Resize Observer

1. Introduction

This section is non-normative.

Responsive Web Components need to respond to Element's size changes. An example is an Element that displays a map:

• it displays a map by tiling its content box with Element tiles.

• when resized, it must redo the tiling.

Responsive Web Applications can already respond to viewport size changes. This is done with CSS media queries, or window.resize event.

The ResizeObserver API is an interface for observing changes to Element’s size. It is an Element's counterpart to window.resize event.

ResizeObserver’s notifications can be used to respond to changes in Element's size. Some interesting facts about these observations:

• Observation will fire when watched Element is inserted/removed from DOM.

• Observation will fire when watched Element display gets set to none.

• Observations do not fire for non-replaced inline Elements.

• Observations will not be triggered by CSS transforms.

• Observation will fire when observation starts if Element is being rendered, and Element’s size is not 0,0.

<canvas id="elipse" style="display:block"></canvas>
<div id="menu" style="display:block;width:100px">
    <img src="hamburger.jpg" style="width:24px;height:24px">
    <p class="title">menu title</p>
</div>

// In response to resize, elipse paints an elipse inside a canvas
document.querySelector('#elipse').handleResize = entry => {
    entry.target.width = entry.borderBoxSize[0].inlineSize;
    entry.target.height = entry.borderBoxSize[0].blockSize;
    let rx = Math.floor(entry.target.width / 2);
    let ry = Math.floor(entry.target.height / 2);
    let ctx = entry.target.getContext('2d');
    ctx.beginPath();
    ctx.ellipse(rx, ry, rx, ry, 0, 0, 2 * Math.PI);
    ctx.stroke();
}
// In response to resize, change title visibility depending on width
document.querySelector('#menu').handleResize = entry => {
    let title = entry.target.querySelector(".title")
    if (entry.borderBoxSize[0].inlineSize < 40)
        title.style.display = "none";
    else
        title.style.display = "inline-block";
}

var ro = new ResizeObserver( entries => {
  for (let entry of entries) {
    let cs = window.getComputedStyle(entry.target);
    console.log('watching element:', entry.target);
    console.log(entry.contentRect.top,' is ', cs.paddingTop);
    console.log(entry.contentRect.left,' is ', cs.paddingLeft);
    console.log(entry.borderBoxSize[0].inlineSize,' is ', cs.width);
    console.log(entry.borderBoxSize[0].blockSize,' is ', cs.height);
    if (entry.target.handleResize)
        entry.target.handleResize(entry);
  }
});
ro.observe(document.querySelector('#elipse'));
ro.observe(document.querySelector('#menu'));

2. Resize Observer API

2.1. ResizeObserver interface

The ResizeObserver interface is used to observe changes to Element's size.

It is modeled after MutationObserver and IntersectionObserver.

enum ResizeObserverBoxOptions {
    "border-box", "content-box", "device-pixel-content-box"
};

ResizeObserver can observe different kinds of CSS sizes:

• border-box : size of box border area as defined in CSS2.

• content-box : size of content area as defined in CSS2.

• device-pixel-content-box : size of content area as defined in CSS2, in device pixels, before applying any CSS transforms on the element or its ancestors. This size must contain integer values.

The device-pixel-content-box can be approximated by multiplying devicePixelRatio by the content-box size. However, due to browser-specific subpixel snapping behavior, authors cannot determine the correct way to round this scaled content-box size. How a UA computes the device pixel box for an element is implementation-dependent. One possible implementation could be to multiply the box size and position by the device pixel ratio, then round both the resulting floating-point size and position of the box to integer values, in a way that maximizes the quality of the rendered output.

Note that this size can be affected by position changes to the target, and thus is typically more expensive to observe than the other sizes.

dictionary ResizeObserverOptions {
    ResizeObserverBoxOptions box = "content-box";
};

This section is non-normative. An author may desire to observe more than one CSS box. In this case, author will need to use multiple ResizeObservers.

// Observe the content-box
ro.observe(document.querySelector('#menu'), { box: 'content-box' });

// Observe just the border box. Replaces previous observation.
ro1.observe(document.querySelector('#menu'), { box: 'border-box' });

This does not have any impact on which box dimensions are returned to the defined callback when the event is fired, it solely defines which box the author wishes to observe layout changes on.

[Exposed=(Window),
 Constructor(ResizeObserverCallback callback)]
interface ResizeObserver {
    void observe(Element target, optional ResizeObserverOptions options);
    void unobserve(Element target);
    void disconnect();
};

2.2. ResizeObserverCallback

callback ResizeObserverCallback = void (sequence<ResizeObserverEntry> entries, ResizeObserver observer);

This callback delivers ResizeObserver's notifications. It is invoked by a broadcast active observations algorithm.

2.3. ResizeObserverEntry

[Exposed=Window]
interface ResizeObserverEntry {
    readonly attribute Element target;
    readonly attribute DOMRectReadOnly contentRect;
    readonly attribute sequence<ResizeObserverSize> borderBoxSize;
    readonly attribute sequence<ResizeObserverSize> contentBoxSize;
    readonly attribute sequence<ResizeObserverSize> devicePixelContentBoxSize;
};

contentRect is from the incubation phase of ResizeObserver and is only included for current web compat reasons. It may be deprecated in future levels.

The box size properties are exposed as sequences in order to support elements that have multiple fragments, which occur in multi-column scenarios. However the current definitions of content rect and border box do not mention how those boxes are affected by multi-column layout. In this spec, there will only be a single ResizeObserverSize returned in the sequences, which will correspond to the dimensions of the first column. A future version of this spec will extend the returned sequences to contain the per-fragment size information.

interface ResizeObserverSize {
    readonly attribute unrestricted double inlineSize;
    readonly attribute unrestricted double blockSize;
};

3. Processing Model

3.1. ResizeObservation example struct

This section is non-normative. ResizeObservation is an example struct that can be used in implementation of Resize Observer. It is being included here in order to help provide clarity during the processing model. It effectively holds observation information for a single Element. This interface is not visible to Javascript.

[Constructor(Element target)
]
interface ResizeObservation {
    readonly attribute Element target;
    readonly attribute ResizeObserverBoxOptions observedBox;
    readonly attribute sequence<ResizeObserverSize> lastReportedSizes;
};

3.2. Internal Slot Definitions

3.2.1. Document

Document has a resizeObservers slot that is a list of ResizeObservers in this document. It is initialized to empty.

3.2.2. ResizeObserver

ResizeObserver has a callback slot, initialized by constructor.

ResizeObserver has an observationTargets slot, which is a list of ResizeObservations. It represents all Elements being observed.

ResizeObserver has a activeTargets slot, which is a list of ResizeObservations. It represents all Elements whose size has changed since last observation broadcast that are eligible for broadcast.

ResizeObserver has a skippedTargets slot, which is a list of ResizeObservations. It represents all Elements whose size has changed since last observation broadcast that are not eligible for broadcast

3.3. CSS Definitions

3.3.1. content rect

• width is content width

• height is content height

• top is padding top

• left is padding left

content width spec does not mention how multi-column layout affects content box. In this spec, content width of an Element inside multi-column is the result of getComputedStyle(element).width. This currently evaluates to width of the first column.

Having content rect position be padding-top/left is useful for absolute positioning of target’s children. Absolute position coordinate space origin is topLeft of the padding rect.

Watching content rect means that:

• observation will fire when watched Element is inserted/removed from DOM.

• observation will fire when watched Element display gets set to none.

• non-replaced inline Elements will always have an empty content rect.

• observations will not be triggered by CSS transforms.

Web content can also contain SVG elements. SVG Elements define bounding box instead of a content box. Content rect for SVGGraphicsElements is a rect whose:

• width is bounding box width

• height is bounding box height

• top and left are 0

3.4. Algorithms

3.4.1. Gather active observations at depth

It computes all active observations for a document. To gather active observations at depth, run these steps:

1. Let depth be the depth passed in.

2. For each observer in resizeObservers run these steps:

   1. Clear observer’s activeTargets, and skippedTargets.

   2. For each observation in observer.observationTargets run this step:

      1. If observation.isActive() is true

         1. Let targetDepth be result of calculate depth for node for observation.target.

         2. If targetDepth is greater than depth then add observation to activeTargets.

         3. Else add observation to skippedTargets.

3.4.2. Has active observations

To determine if Document has active observations run these steps:

1. For each observer in resizeObservers run this step:

   1. If observer.activeTargets is not empty, return true.

2. return false.

3.4.3. Has skipped observations

To determine if Document has skipped observations run these steps:

1. For each observer in resizeObservers run this step:

   1. If observer.skippedTargets is not empty, return true.

2. return false.

3.4.4. Create and populate a ResizeObserverEntry

1. Let this be a new ResizeObserverEntry.

2. Set this.target slot to target.

3. Set this.borderBoxSize slot to result of computing size given target and observedBox of "border-box".

4. Set this.contentBoxSize slot to result of computing size given target and observedBox of "content-box".

5. Set this.devicePixelContentBoxSize slot to result of computing size given target and observedBox of "device-pixel-content-box".

6. Set this.contentRect to logical this.contentBoxSize given target and observedBox of "content-box".

7. If target is not an SVG element do these steps:

   1. Set this.contentRect.top to target.padding top.

   2. Set this.contentRect.left to target.padding left.

8. If target is an SVG element do these steps:

   1. Set this.contentRect.top and this.contentRect.left to 0.

3.4.5. Broadcast active observations

broadcast active observations delivers all active observations in a document, and returns the depth of the shallowest broadcast target depth.

To broadcast active observations for a document, run these steps:

1. Let shallowestTargetDepth be ∞

2. For each observer in document.resizeObservers run these steps:

   1. If observer.activeTargets slot is empty, continue.

   2. Let entries be an empty list of ResizeObserverEntryies.

   3. For each observation in activeTargets perform these steps:

      1. Let entry be the result of running create and populate a ResizeObserverEntry given observation.target.

      2. Add entry to entries.

      3. Set observation.lastReportedSizes to matching entry sizes.

         1. Matching sizes are entry.borderBoxSize if observation.observedBox is "border-box"

         2. Matching sizes are entry.contentBoxSize if observation.observedBox is "content-box"

         3. Matching sizes are entry.devicePixelContentBoxSize if observation.observedBox is "device-pixel-content-box"

      4. Set targetDepth to the result of calculate depth for node for observation.target.

      5. Set shallowestTargetDepth to targetDepth if targetDepth < shallowestTargetDepth

   4. Invoke observer.callback with entries.

   5. Clear observer.activeTargets.

3. Return shallowestTargetDepth.

3.4.6. Deliver Resize Loop Error

To deliver resize loop error notification run these steps:

1. Create a new ErrorEvent.

2. Initialize event’s message slot to "ResizeObserver loop completed with undelivered notifications.".

3. Report the exception event.

3.4.7. Calculate depth for node

To calculate depth for node, given a node, run these steps:

1. Let p be the parent-traversal path from node to a root Element of this element’s flattened DOM tree.

2. Return number of nodes in p.

3.4.8. Calculate box size, given target and observed box

This algorithm computes target Element's observed box size. Type of box is described by ResizeObserverBoxOptions. SVG Elements are an exception. SVG size is always its bounding box size, because SVG elements do not use standard CSS box model.

To calculate box size, given target and observedBox, run these steps:

1. If target is an SVGGraphicsElement

   1. Set computedSize.inlineSize to target’s bounding box inline length.

   2. Set computedSize.blockSize to target’s bounding box block length.

2. If target is not an SVGGraphicsElement

   1. If observedBox is "border-box"

      1. Set computedSize.inlineSize to target’s border area inline length.

      2. Set computedSize.blockSize to target’s border area block length.

   2. If observedBox is "content-box"

      1. Set computedSize.inlineSize to target’s content area inline length.

      2. Set computedSize.blockSize to target’s content area block length.

   3. If observedBox is "device-pixel-content-box"

      1. Set computedSize.inlineSize to target’s content area inline length, in integral device pixels.

      2. Set computedSize.blockSize to target’s content area block length, in integral device pixels.

   4. return computedSize.

3.5. ResizeObserver Lifetime

A ResizeObserver will remain alive until both of these conditions are met:

• there are no scripting references to the observer.

• the observer is not observing any targets.

3.6. External Spec Integrations

3.6.1. HTML Processing Model: Event Loop

ResizeObserver processing happens inside the step 7.12 of the HTML Processing Model event loop.

Step 12 is currently underspecified as:

For each fully active Document in docs, update the rendering or user interface of that Document and its browsing context to reflect the current state..

Existing step 12 can be fully specified as:

For each fully active Document in docs, run the following steps for that Document and its browsing contents:

1. Recalc styles

2. Update layout

3. Paint

ResizeObserver extends step 12 with resize notifications. It tries to deliver all pending notifications by looping until no pending notifications are available. This can cause an infinite loop.

Infinite loop is prevented by shrinking the set of nodes that can notify at every iteration. In each iteration, only nodes deeper than the shallowest node in previous iteration can notify.

An error is generated if notification loop completes, and there are undelivered notifications. Elements with undelivered notifications will be considered for delivery in the next loop.

Step 12 with ResizeObserver notifications is:

For each fully active Document in docs, run the following steps for that Document and its browsing context:

1. Recalc styles

2. Update layout

3. Set depth to 0

4. Gather active observations at depth depth for Document

5. Repeat while document has active observations

   2. Set depth to broadcast active observations.

   3. Recalc styles

   4. Update layout

   5. Gather active observations at depth depth for Document

6. If Document has skipped observations then deliver resize loop error notification

7. Update the rendering or user interface of Document and its browsing context to reflect the current state.