit/libudev/console kit integration to discover seats,
33 that is, groups of input devices and outputs that provide a
34 means for one user to interact with the system. That is,
35 typically a mouse, keyboard and a screen. The input devices
36 will just be evdev devices, the outputs will be a drm device
37 filename and the specific outputs accessible throught that drm
40 - send drm device in connection info, probably just udev path.
42 - cairo-drm; wayland needs cairo-drm one way or another:
44 - chris wilson (ickle) is doing cairo-drm for i915 now, basically
45 the pixman-drm idean, but inside cairo instead.
47 - pixman-drm; move the ddx driver batchbuffer logic into libdrm
48 and write native, direct rendering acceleration code in
49 pixman-drm. is a clean approach in that we avoid the mess of
50 the global GL context leaking through to applications, and we
51 can bootstrap this project by pulling in the EXA hooks from the
54 - use open gl behind the scenes a la glitz.
56 - should be possible to provide a realistic api and then stub out
57 the implementation with pwrite and pread so gtk+ port can proceed.
59 - XKB like client side library for translating keyboard events to
60 more useful keycodes and modifiers etc. Will probably be shared
61 between toolkits as a low-level library.
65 - eek, so much X legacy stuff there...
67 - draw window decorations in gtkwindow.c
69 - start from alexl's client-side-windows branch
71 - Details about pointer grabs. wayland doesn't have active grabs,
72 menus will behave subtly different. Under X, clicking a menu
73 open grabs the pointer and clicking outside the window pops down
74 the menu and swallows the click. without active grabs we can't
75 swallow the click. I'm sure there much more...
77 - Port Qt? There's already talk about this on the list.
81 - move most of the code from xf86-video-intel into a Xorg wayland
84 - don't ask KMS for available output and modes, use the info from
85 the wayland server. then stop mooching off of drmmode.c.
87 - map multiple wayland input devices to MPX in Xorg.
89 - rootless; avoid allocating and setting the front buffer, draw
90 window decorations in the X server (!), how to map input?
92 - gnome-shell as a wayland session compositor
94 - runs as a client of the wayland session compositor, uses
95 clutter+egl on wayland
97 - talks to an Xorg server as the compositing and window manager
98 for that server and renders the output to a wayland surface.
99 the Xorg server should be modified to take input from the system
100 compositor through gnome-shell, but not allocate a front buffer.
102 - make gnome-shell itself a nested wayland server and allow native
103 wayland clients to connect and can native wayland windows with
104 the windows from the X server.
106 - qemu as a wayland client; session surface as X case
108 - qemu has too simple acceleration, so a Wayland backend like the
109 SDL/VNC ones it has now is trivial.
111 - paravirt: forward wayland screen info as mmio, expose gem ioctls as mmio
113 - mapping vmem is tricky, should try to only use ioctl (pwrite+pread)
115 - not useful for Windows without a windows paravirt driver.
117 - two approaches: 1) do a toplevel qemu window, or 2) expose a
118 wayland server in the guest that forwards to the host wayland
119 server, ie a "remote" compositor, but with the gem buffers
120 shared. could do a wl_connection directly on mmio memory, with
121 head and tail pointers. use an alloc_head register to indicate
122 desired data to write, if it overwrites tail, block guest. just
123 a socket would be easier.
125 - moblin as a wayland compositor
127 - clutter as a wayland compositors
131 - make libwayland-client less ghetto
133 - sparse based idl compiler
137 - xml based description instead?
139 - actually make batch/commit batch up commands
141 - protocol for setting the cursor image
143 - should we have a mechanism to attach surface to cursor for
144 guaranteed non-laggy drag?
146 - drawing cursors, moving them, cursor themes, attaching surfaces
147 to cursors. How do you change cursors when you mouse over a
148 text field if you don't have subwindows? This is what we do: a
149 client can set a cursor for a surface and wayland will set that
150 on enter and revert to default on leave
152 - server should be able to discard surfaces, and send a re-render
153 event to clients to get them to render and provide the surface
154 again. for wayland system compositor vt switcing, for example, to
155 be able to throw away the surfaces in the session we're switching
156 away from. for minimized windows that we don't want live thumb
159 - auth; We need to generate a random socket name and advertise that
160 on dbus along with a connection cookie. Something like a method
161 that returns the socket name and a connection cookie. The
162 connection cookie is just another random string that the client
163 must pass to the wayland server to become authenticated. The
164 Wayland server generates the cookie on demand when the dbus method
165 is called and expires it after 5s or so.
167 - or just pass the fd over dbus
169 - drm bo access control, authentication, flink_to
171 - enter/leave events from the input devices
173 - gain, lose keyboard focus events; this event carries information
174 about which keys are currently held down as a surface gains focus
175 so the client can deduce modifier state.
177 - Range protocol may not be sufficient... if a server cycles through
178 2^32 object IDs we don't have a way to handle wrapping. And since
179 we hand out a range of 256 IDs to each new clients, we're just
180 talking about 2^24 clients. That's 31 years with a new client
181 every minute... Maybe just use bigger ranges, then it's feasible
182 to track and garbage collect them when a client dies.
184 - multi gpu, needs queue and seqno to wait on in requests
186 - opaque region, window rect
188 - save and restore state on crash, clients reconnect, re-render
191 - how do apps share the glyph cache? what is the glyph cache, how
192 does it work? pixbuf cache?
194 - synaptics, 3-button emulation, scim
196 - what to do when protocol out buffer fills up? Just block on write
197 would work I guess. Clients are supposed to throttle using the
198 bread crumb events, so we shouldn't get into this situation.
200 - when a surface is the size of the screen and on top, we can set the
201 scanout buffer to that surface directly. Like compiz unredirect
202 top-level window feature. Except it won't have any protocol state
203 side-effects and the client that owns the surface won't know. We
204 lose control of updates. Should work well for X server root window