wayland/TODO

Core wayland protocol

 - scanner: wl_* prefix removal: split it out into a namespace part so
   we can call variables "surface" instead of "wl_surface"?

 - we need surface.enter/leave events to be emitted as the surface
   enters and leaves outputs.  This lets us track which output(s) a
   surface is currently showing on, which affects how we render
   (subpixel information, dpi, rotation).  By using enter/leave
   events, a surface can be on multiple output.

 - We need rotation information in the output (multiples of 90
   degress) and we'll need a way for a client to communicate that it
   has rendered it's buffer according to the output rotation.  The
   goal is to be able to pageflip directly to the client buffer, and
   for that we need the client to render accordingly and the
   compositor needs to know that it did.

 - Atomicity.  Currently a lot of the atomicity in Wayland relies on
   how we batch up all requests in a protocol buffer and only flushes
   in the "blockhandler" in the client.  Consensus was that we need
   something more reliable and explicit.  The suggestion is that we
   make surface.attach a synchronization point such that everything
   before that is batched and applied atomically when the
   surface.attach request comes in.  For cases where we need atomicity
   beyond a surface.attach, we can add an atomic grouping mechanism,
   that can group together multiple surface.attach requests into a
   bigger atomic change.  To be researched a bit.

 - We should make pointer sprites regular surfaces.  Something like
   input_device.set_sprite(surface).  This also make client side
   animated cursors simple/possible, since we now get a frame event
   that can drive the animation.

 - Maybe try to make remote wayland actually happen, to see if there
   is something in the protocol/architecute that makes it harder than
   it should be.

 - Reconsider data types for coordinates in events.  double, floats or
   fixed point.  Transformed and/or accelerated input generates
   sub-pixel positions.  24.8 fixed point could work.  Need to think
   about the range of coordinates we need.  Different from X problem,
   since we don't have sub-windows, which is where X hits the 16 bit
   limitations.  Monitor and window sizes haven't yet out-grown the 16
   bit range.

 - Key events need a bit more work/thinking/redesign. As it is we need
   a mechanism to change and synchronize keymaps, repeat rates.  But
   as we started talking, we decided that we needed to go back and
   research what other systems do instead of just essentially copying
   the X model.  Sending out unicode events in addition to keycode
   events has a lot of benefits (OSK can send out unicode events
   instead of fake keycode events, apps become much simpler...)  Move
   keymap handling and repeat to the server?  Needs more research.

 - Pointer axis events need modifiers (ctrl-scroll eg), but we either
   need to send the modifier state with each axis/scroll event or send
   keys down on pointer_focus and subsequent key events... or just key
   events for modifier keys... or for the non-repeating subset?

 - Input protocol restructuring: break up events into wl_pointer
   (enter/leave/motion/button/axis events, set_pointer_surface
   request), wl_keyboard (enter/leave/key events... what
   else... unicode event, set_map request? pending kb work), and
   wl_touch (down/up/motion/cancel events) interfaces.  Rename
   wl_input_device to wl_seat.  wl_seat has zero or one of each, and
   will announce this at bind time.  Raw devices are also tied to a
   wl_seat, but we may not do that for 1.0, we just need to make sure
   wl_seat has a forward compatible way to announce them.

 - Add timestamp to touch_cancel, add touch id to touch_cancel (?)

 - Serial numbers.  The wayland protocol, as X, uses timestamps to
   match up certain requests with input events.  The problem is that
   sometimes an event happens that triggers a timestamped event.  For
   example, a surface goes away and a new surface receives a
   pointer.enter event.  These events are normally timestamped with
   the evdev event timestamp, but in this case, we don't have a evdev
   timestamp.  So we have to go to gettimeofday (or clock_gettime())
   and then we don't know if it's coming from the same time source
   etc.  And we don't really need a real time timestamp, we just need
   a serial number that encodes the order of events inside the server.
   So we need to introduce a serial number mechanism (uint32_t,
   maintained in libwayland-server.so) that we can use to order
   events, and have a look at the events we send out and decide
   whether they need serial number or timestamp or both.  We still
   need real-time timestamps for actual input device events (motion,
   buttons, keys, touch), to be able to reason about double-click
   speed and movement speed.  The serial number will also give us a
   mechanism to key together events that are "logically the same" such
   as a unicode event and a keycode event, or a motion event and a
   relative event from a raw device.

 - The output protocol needs to send all the ugly timing details for the modes.

ICCCM

 - clipboard manager interface?  what's needed?  just notification
   that the selection has gone away.  should the clipboard manager be
   able to take over the selection "seamlessly", ie, with the same
   timestamp etc?  Doesn't seem like we need that, the clipboard will
   have to set a new data_source anyway, with the subset of mimetypes
   it offers (the clipboad manager may only offer a subset of the
   types offered by the original data_source)

 - mime-type guidelines for data_source (ie, both dnd and selection):
   recommended types for text or images, types that a clipboard
   manager must support, mime-types must be listed in preferred order

 - TRANSIENT_FOR handled by wl_shell_surface, WM_CLASS replaced by
   .desktop file filename (absolute path if non-standard location)
   WM_CLASS used for grouping windows in one button in a panel, for
   example.  So we'll need a request to set that.

 - we need a "no kb focus please" mechanism.  Or should this be
   implicit in a specific surface type?

EWMH

 - configure should provide dx_left, dx_right, dy_top, dy_bottom, or
   dx, dy, width and height.

 - _NET_WM_NAME (shell_surface.set_title(utf8)), _NET_WM_ICON Is this
   just another wl_surface? Do we need this if we have the .desktop
   file?  How to set multiple sizes?

 - ping event, essentially the opposite of the display.sync request.
   Compositor can ping clients to see if they're alive (typically when
   sending input events to a client surface)

 - move to workspace, keep on top, on all workspaces, minimize etc
   requests for implementing client side window menu? or just make a
   "show window menu" request to let the compositor display and manage
   a popup window?

 - window move and resize functionality for kb and touch.

 - dnd loose ends: self-dnd: initiate dnd with a null data-source,
   compositor will not offer to other clients, client has to know
   internally what's offered and how to transfer data. no fd passing.

 - Protocol for specifying title bar rectangle (for moving
   unresponsive apps).  Rectangle for close button, so we can popup
   force-close dialog if application doesn't respond to ping event
   when user clicks there.  We could use the region mechanism here
   too.

 - popup placement protocol logic.

 - subsurface mechanism.  we need this for cases where we would use an
   X subwindow for gl or video other different visual type.

EGL/gbm

 - Land Anders gbm_surface patches.

 - Don't wl_display_iterate in eglSwapBuffer, send an eventfd fd?

 - Land Robert Braggs EGL extensions: frame age, swap with damage

 - Make it possible to share buffers from compositor to clients.
   Tricky part here is how to indicate to EGL on the server side that
   it should make an EGLImage available to a client.  We'll need a
   "create a wl_buffer for this EGLImage for this client" kind of
   entry point.

 - Protocol for arbitrating access to scanout buffers (physically
   contiguous memory).  When a client goes fullscreen (or ideally as
   the compositor starts the animation that will make it fullscreen)
   we send a "give up your scanout buffer" to the current fullscreen
   client (if any) and when the client acks that we send a "try to
   allocate a scanout buffer now" event to the fullscreen-to-be
   client.

Misc

 - glyph cache

    - Needs a mechanism to pass buffers to client.

      buffer = drm.create_buffer(); /* buffer with stuff in it */

      cache.upload(buffer, x, y, width, height, int hash)

      drm.buffer: id, name, stride etc /* event to announce cache buffer */

      cache.image: hash, buffer, x, y, stride /* event to announce
					      * location in cache */

      cache.reject: hash   /* no upload for you! */

      cache.retire: buffer /* cache has stopped using buffer, please
			    * reupload whatever you had in that buffer */

 - A "please suspend" event from the compositor, to indicate to an
   application that it's no longer visible/active.  Or maybe discard
   buffer, as in "wayland discarded your buffer, it's no longer
   visible, you can stop updating it now.", reattach, as in "oh hey,
   I'm about to show your buffer that I threw away, what was it
   again?".  for wayland system compositor vt switcing, for example,
   to be able to throw away the surfaces in the session we're
   switching away from.  for minimized windows that we don't want live
   thumb nails for. etc.

libxkbcommon

  - pull in actions logic from xserver

  - pull in keycode to keysym logic from libX11

  - expose alloc functions in libxkbcommon, drop xserver funcs?

  - pull the logic to write the xkb file from xkb_desc and names into
    libxkbcommon and just build up the new xkb_desc instead of
    dump+parse? (XkbWriteXKBKeymapForNames followed by
    xkb_compile_keymap_from_string in XkbDDXLoadKeymapByNames)

  - pull in keysym defs as XKB_KEY_BackSpace

  - figure out what other X headers we can get rid of, make it not
    need X at all (except when we gen the keysyms).

  - Sort out namespace pollution (XkbFoo macros, atom funcs etc).

  - Sort out 32 bit vmods and serialization


Clients and ports

 - port gtk+

    - draw window decorations in gtkwindow.c

    - Details about pointer grabs. wayland doesn't have active grabs,
      menus will behave subtly different.  Under X, clicking a menu
      open grabs the pointer and clicking outside the window pops down
      the menu and swallows the click.  without active grabs we can't
      swallow the click.  I'm sure there much more...

    - dnd, copy-paste

 - Investigate DirectFB on Wayland (or is that Wayland on DirectFB?)

 - SDL port, bnf has work in progress here:
   http://cgit.freedesktop.org/~bnf/sdl-wayland/

 - libva + eglimage + kms integration


Ideas

 - A wayland settings protocol to tell clients about themes (icons,
   cursors, widget themes), fonts details (family, hinting
   preferences) etc.  Just send all settings at connect time, send
   updates when a setting change.  Getting a little close to gconf
   here, but could be pretty simple:

     interface "settings":
       event int_value(string name, int value)
       event string_value(string name, string value)

   but maybe it's better to just require that clients get that from
   somewhere else (gconf/dbus).


Crazy ideas

 - AF_WAYLAND - A new socket type.  Eliminate compositor context
   switch by making kernel understand enough of wayland that it can
   forward input events as wayland events and do page flipping in
   response to surface_attach requests:

    - ioctl(wayland_fd, "surface_attach to object 5 should do a kms page
			 flip on ctrc 2");

    - what about multiple crtcs? what about frame event for other
      clients?

    - forward these input devices to the client

    - "scancode 124 pressed or released with scan codes 18,22 and 30
       held down gives control back to userspace wayland.

    - what about maintaining cursor position? what about pointer
      acceleration?  maybe this only works in "client cursor mode",
      where wayland hides the cursor and only sends relative events?
      Solves the composited cursor problem.  How does X show its
      cursor then?

    - Probably not worth it.