This specification describes the method for enabling the author to define and use new types of DOM elements in a document. It is a copy of the relevant parts of [[!HTML]] up through commit 6e17d01489060988451fc6607089ef4e2d00bbfa and [[!WHATWG-DOM]] through commit 182e89f5355cbf90d263f7fffef812e7012dd945. Section titles are prefixed with "HTML:" or "DOM:" to indicate which spec they copy from.

Any point, at which a conforming UA must make decisions about the state or reaction to the state of the conceptual model, is captured as algorithm. The algorithms are defined in terms of processing equivalence. The processing equivalence is a constraint imposed on the algorithm implementors, requiring the output of the both UA-implemented and the specified algorithm to be exactly the same for all inputs.

The IDL fragments in this specification must be interpreted as required for conforming IDL fragments, as described in the Web IDL specification [[!WEBIDL]].

HTML: Custom elements

Introduction

This section is non-normative.

Custom elements provide a way for authors to build their own fully-featured DOM elements. Although authors could always use non-standard elements in their documents, with application-specific behaviour added after the fact by scripting or similar, such elements have historically been non-conforming and not very functional. By defining a custom element, authors can inform the parser how to properly construct an element and how elements of that class should react to changes.

Custom elements are part of a larger effort to "rationalise the platform", by explaining existing platform features (like the elements of HTML) in terms of lower-level author-exposed extensibility points (like custom element definition). Although today there are many limitations on the capabilities of custom elements—both functionally and semantically—that prevent them from fully explaining the behaviours of HTML's existing elements, we hope to shrink this gap over time.

Creating an autonomous custom element

This section is non-normative.

For the purposes of illustrating how to create an autonomous custom element, let's define a custom element that encapsulates rendering a small icon for a country flag. Our goal is to be able to use it like so:

<flag-icon country="nl"></flag-icon>

To do this, we first declare a class for the custom element, extending HTMLElement:

class FlagIcon extends HTMLElement {
  constructor() {
    super();
    this._countryCode = null;
  }

  static get observedAttributes() { return ["country"]; }

  attributeChangedCallback(name, oldValue, newValue) {
    // name will always be "country" due to observedAttributes
    this._countryCode = newValue;
    this._updateRendering();
  }
  connectedCallback() {
    this._updateRendering();
  }

  get country() {
    return this._countryCode;
  }
  set country(v) {
    this.setAttribute("country", v);
  }

  _updateRendering() {
    // Left as an exercise for the reader. But, you'll probably want to
    // check this.ownerDocument.defaultView to see if we've been
    // inserted into a document with a browsing context, and avoid
    // doing any work if not.
  }
}

We then need to use this class to define the element:

customElements.define("flag-icon", FlagIcon);

At this point, our above code will work! The parser, whenever it sees the flag-icon tag, will construct a new instance of our FlagIcon class, and tell our code about its new country attribute, which we then use to set the element's internal state and update its rendering (when appropriate).

You can also create flag-icon elements using the DOM API:

const flagIcon = document.createElement("flag-icon")
flagIcon.country = "jp"
document.body.appendChild(flagIcon)

Finally, we can also use the custom element constructor itself. That is, the above code is equivalent to:

const flagIcon = new FlagIcon()
flagIcon.country = "jp"
document.body.appendChild(flagIcon)
Creating a customized built-in element

This section is non-normative.

Customized built-in elements are a distinct kind of custom element, which are defined slightly differently and used very differently compared to autonomous custom elements. They exist to allow reuse of behaviours from the existing elements of HTML, by extending those elements with new custom functionality. This is important since many of the existing behaviours of HTML elements can unfortunately not be duplicated by using purely autonomous custom elements. Instead, customized built-in elements allow the installation of custom construction behaviour, lifecycle hooks, and prototype chain onto existing elements, essentially "mixing in" these capabilities on top of the already-existing element.

Customized built-in elements require a distinct syntax from autonomous custom elements because user agents and other software key off an element's local name in order to identify the element's semantics and behaviour. That is, the concept of customized built-in elements building on top of existing behaviour depends crucially on the extended elements retaining their original local name.

In this example, we'll be creating a customized built-in element named plastic-button, which behaves like a normal button but gets fancy animation effects added whenever you click on it. We start by defining a class, just like before, although this time we extend HTMLButtonElement instead of HTMLElement:

class PlasticButton extends HTMLButtonElement {
  constructor() {
    super();

    this.addEventListener("click", () => {
      // Draw some fancy animation effects!
    });
  }
}

When defining our custom element, we have to also specify the extends option:

customElements.define("plastic-button", PlasticButton, { extends: "button" });

In general, the name of the element being extended cannot be determined simply by looking at what element interface it extends, as many elements share the same interface (such as q and blockquote both sharing HTMLQuoteElement).

To use our customized built-in element, we use the is attribute on a button element:

<button is="plastic-button">Click Me!</button>

Trying to use a customized built-in element as an autonomous custom element will not work; that is, <plastic-button>Click me?</plastic-button> will simply create an HTMLElement with no special behaviour.

If you need to create a type-extended element programmatically, you can use the following form of createElement():

const plasticButton = document.createElement("button", { is: "plastic-button" });
plasticButton.textContent = "Click me!";

And as before, the constructor will also work:

const plasticButton2 = new PlasticButton();
console.log(plasticButton2.localName);          // will output "button"
console.log(plasticButton2.getAttribute("is")); // will output "plastic-button"

Notably, all the of the ways in which button is special apply to such "plastic buttons" as well: their focus behaviour, ability to participate in form submission, the disabled attribute, and so on.

Drawbacks of autonomous custom elements

This section is non-normative.

As specified below, and alluded to above, simply defining and using an element called taco-button does not mean that such elements represent buttons. That is, tools such as Web browsers, search engines, or accessibility technology will not automatically treat the resulting element as a button just based on its defined name.

To convey the desired button semantics to a variety of users, while still using an autonomous custom element, a number of techniques would need to be employed:

  • The addition of the tabindex attribute would make the taco-button interactive content, thus making it focusable. Note that if the taco-button were to become logically disabled, the tabindex attribute would need to be removed.

  • The addition of various ARIA attributes helps convey semantics to accessibility technology. For example, setting the role attribute to "button" will convey the semantics that this is a button, enabling users to successfully interact with the control using usual button-like interactions in their accessibility technology. Setting the aria-label attribute is necessary to give the button an accessible name, instead of having accessibility technology traverse its child text nodes and announce them. And setting aria-disabled to "true" when the button is logically disabled conveys to accessibility technology the button's disabled state.

  • The addition of event handlers to handle commonly-expected button behaviours helps convey the semantics of the button to Web browser users. In this case, the most relevant event handler would be one that proxies appropriate keydown events to become click events, so that you can activate the button both with keyboard and by clicking.

  • In addition to any default visual styling provided for taco-button elements, the visual styling will also need to be updated to reflect changes in logical state, such as becoming disabled; that is, whatever stylesheet has rules for taco-button will also need to have rules for taco-button[disabled].

With these points in mind, a full-featured taco-button that took on the responsibility of conveying button semantics (including the ability to be disabled) might look something like this:

class TacoButton extends HTMLElement {
  static get observedAttributes() { return ["disabled"]; }

  constructor() {
    super();

    this.addEventListener("keydown", e => {
      if (e.keyCode === 32 || e.keyCode === 13) {
        this.dispatchEvent(new MouseEvent("click", {
          bubbles: true,
          cancelable: true
        }));
      }
    });

    this.addEventListener("click", e => {
      if (this.disabled) {
        e.preventDefault();
        e.stopPropagation();
      }
    });

    this._observer = new MutationObserver(() => {
      this.setAttribute("aria-label", this.textContent);
    });
  }

  connectedCallback() {
    this.setAttribute("role", "button");
    this.setAttribute("tabindex", "0");

    this._observer.observe(this, {
      childList: true,
      characterData: true,
      subtree: true
    });
  }

  disconnectedCallback() {
    this._observer.disconnect();
  }

  get disabled() {
    return this.hasAttribute("disabled");
  }

  set disabled(v) {
    if (v) {
      this.setAttribute("disabled", "");
    } else {
      this.removeAttribute("disabled");
    }
  }

  attributeChangedCallback() {
    // only is called for the disabled attribute due to observedAttributes
    if (this.disabled) {
      this.removeAttribute("tabindex");
      this.setAttribute("aria-disabled", "true");
    } else {
      this.setAttribute("tabindex", "0");
      this.setAttribute("aria-disabled", "false");
    }
  }
}

Even with this rather-complicated element definition, the element is not a pleasure to use for consumers: it will be continually "sprouting" tabindex and aria-* attributes of its own volition. This is because as of now there is no way to specify default accessibility semantics or focus behaviour for custom elements, forcing the use of these attributes to do so (even though they are usually reserved for allowing the consumer to override default behaviour).

In contrast, a simple customized built-in element, as shown in the previous section, would automatically inherit the semantics and behaviour of the button element, with no need to implement these behaviours manually. In general, for any elements with nontrivial behaviour and semantics that build on top of existing elements of HTML, customized built-in elements will be easier to develop, maintain, and consume.

Upgrading elements after their creation

This section is non-normative.

Because element definition can occur at any time, a non-custom element could be created, and then later become a custom element after an appropriate definition is registered. We call this process "upgrading" the element, from a normal element into a custom element.

Upgrades enable scenarios where it may be preferable for custom element definitions to be registered after relevant elements has been initially created, such as by the parser. They allow progressive enhancement of the content in the custom element. For example, in the following HTML document the element definition for img-viewer is loaded asynchronously:

<!DOCTYPE html>
<html lang="en">
<title>Image viewer example</title>

<img-viewer filter="Kelvin">
  <img src="images/tree.jpg" alt="A beautiful tree towering over an empty savannah">
</img-viewer>

<script src="js/elements/img-viewer.js" async></script>

The definition for the img-viewer element here is loaded using a script element marked with the async attribute, placed after the <img-viewer> tag in the markup. While the script is loading, the img-viewer element will be treated as an undefined element, similar to a span. Once the script loads, it will define the img-viewer element, and the existing img-viewer element on the page will be upgraded, applying the custom element's definition (which presumably includes applying an image filter identified by the string "Kelvin", enhancing the image's visual appearance).


Note that upgrades only apply to elements in the document tree. (Formally, elements in a shadow-including document.) An element that is not inserted into a document will stay un-upgraded. An example illustrates this point:

<!DOCTYPE html>
<html lang="en">
<title>Upgrade edge-cases example</title>

<example-element></example-element>

<script>
  "use strict";

  const inDocument = document.querySelector("example-element");
  const outOfDocument = document.createElement("example-element");

  // Before the element definition, both are HTMLElement:
  console.assert(inDocument instanceof HTMLElement);
  console.assert(outOfDocument instanceof HTMLElement);

  class ExampleElement extends HTMLElement {}
  customElements.define("example-element", ExampleElement);

  // After element definition, the in-document element was upgraded:
  console.assert(inDocument instanceof ExampleElement);
  console.assert(!(outOfDocument instanceof ExampleElement));

  document.body.appendChild(outOfDocument);

  // Now that we've moved the element into the document, it too was upgraded:
  console.assert(outOfDocument instanceof ExampleElement);
</script>

Requirements for custom element constructors

When authoring custom element constructors, authors are bound by the following conformance requirements:

Several of these requirements are checked during element creation, either directly or indirectly, and failing to follow them will result in a custom element that cannot be instantiated by the parser or DOM APIs.

Core concepts

A custom element is an element that is custom. Informally, this means that its constructor and prototype are defined by the author, instead of by the user agent. This author-supplied constructor function is called the custom element constructor.

Two distinct types of custom elements can be defined:

  1. An autonomous custom element, which is defined with no extends option. These types of custom elements have a local name equal to their defined name.

  2. A customized built-in element, which is defined with an extends option. These types of custom elements have local name equal to the value passed in their extends option, and their defined name is used as the value of the is attribute.

After a custom element is created, changing the value of the is attribute does not change the element's behaviour, as it is saved on the element as its is value.

Autonomous custom elements have the following element definition:

Categories:
Flow content.
Phrasing content.
Palpable content.
Contexts in which this element can be used:
Where phrasing content is expected.
Content model:
Transparent.
Content attributes:
Global attributes, except the is attribute
Any other attribute that has no namespace (see prose).
DOM interface:
Supplied by the element's author (inherits from HTMLElement)

An autonomous custom element does not have any special meaning: it represents its children. A customized built-in element inherits the semantics of the element that it extends.

Any namespace-less attribute that is relevant to the element's functioning, as determined by the element's author, may be specified on an autonomous custom element, so long as the attribute name is XML-compatible and contains no uppercase ASCII letters. The exception is the is attribute, which must not be specified on an autonomous custom element (and which will have no effect if it is).

Customized built-in elements follow the normal requirements for attributes, based on the elements they extend. To add custom attribute-based behavior, use data-* attributes.


A valid custom element name is a sequence of characters name that meets all of the following requirements:

These requirements ensure a number of goals for valid custom element names:

  • They start with a lowercase ASCII letter, ensuring that the HTML parser will treat them as tags instead of as text.

  • They do not contain any uppercase ASCII letters, ensuring that the user agent can always treat HTML elements ASCII-case-insensitively.

  • They contain a hyphen, used for namespacing and to ensure forward compatibility (since no elements will be added to HTML, SVG, or MathML with hyphen-containing local names in the future).

  • They can always be created with createElement() and createElementNS(), which have restrictions that go beyond the parser's.

Apart from these restrictions, a large variety of names is allowed, to give maximum flexibility for use cases like <math-α> or <emotion-😍>.

A custom element definition describes a custom element and consists of:

A name
A valid custom element name
A local name
A local name
A constructor
A custom element constructor
A prototype
A JavaScript object
A list of observed attributes
A sequence<DOMString>
A collection of lifecycle callbacks
A map, whose three keys are the strings "connectedCallback", "disconnectedCallback", and "attributeChangedCallback". The corresponding values are either a JavaScript function object, or undefined. By default the value of each entry is undefined.
A construction stack
A list, initially empty, that is manipulated by the upgrade an element algorithm and the HTML element constructors. Each entry in the list will be either an element or an already constructed marker.

To look up a custom element definition, given a document, namespace, localName, and is, perform the following steps. They will return either a custom element definition or null:

  1. If namespace is not the HTML namespace, return null.

  2. If document does not have a browsing context, return null.

  3. Let registry be document's associated Window's CustomElementsRegistry object.

  4. If there is custom element definition in registry with name and local name both equal to localName, return that custom element definition.

  5. If there is a custom element definition in registry with name equal to is and local name equal to localName, return that custom element definition.

  6. Return null.

The CustomElementsRegistry interface

Each Window object is associated with a unique instance of a CustomElementsRegistry object, allocated when the Window object is created.

Custom element registries are associated with Window objects, instead of Document objects, since each custom element constructor inherits from the HTMLElement interface, and there is exactly one HTMLElement interface per Window object.

The customElements attribute of the Window interface must return the CustomElementsRegistry object for that Window object.

interface CustomElementsRegistry {
  [CEReactions] void define(DOMString name, Function constructor, optional ElementDefinitionOptions options);
  any get(DOMString name);
  Promise<void> whenDefined(DOMString name);
};

dictionary ElementDefinitionOptions {
  DOMString extends;
};

Every CustomElementsRegistry has a set of custom element definitions, initially empty. In general, algorithms in this specification look up elements in the registry by any of name, local name, or constructor.

Every CustomElementsRegistry also has a set of being-defined names and a set of being-defined constructors, containing valid custom element names and JavaScript function objects, respectively. These are used to ensure that when script executes during element definition, which can reentrantly cause element definition again, the script cannot define another element using the same name or constructor.

Every CustomElementsRegistry also has a when-defined promise map, mapping valid custom element names to promises. It is used to implement the whenDefined() method.

window . customElements . define(name, constructor)
Defines a new custom element, mapping the given name to the given constructor as an autonomous custom element.
window . customElements . define(name, constructor, { extends: baseLocalName })
Defines a new custom element, mapping the given name to the given constructor as a customized built-in element for the element type identified by the supplied baseLocalName. A "NotSupportedError" DOMException will be thrown upon trying to extend a custom element or an unknown element.
window . customElements . get(name)
Retrieves the custom element constructor defined for the given name. Returns undefined if there is no custom element definition with the given name.
window . customElements . whenDefined(name)
Returns a promise that will be fulfilled when a custom element becomes defined with the given name. (If such a custom element is already defined, the returned promise will be immediately fulfilled.) Returns a promise rejected with a "SyntaxError" DOMException if not given a valid custom element name.

Element definition is a process of adding a custom element definition to the CustomElementsRegistry. This is accomplished by the define() method. When invoked, the define(name, constructor, options) method must run these steps:

  1. If IsConstructor(constructor) is false, then throw a TypeError and abort these steps.

  2. If constructor is an interface object whose corresponding interface either is HTMLElement or has HTMLElement in its set of inherited interfaces, throw a TypeError and abort these steps.

    This prevents passing, for example, HTMLButtonElement as constructor, and thus enabling the creation of HTMLButtonElement instances with a local name that is not button.

    Author-defined custom element constructors, created for example via class extends HTMLElement {} or class extends HTMLButtonElement {}, will pass this step without throwing a TypeError, since they are not interface objects.

    Interface objects that do not inherit from HTMLElement (for example TextTrack), or classes defined in the JavaScript specification (such as Set), will pass this step without throwing a TypeError. However, they will later cause create an element to fail.

  3. If name is not a valid custom element name, then throw a "SyntaxError" DOMException and abort these steps.

  4. If this CustomElementsRegistry contains an entry with name name, then throw a "NotSupportedError" DOMException and abort these steps.

  5. If this CustomElementsRegistry's set of being-defined names contains name, then throw a "NotSupportedError" DOMException and abort these steps.

  6. If this CustomElementsRegistry contains an entry with constructor constructor, then throw a "NotSupportedError" DOMException and abort these steps.

  7. If this CustomElementsRegistry's set of being-defined constructors contains constructor, then throw a "NotSupportedError" DOMException and abort these steps.

  8. Let localName be name.

  9. Let extends be the value of the extends member of options, or null if no such member exists.

  10. If extends is not null, then:

    1. If extends is a valid custom element name, then throw a "NotSupportedError" DOMException.

    2. If the element interface for extends and the HTML namespace is HTMLUnknownElement (e.g., if extends does not indicate an element definition in this specification), then throw a "NotSupportedError" DOMException.

    3. Set localName to extends.

  11. Add name to this CustomElementsRegistry's set of being-defined names.

  12. Add constructor to this CustomElementsRegistry's set of being-defined constructors.

  13. Prepare to run script with the current settings object.

  14. Run the following substeps while catching any exceptions:

    1. Let prototype be Get(constructor, "prototype"). Rethrow any exceptions.

    2. If Type(prototype) is not Object, then throw a TypeError exception.

    3. Let connectedCallback be Get(prototype, "connectedCallback"). Rethrow any exceptions.

    4. If connectedCallback is not undefined, and IsCallable(connectedCallback) is false, then throw a TypeError exception.

    5. Let disconnectedCallback be Get(prototype, "disconnectedCallback"). Rethrow any exceptions.

    6. If disconnectedCallback is not undefined, and IsCallable(disconnectedCallback) is false, then throw a TypeError exception.

    7. Let attributeChangedCallback be Get(prototype, "attributeChangedCallback"). Rethrow any exceptions.

    8. Let observedAttributes be an empty sequence<DOMString>.

    9. If attributeChangedCallback is not undefined, then:

      1. If IsCallable(attributeChangedCallback) is false, then throw a TypeError exception.

      2. Let observedAttributesIterable be Get(constructor, "observedAttributes"). Rethrow any exceptions.

      3. If observedAttributesIterable is not undefined, then set observedAttributes to the result of converting observedAttributesIterable to a sequence<DOMString>. Rethrow any exceptions.

    Then, perform the following substeps, regardless of whether the above steps threw an exception or not:

    1. Clean up after running script with the current settings object.

    2. Remove name from this CustomElementsRegistry's set of being-defined names.

    3. Remove constructor from this CustomElementsRegistry's set of being-defined constructors.

    Finally, if the first set of steps threw an exception, then rethrow that exception, and terminate this algorithm. Otherwise, continue onward.

  15. Let definition be a new custom element definition with name name, local name localName, constructor constructor, prototype prototype, observed attributes observedAttributes, and lifecycle callbacks connectedCallback, disconnectedCallback, and attributeChangedCallback (stored by their corresponding name).

  16. Add definition to this CustomElementsRegistry.

  17. Let document be this CustomElementsRegistry's relevant global object's Document object.

  18. Let upgrade candidates be all elements that are shadow-including descendants of document, whose namespace is the HTML namespace and whose local name is localName, in shadow-including tree order. Additionally, if extends is non-null, only include elements whose is value is equal to name.

  19. For each element element in upgrade candidates, enqueue a custom element upgrade reaction given element and definition.

  20. If this CustomElementsRegistry's when-defined promise map contains an entry with key name:

    1. Let promise be the value of that entry.

    2. Resolve promise with undefined.

    3. Delete the entry with key name from this CustomElementsRegistry's when-defined promise map.

When invoked, the get(name) method must run these steps:

  1. If this CustomElementsRegistry contains an entry with name name, then return that entry's constructor.

  2. Otherwise, return undefined.

When invoked, the whenDefined(name) method must run these steps:

  1. If name is not a valid custom element name, then return a new promise rejected with a "SyntaxError" DOMException and abort these steps.

  2. If this CustomElementsRegistry contains an entry with name name, then return a new promise resolved with undefined and abort these steps.

  3. Let map be this CustomElementsRegistry's when-defined promise map.

  4. If map does not contain an entry with key name, create an entry in map with key name and whose value is a new promise.

  5. Let promise be the value of the entry in map with key name.

  6. Return promise.

The whenDefined() method can be used to avoid performing an action until all appropriate custom elements are defined. In this example, we combine it with the :defined pseudo-class to hide a dynamically-loaded article's contents until we're sure that all of the autonomous custom elements it uses are defined.

articleContainer.hidden = true;

fetch(articleURL)
  .then(response => response.text())
  .then(text => {
    articleContainer.innerHTML = text;

    return Promise.all(
      [...articleContainer.querySelectorAll(":not(:defined)")]
        .map(el => customElements.whenDefined(el.localName))
    );
  })
  .then(() => {
    articleContainer.hidden = false;
  });

Upgrades

To upgrade an element, given as input a custom element definition definition and an element element, run the following steps:

  1. For each attribute in element's attribute list, in order, enqueue a custom element callback reaction with element, callback name "attributeChangedCallback", and an argument list containing attribute's local name, null, attribute's value, and attribute's namespace.

  2. If element is currently in a shadow-including document, then enqueue a custom element callback reaction with element, callback name "connectedCallback", and an empty argument list.

  3. If element is custom, abort these steps.

    This can occur due to reentrant invocation of this algorithm, as in the following example:

    <!DOCTYPE html>
    <x-foo id="a"></x-foo>
    <x-foo id="b"></x-foo>
    
    <script>
    // Defining enqueues upgrade reactions for both "a" and "b"
    customElements.define("x-foo", class extends HTMLElement {
      constructor() {
        super();
    
        const b = document.querySelector("#b");
        b.remove();
    
        // While this constructor is running for "a", "b" is still
        // undefined, and so inserting it into the document will enqueue a
        // second upgrade reaction for "b" in addition to the one enqueued
        // by defining x-foo.
        document.body.appendChild(b);
      }
    })
    </script>
    

    This step will thus bail out the algorithm early when upgrade an element is invoked with "b" a second time.

  4. Add element to the end of definition's construction stack.

  5. Let C be definition's constructor.

  6. Let constructResult be Construct(C).

    If C non-conformantly uses an API decorated with the [CEReactions] extended attribute, then the reactions enqueued at the beginning of this algorithm will execute during this step, before C finishes and control returns to this algorithm. Otherwise, they will execute after C and the rest of the upgrade process finishes.

  7. Remove the last entry from the end of definition's construction stack.

    Assuming C calls super() (as it will if it is conformant), and that the call succeeds, this will be the already constructed marker that replaced the element we pushed at the beginning of this algorithm. (The HTML element constructor carries out this replacement.)

    If C does not call super() (i.e. it is not conformant), or if any step in the HTML element constructor throws, then this entry will still be element.

  8. If constructResult is an abrupt completion, then return constructResult (i.e., rethrow the exception).

  9. If SameValue(constructResult.[[\value]], element) is false, then throw an "InvalidStateError" DOMException and terminate these steps.

    This can occur if C constructs another instance of the same custom element before calling super(), or if C uses JavaScript's return-override feature to return an arbitrary object from the constructor.

  10. Set element's custom element state to "custom".

To try to upgrade an element, given as input an element element, run the following steps:

  1. Let definition be the result of looking up a custom element definition given element's node document, element's namespace, element's local name, and element's is value.

  2. If definition is not null, then enqueue a custom element upgrade reaction given element and definition.

Custom element reactions

A custom element possesses the ability to respond to certain occurrences by running author code:

We call these reactions collectively custom element reactions.

The way in which custom element reactions are invoked is done with special care, to avoid running author code during the middle of delicate operations. Effectively, they are delayed until "just before returning to user script". This means that for most purposes they appear to execute synchronously, but in the case of complicated composite operations (like cloning, or range manipulation), they will instead be delayed until after all the relevant user agent processing steps have completed, and then run together as a batch.

Additionally, the precise ordering of these reactions is managed via a somewhat-complicated stack-of-queues system, described below. The intention behind this system is to guarantee that custom element reactions always are invoked in the same order as their triggering actions, at least within the local context of a single custom element. (Because custom element reaction code can perform its own mutations, it is not possible to give a global ordering guarantee across multiple elements.)


Each unit of related similar-origin browsing contexts has a custom element reactions stack, which is initially empty. Each item in the stack is an element queue, which is initially empty as well; the element queue at the top of the stack is called the current element queue. Each item in an element queue is an element. (The elements are not necessarily custom yet, since this queue is used for upgrades as well.)

All elements have an associated custom element reaction queue, initially empty. Each item in the custom element reaction queue is of one of two types:

This is all summarised in the following schematic diagram:

A custom elements reaction stack consists of a stack of element queues. Zooming in on a particular queue, we see that it contains a number of elements (in our example, <x-a>, then <x-b>, then <x-c>). Any particular element in the queue then has a custom element reaction queue. Zooming in on the custom element reaction queue, we see that it contains a variety of queued-up reactions (in our example, upgrade, then attribute changed, then another attribute changed, then connected).

To enqueue a custom element callback reaction, given a custom element element, a callback name callbackName, and a list of arguments args, run the following steps:

  1. Let document be element's node document.

  2. Let definition be the result of looking up a custom element definition given document, element's namespace, element's local name, and element's is value.

    definition will never be null; this algorithm will only be called when such a definition exists.

  3. Let callback be the value of the entry in definition's lifecycle callbacks with key callbackName.

  4. If callback is undefined, then abort these steps.

  5. If callbackName is "attributeChangedCallback", then:

    1. Let attributeName be the first element of args.

    2. If definition's observed attributes does not contain attributeName, then abort these steps.

  6. Add a new callback reaction to element's custom element reaction queue, with callback function callback and arguments args.

  7. Add element to the current element queue.

To enqueue a custom element upgrade reaction, given an element element and custom element definition definition, run the following steps:

  1. Add a new upgrade reaction to element's custom element reaction queue, with custom element definition definition.

  2. Add element to the current element queue.

To invoke custom element reactions in an element queue queue, run the following steps:

  1. For each custom element element in queue:

    1. Let reactions be element's custom element reaction queue.

    2. Repeat until reactions is empty:

      1. Remove the first element of reactions, and let reaction be that element. Switch on reaction's type:

        upgrade reaction
        Upgrade element using reaction's custom element definition.
        callback reaction
        Invoke reaction's callback function with reaction's arguments, and with element as the callback this value.

        If this throws any exception, then report the exception.


To ensure custom element reactions are triggered appropriately, we introduce the [CEReactions] IDL extended attribute. It indicates that the relevant algorithm is to be supplemented with additional steps in order to appropriately track and invoke custom element reactions.

The [CEReactions] extended attribute must take no arguments, and must not appear on anything other than an operation, attribute, setter, or deleter. Additionally, it must not appear on readonly attributes, unless the readonly attribute is also annotated with [PutForwards].

Operations, attributes, setters, or deleters annotated with the [CEReactions] extended attribute must run the following steps surrounding the main algorithm specified for the operation, setter, deleter, or for the attribute's setter:

Before executing the algorithm's steps
Push a new element queue onto the custom element reactions stack.
After executing the algorithm's steps
Pop the element queue from the custom element reactions stack, and invoke custom element reactions in that queue.

The intent behind this extended attribute is somewhat subtle. One way of accomplishing its goals would be to say that every operation, attribute, setter, and deleter on the platform should have these steps inserted, and to allow implementers to optimize away unnecessary cases (where no DOM mutation is possible that could cause custom element reactions to occur).

However, in practice this imprecision could lead to non-interoperable implementations of custom element reactions, as some implementations might forget to invoke these steps in some cases. Instead, we settled on the approach of explicitly annotating all relevant IDL constructs, as a way of ensuring interoperable behavior and helping implementations easily pinpoint all cases where these steps are necessary.

HTML: HTML element constructors

To support the custom elements feature, all HTML elements have special constructor behavior. This is indicated via the [HTMLConstructor] IDL extended attribute. It indicates that the interface object for the given interface will have a specific behavior when called, as defined in detail below.

The [HTMLConstructor] extended attribute must take no arguments, and must not appear on anything other than an interface. It must appear only once on an interface, and the interface must not be annotated with the [Constructor] or [NoInterfaceObject] extended attributes. (However, the interface may be annotated with [NamedConstructor]; there is no conflict there.) It must not be used on a callback interface.

Interface objects for interfaces annotated with the [HTMLConstructor] extended attribute must run the following steps as the function body behavior for both [[\Call]] and [[\Construct]] invocations of the corresponding JavaScript function object. When invoked with [[\Call]], the NewTarget value is undefined, and so the algorithm below will immediately throw. When invoked with [[\Construct]], the [[\Construct]] newTarget parameter provides the NewTarget value.

  1. Let registry be the current global object's CustomElementsRegistry object.

  2. Let definition be the entry in registry with constructor equal to NewTarget. If there is no such definition, then throw a TypeError and abort these steps.

    Since there can be no entry in registry with a constructor of undefined, this step also prevents HTML element constructors from being called as functions (since in that case NewTarget will be undefined).

  3. Let prototype be definition's prototype.

  4. If definition's construction stack is empty, then:

    1. Let element be a new element that implements the interface to which this constructor corresponds, with no attributes, namespace set to the HTML namespace, local name set to definition's local name, and node document set to the current global object's Document object.

    2. Perform element.[[\SetPrototypeOf]](prototype). Rethrow any exceptions.

    3. Set element's custom element state to "custom".

    4. Return element.

    This occurs when author script constructs a new custom element directly, e.g. via new MyCustomElement().

  5. Let element be the last entry in definition's construction stack.

  6. If element is an already constructed marker, then throw an "InvalidStateError" DOMException and abort these steps.

    This can occur when the author code inside the custom element constructor non-conformantly creates another instance of the class being constructed, before calling super():

    let doSillyThing = false;
    
    class DontDoThis extends HTMLElement {
      constructor() {
        if (doSillyThing) {
          doSillyThing = false;
          new DontDoThis();
          // Now the construction stack will contain an already constructed marker.
        }
    
        // This will then fail with an "InvalidStateError" DOMException:
        super();
      }
    }
    

    This can also occur when author code inside the custom element constructor non-conformantly calls super() twice, since per the JavaScript specification, this actually executes the superclass constructor (i.e. this algorithm) twice, before throwing an error:

    class DontDoThisEither extends HTMLElement {
      constructor() {
        super();
    
        // This will throw, but not until it has already called into the HTMLElement constructor
        super();
      }
    }
    
  7. Perform element.[[\SetPrototypeOf]](prototype). Rethrow any exceptions.

  8. Replace the last entry in definition's construction stack with an already constructed marker.

  9. Return element.

    This step is normally reached when upgrading a custom element; the existing element is returned, so that the super() call inside the custom element constructor assigns that existing element to this.


In addition to the constructor behavior implied by [HTMLConstructor], some elements also have named constructors (which are really factory functions with a modified prototype property).

Named constructors for HTML elements can also be used in an extends clause when defining a custom element constructor:

class AutoEmbiggenedImage extends Image {
  constructor(width, height) {
    super(width * 10, height * 10);
  }
}

customElements.define("auto-embiggened", AutoEmbiggeningImage, { extends: "img" });

const image = new AutoEmbiggenedImage(15, 20);
console.assert(image.width === 150);
console.assert(image.height === 200);

Miscellaneous patches

HTML: The Window object

HTML's Window object definition must be extended as follows:

partial interface Window {
    readonly attribute CustomElementsRegistry customElements;
};

HTML: Pseudo-classes

:defined

The :defined pseudo-class must match any element that is defined.

HTML: Creating and inserting nodes

When the HTML parser requires the UA to create an element for a token in a particular given namespace and with a particular intended parent, the UA must run the following steps:

  1. Let document be intended parent's node document.

  2. Let local name be the tag name of the token.

  3. Let is be the value of the "is" attribute in the given token, if such an attribute exists, or null otherwise.

  4. Let definition be the result of looking up a custom element definition given document, given namespace, local name, and is.

  5. If definition is non-null and the parser was not originally created for the HTML fragment parsing algorithm, then let will execute script be true. Otherwise, let it be false.

  6. If will execute script is true, then:

    1. Increment the parser's script nesting level.

    2. Set the parser pause flag to true.

    3. If the JavaScript execution context stack is empty, then perform a microtask checkpoint.

    4. Push a new element queue onto the custom element reactions stack.

  7. Let element be the result of creating an element given document, localName, given namespace, null, and is. If will execute script is true, set the synchronous custom elements flag; otherwise, leave it unset.

    This will cause custom element constructors to run, if will execute script is true. However, even if this causes new characters to be inserted into the tokenizer, the parser will not be executed reentrantly, since the parser pause flag is true. Similarly, blowing away the document is not possible, since the script nesting level is greater than zero.

    If this step throws an exception, then report the exception, and let element be instead a new element that implements HTMLUnknownElement, with no attributes, namespace set to given namespace, namespace prefix set to null, custom element state "undefined", and node document set to document.

  8. Append each attribute in the given token to element.

    This can enqueue a custom element callback reaction for the attributeChangedCallback, which might run immediately (in the next step).

    Even though the is attribute governs the creation of a customized built-in element, it is not present during the execution of the relevant custom element constructor; it is appended in this step, along with all other attributes.

  9. If will execute script is true, then:

    1. Let queue be the result of popping the current element queue from the custom element reactions stack. (This will be the same element queue as was pushed above.)

    2. Invoke custom element reactions in queue.

    3. Decrement the script nesting level by one.

    4. If the parser's script nesting level is zero, then set the parser pause flag to false.

  10. If element has an xmlns attribute in the XMLNS namespace whose value is not exactly the same as the element's namespace, that is a parse error. Similarly, if element has an xmlns:xlink attribute in the XMLNS namespace whose value is not the XLink Namespace, that is a parse error.

  11. If element is a resettable element, invoke its reset algorithm. (This initialises the element's value and checkedness based on the element's attributes.)

  12. If element is a form-associated element, and the form element pointer is not null, and there is no template element on the stack of open elements, and element is either not listed or doesn't have a form attribute, and the intended parent is in the same tree as the element pointed to by the form element pointer, associate element with the form element pointed to by the form element pointer, and suppress the running of the reset the form owner algorithm when the parser subsequently attempts to insert the element.

  13. Return element.

HTML: Parsing XHTML documents

When creating DOM nodes representing elements, the create an element for a token algorithm or some equivalent that operates on appropriate XML datastructures must be used, to ensure the proper element interfaces are created and that custom elements are set up correctly.

HTML: Content model

HTML has several sections that need to be updated when introducing a new element, or in this case class of elements. The necessary changes are:

HTML: Wide-ranging patches

All IDL attributes in HTML which could potentially cause DOM mutations have been updated to be annotated with the [CEReactions] extended attribute.

All HTML interfaces representing a HTML element have been updated to be annotated with the [HTMLConstructor] extended attribute.

All places in HTML that create elements have been updated to use the new create an element algorithm.

DOM: Elements

Elements have an associated namespace, namespace prefix, local name, custom element state, and is value. When an element is created, all of these values are initialized.

An element’s custom element state is one of "undefined", "uncustomized", or "custom". An element whose custom element state is "uncustomized" or "custom" is said to be defined. An element whose custom element state is "custom", is said to be custom.

Whether or not an element is defined is used to determine the behavior of the :defined pseudo-class. Whether or not an element is custom is used to determine the behavior of the mutation algorithms.

The following code illustrates elements in each of these three states:

<!DOCTYPE html>
<script>
  window.customElements.define("sw-rey", class extends HTMLElement {})
  window.customElements.define("sw-finn", class extends HTMLElement {}, { extends: "p" })
  window.customElements.define("sw-kylo", class extends HTMLElement {
    constructor() {
      // super() intentionally omitted for this example
    }
  })
</script>

<!-- "undefined" (not defined, not custom) -->
<sw-han></sw-han>
<sw-kylo></sw-kylo>
<p is="sw-luke"></p>
<p is="asdf"></p>

<!-- "uncustomized" (defined, not custom) -->
<p></p>
<asdf></asdf>

<!-- "custom" (defined, custom) -->
<sw-rey></sw-rey>
<p is="sw-finn"></p>

To create an element, given a document, localName, namespace, and optional prefix, is, and synchronous custom elements flag, run these steps:

  1. If prefix was not given, let prefix be null.

  2. If is was not given, let is be null.

  3. Let result be null.

  4. Let definition be the result of looking up a custom element definition given document, namespace, localName, and is.

  5. If definition is non-null, and definition’s name is not equal to its local name (i.e., definition represents a customized built-in element), then:

    1. Let interface be the element interface for localName and the HTML namespace.

    2. Set result to a new element that implements interface, with no attributes, namespace set to the HTML namespace, namespace prefix set to prefix, local name set to localName, custom element state set to "undefined", is value set to is, and node document set to document.

    3. If the synchronous custom elements flag is set, upgrade element using definition.

    4. Otherwise, enqueue a custom element upgrade reaction given result and definition.

  6. Otherwise, if definition is non-null, then:

    1. If the synchronous custom elements flag is set:

      1. Let C be definition’s constructor.

      2. Set result to Construct(C). Rethrow any exceptions.

      3. If result does not implement the HTMLElement interface, throw a TypeError.

        This is meant to be a brand check to ensure that the object was allocated by the HTMLElement constructor. See webidl #97 about making this more precise.

      4. If result’s attribute list is not empty, then throw a NotSupportedError.

      5. If result has children, then throw a NotSupportedError.

      6. If result’s parent is not null, then throw a NotSupportedError.

      7. If result’s node document is not document, then throw a NotSupportedError.

      8. If result’s namespace is not the HTML namespace, then throw a NotSupportedError.

      9. If result’s local name is not equal to localName, then throw a NotSupportedError.

      10. Set result’s namespace prefix to prefix.

      11. Set result’s is value to null.

    2. Otherwise:

      1. Set result to a new element that implements the HTMLElement interface, with no attributes, namespace set to the HTML namespace, namespace prefix set to prefix, local name set to localName, custom element state set to "undefined", is value set to null, and node document set to document.

      2. Enqueue a custom element upgrade reaction given result and definition.

  7. Otherwise:

    1. Let interface be the element interface for localName and namespace.

    2. Set result to a new element that implements interface, with no attributes, namespace set to namespace, namespace prefix set to prefix, local name set to localName, custom element state set to "uncustomized", is value set to is, and node document set to document.

    3. If document has a browsing context, and namespace is the HTML namespace, and either localName is a valid custom element name or is is is non-null, set result’s custom element state to "undefined".

  8. Return result.

To change an attribute attribute from an element element to value, run these steps:

  1. Queue a mutation record of "attributes" for element with name attribute’s local name, namespace attribute’s namespace, and oldValue attribute’s value.
  2. If element is custom, then enqueue a custom element callback reaction with element, callback name "attributeChangedCallback", and an argument list containing attribute’s local name, attribute’s value, value, and attribute’s namespace.
  3. Run the attribute change steps with element, attribute’s local name, attribute’s value, value, and attribute’s namespace.

  4. Set attribute’s value to value.

To append an attribute attribute to an element element, run these steps:

  1. Queue a mutation record of "attributes" for element with name attribute’s local name, namespace attribute’s namespace, and oldValue null.
  2. If element is custom, then enqueue a custom element callback reaction with element, callback name "attributeChangedCallback", and an argument list containing attribute’s local name, null, attribute’s value, and attribute’s namespace.
  3. Run the attribute change steps with element, attribute’s local name, null, attribute’s value, and attribute’s namespace.

  4. Append the attribute to the element’s attribute list.
  5. Set attribute’s element to element.

To remove an attribute attribute from an element element, run these steps:

  1. Queue a mutation record of "attributes" for element with name attribute’s local name, namespace attribute’s namespace, and oldValue attribute’s value.
  2. If element is custom, then enqueue a custom element callback reaction with element, callback name "attributeChangedCallback", and an argument list containing attribute’s local name, attribute’s value, null, and attribute’s namespace.
  3. Run the attribute change steps with element, attribute’s local name, attribute’s value, null, and attribute’s namespace.

  4. Remove attribute from the element’s attribute list.
  5. Set attribute’s element to null.

To replace an attribute oldAttr by an attribute newAttr in an element element, run these steps:

  1. Queue a mutation record of "attributes" for element with name oldAttr’s local name, namespace oldAttr’s namespace, and oldValue oldAttr’s value.

  2. If element is custom, then enqueue a custom element callback reaction with element, callback name "attributeChangedCallback", and an argument list containing oldAttr’s local name, oldAttr’s value, newAttr’s value, and oldAttr’s namespace.
  3. Run the attribute change steps with element, oldAttr’s local name, oldAttr’s value, newAttr’s value, and oldAttr’s namespace.

  4. Replace oldAttr by newAttr in the element’s attribute list.

  5. Set oldAttr’s element to null.

  6. Set newAttr’s element to element.

DOM: Cloning

To clone a node, with an optional document and clone children flag, run these steps:

  1. If document is not given, let document be node’s node document.

  2. If node is an element, then:

    1. Let copy be the result of creating an element, given document, node’s local name, node’s namespace, node’s namespace prefix, and the value of node’s is attribute if present (or null if not). The synchronous custom elements flag should be unset.

    2. For each attribute in node’s attribute list, in order, run these substeps:

      1. Let copyAttribute be a new attribute.

      2. Set copyAttribute’s namespace, namespace prefix, local name, and value, to those of attribute.

      3. Append copyAttribute to copy.

  3. Otherwise, let copy be a node that implements the same interfaces as node, and fulfills these additional requirements, switching on node:

    Document

    Set copy’s encoding, content type, URL, type, and mode, to those of node.

    DocumentType

    Set copy’s name, public ID, and system ID, to those of node.

    Text
    Comment
    Set copy’s data, to that of node.
    ProcessingInstruction
    Set copy’s target and data to those of node.
    Any other node
  4. Set copy’s node document and document to copy, if copy is a document, and set copy’s node document to document otherwise.

  5. Run any cloning steps defined for node in other applicable specifications and pass copy, node, document and the clone children flag if set, as parameters.
  6. If the clone children flag is set, clone all the children of node and append them to copy, with document as specified and the clone children flag being set.
  7. Return copy.

DOM: Mutation algorithms

To insert a node into a parent before a child, with an optional suppress observers flag, run these steps:

  1. Let count be the number of children of node if it is a DocumentFragment node, and one otherwise.
  2. If child is non-null, run these substeps:
    1. For each range whose start node is parent and start offset is greater than child’s index, increase its start offset by count.
    2. For each range whose end node is parent and end offset is greater than child’s index, increase its end offset by count.
  3. Let nodes be node’s children if node is a DocumentFragment node, and a list containing solely node otherwise.
  4. If node is a DocumentFragment node, remove its children with the suppress observers flag set.
  5. If node is a DocumentFragment node, queue a mutation record of "childList" for node with removedNodes nodes.

    This step intentionally does not pay attention to the suppress observers flag.

  6. For each node in nodes, in tree order, run these substeps:

    1. Insert node into parent before child or at the end of parent if child is null.

    2. If parent is a shadow host and node is a slotable, then assign a slot for node.

    3. If parent is a slot whose assigned nodes is the empty list, then run signal a slot change for parent.

    4. Run assign slotables for a tree with node’s tree and a set containing each inclusive descendant of node that is a slot.

    5. For each shadow-including inclusive descendant inclusiveDescendant of node, in shadow-including tree order, run these subsubsteps:

      1. Run the insertion steps with inclusiveDescendant.

      2. If inclusiveDescendant is in a shadow-including document, then:

        1. If inclusiveDescendant is custom, then enqueue a custom element callback reaction with inclusiveDescendant, callback name "connectedCallback", and an empty argument list.

        2. Otherwise, try to upgrade inclusiveDescendant.

          If this successfully upgrades inclusiveDescendant, its connectedCallback will be enqueued automatically during the upgrade an element algorithm.

  7. If suppress observers flag is unset, queue a mutation record of "childList" for parent with addedNodes nodes, nextSibling child, and previousSibling child’s previous sibling or parent’s last child if child is null.

To remove a node from a parent, with an optional suppress observers flag, run these steps:

  1. Let index be node’s index.
  2. For each range whose start node is an inclusive descendant of node, set its start to (parent, index).
  3. For each range whose end node is an inclusive descendant of node, set its end to (parent, index).
  4. For each range whose start node is parent and start offset is greater than index, decrease its start offset by one.
  5. For each range whose end node is parent and end offset is greater than index, decrease its end offset by one.
  6. For each NodeIterator object iterator whose root’s node document is node’s node document, run the NodeIterator pre-removing steps given node and iterator.

  7. Let oldPreviousSibling be node’s previous sibling.
  8. Let oldNextSibling be node’s next sibling.
  9. Remove node from its parent.
  10. If node is assigned, then run assign slotables for node’s assigned slot.

  11. If parent is a slot whose assigned nodes is the empty list, then run signal a slot change for parent.

  12. If node has an inclusive descendant that is a slot, then:

    1. Run assign slotables for a tree with parent’s tree.

    2. Run assign slotables for a tree with node’s tree and a set containing each inclusive descendant of node that is a slot.

  13. Run the removing steps with node and parent.

  14. If node is custom, then enqueue a custom element callback reaction with node, callback name "disconnectedCallback", and an empty argument list.

    It is intentional for now that custom elements do not get parent passed. This might change in the future if there is a need.

  15. For each shadow-including descendant descendant of node, in shadow-including tree order, run these substeps:

    1. Run the removing steps with descendant.

    2. If descendant is custom, then enqueue a custom element callback reaction with descendant, callback name "disconnectedCallback", and an empty argument list.

  16. For each inclusive ancestor inclusiveAncestor of parent, if inclusiveAncestor has any registered observers whose options' subtree is true, then for each such registered observer registered, append a transient registered observer whose observer and options are identical to those of registered and source which is registered to node’s list of registered observers.
  17. If suppress observers flag is unset, queue a mutation record of "childList" for parent with removedNodes a list solely containing node, nextSibling oldNextSibling, and previousSibling oldPreviousSibling.

DOM: Wide-ranging patches

All IDL attributes in DOM which could potentially cause DOM mutations have been updated to be annotated with the [CEReactions] extended attribute.

All places in DOM that create elements have been updated to use the new create an element algorithm. At the time of this writing, apart from the above createElement() and createElementNS() definitions, only createHTMLDocument() creates elements.

Acknowledgments

David Hyatt developed XBL 1.0, and Ian Hickson co-wrote XBL 2.0. These documents provided tremendous insight into the problem of behavior attachment and greatly influenced this specification.

Alex Russell and his considerable forethought triggered a new wave of enthusiasm around the subject of behavior attachment and how it can be applied practically on the Web.

Dominic Cooney, Hajime Morrita, and Roland Steiner worked tirelessly to scope the problem within the confines of the Web platform and provided a solid foundation for this document.

Steve Faulkner, The Paciello Group, for what formed the basis of the introductory examples.

The <flag-icon> example was inspired by a custom element by Steven Skelton. (MIT)

The editor would also like to thank Alex Komoroske, Andres Rios, Anne van Kesteren, Boris Zbarsky, Daniel Buchner, Edward O'Connor, Erik Arvidsson, Elliott Sprehn, Hayato Ito, Jan Miksovsky, Jonas Sicking, Koji Ishii, Olli Pettay, Rafael Weinstein, Ryosuke Niwa, Scott Miles, Simon Pieters, Steve Orvell, Tab Atkins, Tim Perry, and William Chen for their comments and contributions to this specification.