path: root/protocol/agl-shell.xml

<?xml version="1.0" encoding="UTF-8"?>
<protocol name="agl_shell">
  <copyright>
    Copyright © 2019, 2022 Collabora, Ltd.

    Permission is hereby granted, free of charge, to any person obtaining a
    copy of this software and associated documentation files (the "Software"),
    to deal in the Software without restriction, including without limitation
    the rights to use, copy, modify, merge, publish, distribute, sublicense,
    and/or sell copies of the Software, and to permit persons to whom the
    Software is furnished to do so, subject to the following conditions:

    The above copyright notice and this permission notice (including the next
    paragraph) shall be included in all copies or substantial portions of the
    Software.

    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
    THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
    FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
    DEALINGS IN THE SOFTWARE.
  </copyright>
  <interface name="agl_shell" version="6">
    <description summary="user interface for Automotive Grade Linux platform">
      Starting with version 2 of the protocol, the client is required to wait
      for the 'bound_ok' or 'bound_fail' events in order to proceed further.

      In case the client gets a 'bound_fail' event then it should consider that
      there's another client already bound to the agl_shell protocol.
      A client that receives a 'bound_ok' event should consider that there's
      no other client already bound to the interface and can proceed further.

      If the client uses an older version of the protocol it will receive
      automatically an error and the compositor will terminate the connection,
      if there's another client already bound the interface.

      If the client receives the 'bound_fail' event and attempts to use the
      interface further it will receive an error and the compositor will
      terminate the connection. After the 'bound_fail' event was received the
      client should call the destructor, which has been added with version 2
      of the protocol. The client is free to try at a later point in time to
      see if it will receive the 'bound_ok' event, but there's no explicit way
      of finding out when that event will be delivered.
      It is assumed that it can infer that information through other
      means/other channels.
    </description>

    <enum name="error">
      <entry name="invalid_argument" value="0"/>
      <entry name="background_exists" value="1"/>
      <entry name="panel_exists" value="2"/>
    </enum>

    <enum name="edge">
      <entry name="top" value="0"/>
      <entry name="bottom" value="1"/>
      <entry name="left" value="2"/>
      <entry name="right" value="3"/>
    </enum>

    <enum name="app_state" since="3">
      <entry name="started" value="0"/>
      <entry name="terminated" value="1"/>
      <entry name="activated" value="2"/>
      <entry name="deactivated" value="3"/>
    </enum>

    <request name="ready">
      <description summary="client is ready to be shown">
        Tell the server that this client is ready to be shown. The server
        will delay presentation during start-up until all shell clients are
        ready to be shown, and will display a black screen instead.
        This gives the client an opportunity to set up and configure several
        surfaces into a coherent interface.

        The client that binds to this interface must send this request, otherwise
        they may stall the compositor unnecessarily.

        If this request is called after the compositor has already finished
        start-up, no operation is performed.
      </description>
    </request>

    <request name="set_background">
      <description summary="set surface as output's background">
        Set the surface to act as the background of an output. After this
        request, the server will immediately send a configure event with
        the dimensions the client should use to cover the entire output.

        The surface must have a "desktop" surface role, as supported by
        libweston-desktop.

        Only a single surface may be the background for any output. If a
        background surface already exists, a protocol error is raised.
      </description>
      <arg name="surface" type="object" interface="wl_surface"/>
      <arg name="output" type="object" interface="wl_output"/>
    </request>

    <request name="set_panel">
      <description summary="set surface as panel">
        Set the surface to act as a panel of an output. The 'edge' argument
        says what edge of the output the surface will be anchored to.
        After this request, the server will send a configure event with the
        corresponding width/height that the client should use, and 0 for the
        other dimension. E.g. if the edge is 'top', the width will be the
        output's width, and the height will be 0.

        The surface must have a "desktop" surface role, as supported by
        libweston-desktop.

        The compositor will take the panel's window geometry into account when
        positioning other windows, so the panels are not covered.

        XXX: What happens if e.g. both top and left are used at the same time?
        Who gets to have the corner?

        Only a single surface may be the panel for an output's edge. If a
        surface already exists on an edge, a protocol error is raised.
      </description>
      <arg name="surface" type="object" interface="wl_surface"/>
      <arg name="output" type="object" interface="wl_output"/>
      <arg name="edge" type="uint" enum="edge"/>
    </request>

    <request name="activate_app">
      <description summary="make client current window">
        Ask the compositor to make a toplevel to become the current/focused
        window for window management purposes.

        See xdg_toplevel.set_app_id from the xdg-shell protocol for a
        description of app_id.

        If multiple toplevels have the same app_id, the result is unspecified.

        XXX: Do we need feedback to say it didn't work? (e.g. client does
        not exist)
      </description>
      <arg name="app_id" type="string"/>
      <arg name="output" type="object" interface="wl_output"/>
    </request>

    <event name="bound_ok" since="2">
     <description summary="event sent if binding was ok">
        Informs the client that it was able to bind the agl_shell
        interface succesfully. Clients are required to wait for this
        event before continuing further.
     </description>
    </event>

    <event name="bound_fail" since="2">
      <description summary="event sent if binding was nok">
        Informs the client that binding to the agl_shell interface was
        unsuccesfull. Clients are required to wait for this event for
        continuing further.
      </description>
    </event>

    <request name="destroy" type="destructor" since="2">
      <description summary="destroys the factory object">
      </description>
    </request>

    <event name="app_state" since="3">
      <description summary="event sent when an application suffered state modification">
        Informs the client that an application has changed its state to another,
        specified by the app_state enum. Client can use this event to track
        current application state. For instance to know when the application has
        started, or when terminated/stopped.
      </description>
      <arg name="app_id" type="string"/>
      <arg name="state" type="uint" enum="app_state"/>
    </event>

    <request name="set_activate_region" since="4">
      <description summary="sets a specific region to activate">
      A hint for the compositor to use a custom area, rather than
      inferring the activation area. If any panels are used
      the compositor computes the activation area by subtracting the
      panels geometry area. If no panels are used then the entire output
      is being used. This request changes that as to hint the compositor
      to use the supplied rectangle and ignore any potential panels
      that might been set-up previously.

      In order for this request to take effect it will need to happen
      before the 'ready' request in order for the compositor to make use of it.
      Note that any 'set_panel' request be will not be honored, if this request
      has been called.

      The x and y coordinates use the top-left corner as the origin. The
      rectangle area shouldn't exceed the output area, while an area smaller
      than the output, would basically result in showing up the background
      surface.
      </description>
      <arg name="output" type="object" interface="wl_output"/>
      <arg name="x" type="int" summary="x position of rectangle"/>
      <arg name="y" type="int" summary="y position of rectangle"/>
      <arg name="width" type="int" summary="width of rectangle"/>
      <arg name="height" type="int" summary="height of rectangle"/>
    </request>

    <request name="deactivate_app" since="5">
      <description summary="de-activate/hide window identified by app_id">
        Ask the compositor to hide the toplevel window for window
        management purposes. Depending on the window role, this request
        will either display the previously active window (or the background
        in case there's no previously active surface) or temporarily (or
        until a 'activate_app' is called upon) hide the surface.

        All the surfaces are identifiable by using the app_id, and no actions
        are taken in case the app_id is not/was not present.

        See xdg_toplevel.set_app_id from the xdg-shell protocol for a
        description of app_id.
      </description>
      <arg name="app_id" type="string"/>
    </request>

    <request name="set_app_float" since="6">
      <description summary="set the window identified by app_id as float">
        Makes the application identified by app_id as floating. If the
        application's window is already mapped, in a maximized, normal state,
        it would transition to the float state.

        For applications that want to modify their own state, this request
        must be done before the initial surface commit in order to take effect.

        If the application is already in floating state, this request wouldn't
        do anything.

        There's no persistence of this request, once the application terminated
        you'll to issue this request again for that particular app_id.

        The x, and y values would be initial position of the window where the
        window surface will be placed.

        See xdg_toplevel.set_app_id from the xdg-shell protocol for a
        description of app_id.
      </description>
      <arg name="app_id" type="string"/>
      <arg name="x" type="int" summary="x position"/>
      <arg name="y" type="int" summary="y position"/>
    </request>

    <request name="set_app_normal" since="6">
      <description summary="set the window identified by app_id as normally started">
      Returns the application identified by app_id as it was in the normal state.
      This is useful to come back from other states to the maximized state, the
      normal state applications are started.
      </description>
      <arg name="app_id" type="string"/>
    </request>
  </interface>

  <interface name="agl_shell_ext" version="1">
    <description summary="extended user interface for Automotive Grade Linux platform">
      This interface allows another client bind to the agl_shell interface,
      while there's another shell client already present.

      The client should first bind to this interface and then inform the
      compositor with the 'doas_shell_client' request and it wants to bind to
      the agl_shell interface. The client is still expected, if using a new
      version of the agl_shell interface, to wait for the 'bound_ok' and
      'bound_fail' events before issueing any other requests/events.

      Note that this interface has its limitations, and the compositor would
      still refuse the act for 'set_panel' or 'set_background' requests
      of the agl_shell interface if there's already a client that used them.

      Any other requests or events should be delievered and handled as it would
      a client bound to the agl_shell interface.
    </description>

    <enum name="doas_shell_client_status">
      <entry name="success" value="0"/>
      <entry name="failed" value="1"/>
    </enum>

    <request name="destroy" type="destructor">
      <description summary="destroys the factory object">
        Call the destructor once you're ready with agl_shell_ext interface.
        This would reset the state and would make any requests made
        on the agl_shell interface be terminated. The client would need 
        to bind again the agl_shell_ext and issue a 'doas_shell_client'
        request.
      </description>
    </request>

    <request name="doas_shell_client">
      <description summary="Informs the compositor it wants to bind to the
      agl_shell interface">
        Prior to binding to agl_shell interface, this request would inform
        the compositor that it wants to gain access the agl_shell interface.
        The client is expected to wait for 'doas_shell_client_done' event and 
        check for a successful status before going further with binding to
        the agl_shell interface.
      </description>
    </request>

    <event name="doas_done">
      <description summary="event sent as a reply to doas_shell_client">
        The client should check the status event to verify that the
        compositor was able to handle the request.
      </description>
      <arg name="status" type="uint" enum="doas_shell_client_status"/>
    </event>
  </interface>
</protocol>