3 - scanner: wl_* prefix removal: split it out into a namespace part so
4 we can call variables "surface" instead of "wl_surface"?
6 - we need surface.enter/leave events to be emitted as the surface
7 enters and leaves outputs. This lets us track which output(s) a
8 surface is currently showing on, which affects how we render
9 (subpixel information, dpi, rotation). By using enter/leave
10 events, a surface can be on multiple output.
12 - We need rotation information in the output (multiples of 90
13 degress) and we'll need a way for a client to communicate that it
14 has rendered it's buffer according to the output rotation. The
15 goal is to be able to pageflip directly to the client buffer, and
16 for that we need the client to render accordingly and the
17 compositor needs to know that it did.
19 - Atomicity. Currently a lot of the atomicity in Wayland relies on
20 how we batch up all requests in a protocol buffer and only flushes
21 in the "blockhandler" in the client. Consensus was that we need
22 something more reliable and explicit. The suggestion is that we
23 make surface.attach a synchronization point such that everything
24 before that is batched and applied atomically when the
25 surface.attach request comes in. For cases where we need atomicity
26 beyond a surface.attach, we can add an atomic grouping mechanism,
27 that can group together multiple surface.attach requests into a
28 bigger atomic change. To be researched a bit.
30 - We should make pointer sprites regular surfaces. Something like
31 input_device.set_sprite(surface). This also make client side
32 animated cursors simple/possible, since we now get a frame event
33 that can drive the animation.
35 - Maybe try to make remote wayland actually happen, to see if there
36 is something in the protocol/architecute that makes it harder than
39 - Reconsider data types for coordinates in events. double, floats or
40 fixed point. Transformed and/or accelerated input generates
41 sub-pixel positions. 24.8 fixed point could work. Need to think
42 about the range of coordinates we need. Different from X problem,
43 since we don't have sub-windows, which is where X hits the 16 bit
44 limitations. Monitor and window sizes haven't yet out-grown the 16
47 - Key events need a bit more work/thinking/redesign. As it is we need
48 a mechanism to change and synchronize keymaps, repeat rates. But
49 as we started talking, we decided that we needed to go back and
50 research what other systems do instead of just essentially copying
51 the X model. Sending out unicode events in addition to keycode
52 events has a lot of benefits (OSK can send out unicode events
53 instead of fake keycode events, apps become much simpler...) Move
54 keymap handling and repeat to the server? Needs more research.
56 - Pointer axis events need modifiers (ctrl-scroll eg), but we either
57 need to send the modifier state with each axis/scroll event or send
58 keys down on pointer_focus and subsequent key events... or just key
59 events for modifier keys... or for the non-repeating subset?
61 - Input protocol restructuring: break up events into wl_pointer
62 (enter/leave/motion/button/axis events, set_pointer_surface
63 request), wl_keyboard (enter/leave/key events... what
64 else... unicode event, set_map request? pending kb work), and
65 wl_touch (down/up/motion/cancel events) interfaces. Rename
66 wl_input_device to wl_seat. wl_seat has zero or one of each, and
67 will announce this at bind time. Raw devices are also tied to a
68 wl_seat, but we may not do that for 1.0, we just need to make sure
69 wl_seat has a forward compatible way to announce them.
71 - Add timestamp to touch_cancel, add touch id to touch_cancel (?)
73 - Serial numbers. The wayland protocol, as X, uses timestamps to
74 match up certain requests with input events. The problem is that
75 sometimes an event happens that triggers a timestamped event. For
76 example, a surface goes away and a new surface receives a
77 pointer.enter event. These events are normally timestamped with
78 the evdev event timestamp, but in this case, we don't have a evdev
79 timestamp. So we have to go to gettimeofday (or clock_gettime())
80 and then we don't know if it's coming from the same time source
81 etc. And we don't really need a real time timestamp, we just need
82 a serial number that encodes the order of events inside the server.
83 So we need to introduce a serial number mechanism (uint32_t,
84 maintained in libwayland-server.so) that we can use to order
85 events, and have a look at the events we send out and decide
86 whether they need serial number or timestamp or both. We still
87 need real-time timestamps for actual input device events (motion,
88 buttons, keys, touch), to be able to reason about double-click
89 speed and movement speed. The serial number will also give us a
90 mechanism to key together events that are "logically the same" such
91 as a unicode event and a keycode event, or a motion event and a
92 relative event from a raw device.
94 - The output protocol needs to send all the ugly timing details for the modes.
98 - clipboard manager interface? what's needed? just notification
99 that the selection has gone away. should the clipboard manager be
100 able to take over the selection "seamlessly", ie, with the same
101 timestamp etc? Doesn't seem like we need that, the clipboard will
102 have to set a new data_source anyway, with the subset of mimetypes
103 it offers (the clipboad manager may only offer a subset of the
104 types offered by the original data_source)
106 - mime-type guidelines for data_source (ie, both dnd and selection):
107 recommended types for text or images, types that a clipboard
108 manager must support, mime-types must be listed in preferred order
110 - TRANSIENT_FOR handled by wl_shell_surface, WM_CLASS replaced by
111 .desktop file filename (absolute path if non-standard location)
112 WM_CLASS used for grouping windows in one button in a panel, for
113 example. So we'll need a request to set that.
115 - we need a "no kb focus please" mechanism. Or should this be
116 implicit in a specific surface type?
120 - configure should provide dx_left, dx_right, dy_top, dy_bottom, or
121 dx, dy, width and height.
123 - _NET_WM_NAME (shell_surface.set_title(utf8)), _NET_WM_ICON Is this
124 just another wl_surface? Do we need this if we have the .desktop
125 file? How to set multiple sizes?
127 - ping event, essentially the opposite of the display.sync request.
128 Compositor can ping clients to see if they're alive (typically when
129 sending input events to a client surface)
131 - move to workspace, keep on top, on all workspaces, minimize etc
132 requests for implementing client side window menu? or just make a
133 "show window menu" request to let the compositor display and manage
136 - window move and resize functionality for kb and touch.
138 - dnd loose ends: self-dnd: initiate dnd with a null data-source,
139 compositor will not offer to other clients, client has to know
140 internally what's offered and how to transfer data. no fd passing.
142 - Protocol for specifying title bar rectangle (for moving
143 unresponsive apps). Rectangle for close button, so we can popup
144 force-close dialog if application doesn't respond to ping event
145 when user clicks there. We could use the region mechanism here
148 - popup placement protocol logic.
150 - subsurface mechanism. we need this for cases where we would use an
151 X subwindow for gl or video other different visual type.
155 - Don't wl_display_iterate in eglSwapBuffer, send an eventfd fd?
157 - Land Robert Braggs EGL extensions: frame age, swap with damage
159 - Make it possible to share buffers from compositor to clients.
160 Tricky part here is how to indicate to EGL on the server side that
161 it should make an EGLImage available to a client. We'll need a
162 "create a wl_buffer for this EGLImage for this client" kind of
165 - Protocol for arbitrating access to scanout buffers (physically
166 contiguous memory). When a client goes fullscreen (or ideally as
167 the compositor starts the animation that will make it fullscreen)
168 we send a "give up your scanout buffer" to the current fullscreen
169 client (if any) and when the client acks that we send a "try to
170 allocate a scanout buffer now" event to the fullscreen-to-be
177 - Needs a mechanism to pass buffers to client.
179 buffer = drm.create_buffer(); /* buffer with stuff in it */
181 cache.upload(buffer, x, y, width, height, int hash)
183 drm.buffer: id, name, stride etc /* event to announce cache buffer */
185 cache.image: hash, buffer, x, y, stride /* event to announce
186 * location in cache */
188 cache.reject: hash /* no upload for you! */
190 cache.retire: buffer /* cache has stopped using buffer, please
191 * reupload whatever you had in that buffer */
193 - A "please suspend" event from the compositor, to indicate to an
194 application that it's no longer visible/active. Or maybe discard
195 buffer, as in "wayland discarded your buffer, it's no longer
196 visible, you can stop updating it now.", reattach, as in "oh hey,
197 I'm about to show your buffer that I threw away, what was it
198 again?". for wayland system compositor vt switcing, for example,
199 to be able to throw away the surfaces in the session we're
200 switching away from. for minimized windows that we don't want live
201 thumb nails for. etc.
205 - pull in actions logic from xserver
207 - pull in keycode to keysym logic from libX11
209 - expose alloc functions in libxkbcommon, drop xserver funcs?
211 - pull the logic to write the xkb file from xkb_desc and names into
212 libxkbcommon and just build up the new xkb_desc instead of
213 dump+parse? (XkbWriteXKBKeymapForNames followed by
214 xkb_compile_keymap_from_string in XkbDDXLoadKeymapByNames)
216 - pull in keysym defs as XKB_KEY_BackSpace
218 - figure out what other X headers we can get rid of, make it not
219 need X at all (except when we gen the keysyms).
221 - Sort out namespace pollution (XkbFoo macros, atom funcs etc).
223 - Sort out 32 bit vmods and serialization
230 - draw window decorations in gtkwindow.c
232 - Details about pointer grabs. wayland doesn't have active grabs,
233 menus will behave subtly different. Under X, clicking a menu
234 open grabs the pointer and clicking outside the window pops down
235 the menu and swallows the click. without active grabs we can't
236 swallow the click. I'm sure there much more...
240 - Investigate DirectFB on Wayland (or is that Wayland on DirectFB?)
242 - SDL port, bnf has work in progress here:
243 http://cgit.freedesktop.org/~bnf/sdl-wayland/
245 - libva + eglimage + kms integration
250 - A wayland settings protocol to tell clients about themes (icons,
251 cursors, widget themes), fonts details (family, hinting
252 preferences) etc. Just send all settings at connect time, send
253 updates when a setting change. Getting a little close to gconf
254 here, but could be pretty simple:
256 interface "settings":
257 event int_value(string name, int value)
258 event string_value(string name, string value)
260 but maybe it's better to just require that clients get that from
261 somewhere else (gconf/dbus).
266 - AF_WAYLAND - A new socket type. Eliminate compositor context
267 switch by making kernel understand enough of wayland that it can
268 forward input events as wayland events and do page flipping in
269 response to surface_attach requests:
271 - ioctl(wayland_fd, "surface_attach to object 5 should do a kms page
274 - what about multiple crtcs? what about frame event for other
277 - forward these input devices to the client
279 - "scancode 124 pressed or released with scan codes 18,22 and 30
280 held down gives control back to userspace wayland.
282 - what about maintaining cursor position? what about pointer
283 acceleration? maybe this only works in "client cursor mode",
284 where wayland hides the cursor and only sends relative events?
285 Solves the composited cursor problem. How does X show its
288 - Probably not worth it.