A history of changes to this document can be found at https://github.com/w3c/pointerlock/commits/gh-pages/index.html
Summary of changes since W3C Recommendation 27 October 2016:
The Pointer Lock API provides applications the ability to directly interpret mouse movements as an input method, rather than being limited to only read the position of the mouse cursor. A popular example is that of first person movement controls in three dimensional graphics applications such as games: movement of the mouse is interpreted to control the rotation/direction of the player's camera; no mouse cursor is displayed, and the movement is not limited to the traditional boundaries (such as the user agent's window, or the overall screen) that the mouse cursor is usually subject to, meaning that any mouse movements can be tracked indefinitely in any direction.
Pointer Lock is related to Mouse Capture [[MDN-SETCAPTURE]] (Mouse Capture is unspecified: bug 14600). Capture provides continued event delivery to a target element while a mouse is being dragged, but ceases when the mouse button is released. Pointer Lock differs by being persistent, not limited by screen boundaries, sending events regardless of mouse button state, hiding the cursor, and not releasing until an API call or specific default unlock gesture by the user.
Pointer Lock deals with capturing a single resource and relating it to a single element. This is similar to the Fullscreen API [[FULLSCREEN]], which promotes a single element to be full screen. The Pointer Lock API chooses to pattern the resource capture, state change, and release API as closely as possible after the Fullscreen API.
The Pointer Lock interaction mode was previously referred to as mouse lock. The name was changed as many different controller types besides mice can manipulate the on screen pointing cursor, and they are all impacted.
The [=pointer lock state=] is the state where a single DOM element, which we will call the pointer-lock target, receives all mouse events and the cursor is hidden.
Once in the [=pointer lock state=] the [=user agent=] has a [=pointer-lock target=], a pointer-lock options, which is a {{PointerLockOptions}} and a cursor position which is a pair of numbers representing the location of the system mouse cursor when the Pointer Lock State was entered (the same location that is reported in `screenX`, `screenY`). The [=pointer-lock target=] receives all relevant user generated {{MouseEvent}} events: namely, all user-generated `mousemove`, `mousedown`, `mouseup`, `click`, `dblclick`, `auxclick`, and `wheel` [[ui-events]]. No other elements receive these events while in [=pointer lock state=]. There will be no dispatching of events that require the concept of a mouse cursor: namely, `mouseenter`, `mouseleave`, `mouseover`, `mouseout`, `drag`, and `drop`.
While in the [=pointer lock state=] if the [=pointer-lock options=]' {{PointerLockOptions/unadjustedMovement}} member is `true`, the event coordinates will not be affected by the underlying platform behaviors such as mouse acceleration. In other words, the user agent uses the APIs provided by the underlying platform to guarantee getting the raw events. If the {{PointerLockOptions}}' {{PointerLockOptions/unadjustedMovement}} member is `false`, the user agent relies on the default behavior of the underlying platform regarding the mouse acceleration.
In the [=pointer lock state=], the system mouse cursor is hidden and the window is prevented from losing focus, regardless of mouse movement or button presses. This is directly or indirectly achieved utilizing the underlying operating system API.
Synthetic mouse events created by application script act the same regardless of lock state.
dictionary PointerLockOptions {
boolean unadjustedMovement = false;
};
The options dictionary to customize how the pointer behaves in the locked mode.
If this value is set to `true`, then the pointer movements will not be affected by the underlying platform modifications such as mouse accelaration.
Two events are used to communicate pointer lock state change or an error in changing state. They are named pointerlockchange and pointerlockerror. Refer to algorithm of [[[#requestPointerLock]]] for detail.
Magnification software increases the size of content on the screen. It uses the mouse to move the magnified point of focus around. When a pointer lock is initiated, the magnification software needs to switch to using the keyboard to move the magnified point of focus around instead. When a {{pointerlockchange}} event is fired, web browsers therefore need to make sure the event is communicated to assistive technologies like screen magnifiers.
The process of exiting pointer lock, given an |element:Element|, is as follows:
The {{Element}} interface is extended to provide the ability to request the pointer be locked.
partial interface Element {
Promise<undefined> requestPointerLock(optional PointerLockOptions options = {});
};
const lock_element = document.getElementById("lock_element");
const lock_button = document.getElementById("lock");
lock_button.addEventListener("click", async (event) => {
try {
await lock_element.requestPointerLock({ unadjustedMovement: true });
console.log("successfully locked!");
} catch (e) {
console.log("lock failed with error: ", e);
}
});
A [=parallel queue=] named as lock requests queue is used for queuing all requests. When requestPointerLock() is invoked, perform the following steps:
partial interface Document {
attribute EventHandler onpointerlockchange;
attribute EventHandler onpointerlockerror;
undefined exitPointerLock();
};
An [=event handler idl attribute=] for {{Document/pointerlockchange}} events.
An [=event handler idl attribute=] for {{Document/pointerlockerror}} events.
partial interface mixin DocumentOrShadowRoot {
readonly attribute Element? pointerLockElement;
};
While the pointer is locked, returns the result of [=retargeting=] the element, which is the target for mouse events, against [=this=] element if the result and [=this=] element are in the same tree, otherwise returns null.
Returns null if lock is pending or if pointer is unlocked.
<body>
<div id="host1">
<shadow-root id="root1">
<canvas id="canvas1"></canvas>
</shadow-root>
</div>
<div id="host2">
<shadow-root id="root2">
<canvas id="canvas2"></canvas>
</shadow-root>
</div>
</body>
The example uses fictional `shadow-root` element to denote a [=Element/shadow root=] instance.
If `#canvas1` is the target, `document.pointerLockElement` returns `#host1`, and `root1.pointerLockElement` returns `#canvas1`. The result of [=retargeting=] `#canvas1` against `#root2` is `#host1`, but as `#host1` is not in the same tree as `#root2`, null will be returned for `root2.pointerLockElement`.
partial interface MouseEvent {
readonly attribute double movementX;
readonly attribute double movementY;
};
The attributes {{movementX}} and {{movementY}} must provide the change in position of the pointer, as if the values of `screenX`, `screenY`, were stored between two subsequent `mousemove` events `eNow` and `ePrevious` and the difference taken `movementX = eNow.screenX - ePrevious.screenX`.
{{movementX}} and {{movementY}} must be zero for all mouse events except `mousemove`. All motion data must be delivered via `mousemove` events such that between any two mouse events `earlierEvent` and `currentEvent` the value of `currentEvent.screenX - earlierEvent.screenX` is equivalent to the sum of all {{movementX}} in the events after `earlierEvent`, with the exception of when screenX can not be updated because the pointer is clipped by the [=user agent=] screen boundaries.
{{movementX}} and {{movementY}} must be updated regardless of pointer lock state.
When unlocked, the system cursor can exit and re-enter the [=user agent=] window. If it does so and the [=user agent=] was not the target of operating system mouse move events then the most recent pointer position will be unknown to the [=user agent=] and {{movementX}} / {{movementY}} can not be computed and must be set to zero.
When pointer lock is enabled `clientX`, `clientY`, `screenX`, and `screenY` must hold constant values as if the pointer did not move at all once pointer lock was entered. But {{movementX}} and {{movementY}} must continue to provide the change in position of the pointer as when the pointer is unlocked. There will be no limit to {{movementX}} and {{movementY}} values if the mouse is continuously moved in a single direction. The concept of the mouse cursor will have been removed, and it will not move off the window or be clamped by a screen edge.
The un-initialized value of {{movementX}} and {{movementY}} must be `0`.
Large movement values must not appear in situations when mouse input is interupted, such as the mouse cursor leaving the window and then re-entering at another location. If a [=user agent=] experiences a gap in receiving mouse input data from the operating system then the next generated `mousemove` event must have {{movementX}} and {{movementY}} set to `0`. These gaps may appear for example when the [=user agent=] receives a mouse leaving event at the window system API. As an exception mouse capture may allow the [=user agent=] to continue receiving mouse events when the cursor moves outside the window.
partial dictionary MouseEventInit {
double movementX = 0;
double movementY = 0;
};
A default unlock gesture must always be available that will [=exit pointer lock=] with the [=user agent=]'s [=pointer-lock target=].
Pointer lock must be exited if the [=pointer-lock target=] becomes disconnected, or the [=user agent=], window, or tab loses focus. Moving focus between elements of [=navigable/active document=], including between [=Document/browsing contexts=] , does not [=exit pointer lock=]. E.g. using the keyboard to move focus between contents of frames or iframes will not exit.
Pointer lock must not be exited when fullscreen [[FULLSCREEN]] is entered or exited unless the pointer is required to enable interaction with the [=user agent=] graphical user interface, the [=default unlock gesture=] was used to exit both fullscreen and pointer lock, or window or tab focus was lost.
A player on a first/third person game will need to control the view-port orientation. A widely used method is the use of mouse movements to control the viewing angle. This kind of application can use the Pointer Lock API to allow a complete freedom of control over the viewport's yaw and pitch even when the user is not pressing mouse buttons. Those buttons can be used for other actions while constantly providing navigation via mouse movement.
Users of a three dimensional modeling application will need to rotate models. A application can use the Pointer Lock API to enable the author to rotate the model freely in a drag operation without limiting motion. Without pointer lock a drag would stop providing motion data when the mouse cursor is limited by the edge of the screen.
Similarly, absolute motion panning of a large two dimensional image could be permitted in a single drag operation without cursor / screen limits.
A player on a fast reflexes game controls a paddle to bounce back a ball to the opponent, while allowing the same paddle to execute actions based on different mouse buttons being pressed. The application can use the Pointer Lock API to allow the player to react quickly without being concerned about the mouse cursor leaving the game play area and clicking another system application, thus breaking the game flow.
When modifying numerically magnitudes in applications sometimes the user will prefer to "drag" a numeric control by its button handles to increment or decrement the numeric value. E.g. a spinner with a number entry text box and arrows pointing up and down that can be clicked or dragged on to change the value. An application could use the Pointer Lock API to allow modifying the numeric values beyond what the logical screen bounds allow. The same could apply for a control that fast forwards or rewinds a video or audio stream like a "jog".
Some games use a classical cursor, however they want it to be limited or controlled in some manner. E.g. limited to the bounds of the game, or movable by the game. Locking the pointer enables this if the application creates their own cursor. However HTML and DOM should still be available to use for user interface. Synthetic mouse events should be permitted to allow an application defined cursor to interact with DOM. E.g. the following code should permit a custom cursor to send click events while the pointer is locked:
document.addEventListener("click", function (e) {
if (e._isSynthetic)
return;
// send a synthetic click
var ee = document.createEvent("MouseEvents");
ee._isSynthetic = true;
x = myCursor.x;
y = myCursor.y;
ee.initMouseEvent("click", true, true, null, 1,
x + e.screenX - e.clientX,
y + e.screenY - e.clientY,
x,
y);
var target = document.elementFromPoint(x, y);
if (target)
target.dispatchEvent(ee);
});
Note that synthetic clicks may not be permitted by a [=user agent=] to produce the same default action as a non-synthetic click. However, application handlers can still take action and provide user interface with existing HTML & DOM mechanisms.
Real Time Strategy games often use this technique. When the player moves the pointer to the view-port borders, if they "push" the border with a mouse movement, the view-port is panned over the game area according to how much they move the mouse. When moving the mouse cursor within the bounds of the view port it acts at is typically would on a system. Applications may choose to implement this using pointer lock and the previous use case of "Synthetic cursor interaction with HTML DOM UI" to bring cursor behavior completely under their control.
Games that use pointer lock may desire a traditional UI and system cursor while players prepare in a game lobby. Games usually start after a short timer when all players are ready. Ideally the game could then switch to pointer lock mode without a [=user activation=]. Players should be able to seamlessly move from the game lobby into game navigation.
Game portals, and other sites such as Facebook and Google Plus, host games for users to play. These games may be hosted and served from a different origin from that of the portal site. Embedded games should be able to lock the pointer, even in non-full screen mode.
Security Concerns:
Responses:
Recommendations:
Pointer lock is a required user interaction mode for certain application types, but carries a usability concern if maliciously used. An attacker could remove the ability for a user to control their mouse cursor on their system. [=user agents=] will prevent this by always providing a mechanism to [=exit pointer lock=], by informing the user of how, and by limiting how pointer lock can be entered.
[=user agents=] will determine their own appropriate policies, which may be specialized per device or differ based on user options.
Mouse Capture [[MDN-SETCAPTURE]] handles low security risk mouse event target lock for the duration of a mouse drag gesture. Pointer lock removes the concept of the cursor and directs all events to a given target. They are related, but different.
If a browser implemented both, it would be reasonable to support a combination of traits: The security simplicity of "automatically release lock when mouse up" and the increased functionality of total control over mouse input and removal of the system cursor. The security trait would allow more permissive use of the feature for applications that only required a short burst of pointer lock during a drag event.
This functionality is omitted from the initial version of this spec because it helps the minor use cases in windowed mode but we still do not have an implementation solving the major ones. And, to implement this a browser must implement both, which none does yet. It is not clear if this feature should live on .lock or on .setCapture. If both were implemented, either API could be augmented fairly easily to offer the hybrid functionality.
Even in non locked state, the delta values of mouse movement are useful. Changing the meaning of .client or .screen based on lock state would also cause easy errors in code not carefully monitoring the lock state.
When the pointer is locked 'wheel' events should be sent to the [=pointer-lock target=] element just as 'mousemove' events are. There is a naming conflict with .deltaX/Y/Z as defined in DOM 3 'wheel' event.
There are good motivations to provide a more fine grained approach. E.g. the use case "View-port panning by moving a mouse cursor against the bounds of a view-port" doesn't require hiding the mouse cursor, only bounding it and always having delta values available. Also, this specification defines the movement deltas to be taken from how the system mouse cursor moves, which incorporates operating system filtering and acceleration of the mouse movement data. Applications may desire access to a more raw form of movement data prior to adjustments appropriate for a mouse cursor. Also, raw data may provide better than pixel level accuracy for movement, as well as higher frequency updates. Providing the raw delta movement would also not require special permission or mode from a user, and for some set of applications that do not require bounding the cursor may reduce the security barriers and prompts needed.
There are two justifications for postponing this finer grained approach. The first is a concern of specifying what units mouse movement data are provided in. This specification defines .movementX/Y precisely as the same values that could be recorded when the mouse is not under lock by changes in .screenX/Y. Implementations across multiple [=user agents=] and operating systems will easily be able to meet that requirement and provide application developers and users with a consistent experience. Further, users are expected to have already configured the full system of hardware input and operating system options resulting in a comfortable control the system mouse cursor. By specifying .movementX/Y in the same units mouse lock API applications will be instantly usable to all users because they have already settled their preferences.
Secondly, the implementation of providing movement data and bounding the mouse cursor is more difficult in the fine grained approach. Bundling the features together gives implementations freedom to use a variety of techniques as appropriate on each operating system and is more practical to implement. Direct APIs do not exist on major desktop platforms (Windows, Mac OS X, Linux) to bound the cursor to a specific rectangle, and prototypes have not yet been developed to demonstrate building that behavior by e.g. invisible windows with Xlib or manual cursor movement on Mac. Unaccelerated Delta values have been proposed to be accessed by reading raw Human Interface Device (HID) data. E.g. WM_INPUT messages on Windows, and USB device APIs on Mac OS X / Linux. The challenge here is interpreting and normalizing the units to some consistent and specifiable scale. Also, most APIs considered to date are limited to USB devices.
It would be reasonable to consider adding these capabilities in the future, as the currently specified pointer lock API would be easy to continue to support if the finer grained delta and confinement features were implemented.
The bundled API is selected for implementation practicality, because the desired use cases are supported, and because it will not conflict with future improvements as discussed here.
Not yet, for the same reasons in the previous question: "Why bundle all functionality (hiding cursor, providing mouse deltas) instead of using CSS to hide the cursor, always providing delta values, and offering an API to restrict the cursor movement to a portion of the web page?".
When under pointer lock many mouse events remain relevant, e.g. click, mousedown, etc. These all share the same event data structure MouseEvent. If movement data were reported via a new data structure then a new event would be needed for reporting delta movement. The new data structure would have many parallels to MouseEvent to offer the same conveniences, e.g. button and modifier key states. When handling click, down, and up events would the existing mousedown, mouseup be used? If so, they would provide .clientX/Y and .screenX/Y with no useful data, but would lack the convenience of containing the current movement data. Or, new events would also be required for when the mouse is locked.
Also, movementX/Y are convenient even when the mouse is not locked. This spec requires movement members to always be valid, even when the mouse cursor exists. This reduces code required to track the last cursor state and mouseover/mouseout transitions if applications wish to make use of delta motion of the mouse.
The only negative of adding movementX/Y to MouseEvent appears to be the unused values in clientX/Y and screenX/Y when under pointer lock. This does not seem to be a significant problem.
Therefore the minimal change to add movementX/Y to MouseEvent is selected to reduce API and implementation complexity.
Consider a game with a 3D view controlled by moving the mouse cursor, while the user may still chat with other users via a text console. It is reasonable for the application to accept text input to an element that is different than where mouse events are being dispatched. This is similar to pre-existing behavior of receiving mousemove events over any element while typing into a form on a page.
This specification defines conformance criteria that apply to a single product: the user agent that implements the interfaces that it contains.
Many thanks to lots of people who made contributions to the discussions of this specification: