Because x-element
has zero dependencies it ships with an integrated template
engine. Developers can choose alternatives according to their preference, e.g.
lit-html
, ReactDOM
, etc.
Add a static template function in your x-element
definition in order to
leverage automagical DOM generation and data binding:
static template(html) {
return ({ options, selectedId }) => {
return html`
<select name="my-options">
${options.map(option => [
option.id,
html`<option value="${option.value}" ?selected="${option.id === selectedId}">`,
])}
</select>
`;
};
}
The following bindings are supported:
Binding | Template | Emulates |
---|---|---|
-- | -- | const el = document.createElement('div'); |
attribute | <div foo="${bar}"></div> |
el.setAttribute('foo', bar); |
attribute (boolean) | <div ?foo="${bar}"></div> |
el.setAttribute('foo', ''); // if “bar” is truthy |
-- | -- | el.removeAttribute('foo'); // if “bar” is falsy |
attribute (defined) | <div ??foo="${bar}"></div> |
el.setAttribute('foo', bar); // if “bar” is non-nullish |
-- | -- | el.removeAttribute('foo'); // if “bar” is nullish |
property | <div .foo="${bar}"></div> |
el.foo = bar; |
content | <div>${foo}</div> |
el.append(document.createTextNode(foo)) // if “bar” is text |
-- | -- | (see content binding for composition) |
Important note on serialization during data binding:
Browsers must serialize values assigned to a DOM node’s attributes using
toString()
. It only makes sense to bind serializable values to attributes
(i.e., String
, Number
, and Boolean
). To avoid [object Object]
surprises,
you can specify the anticipated type of an x-element
property which will
throw a runtime error if it’s ever set to a value of an unanticipated type.
The following template languages are supported:
html
svg
A note on non-primitive data:
Because DOM manipulation is slow — template engines do their best to avoid it
(if a value hasn’t changed, DOM manipulation is skipped). To know if a
non-primitive property has changed, a change-by-reference check is performed
(versus a change-by-value check). Therefore, treat non-primitives as if they’re
immutable. Do this: element.property = { ...element.property, foo: 'bar' }
,
and don’t do this: element.property.foo = 'bar'
.
There are only a handful of bindings. And for each of those bindings, there are only a handful of value types that are appropriate. This section will cover everything by binding, and then by value type.
The basic attribute binding simply calls setAttribute
, verbatim.
const bar = 'something';
html`<div foo="${bar}"></div>`;
// <div foo="something"></div>
const bar = 99;
html`<div foo="${bar}"></div>`;
// <div foo="99"></div>
const bar = undefined;
html`<div foo="${bar}"></div>`;
// <div foo="undefined"></div>
const bar = null;
html`<div foo="${bar}"></div>`;
// <div foo="null"></div>
const bar = true;
html`<div foo="${bar}"></div>`;
// <div foo="true"></div>
const bar = false;
html`<div foo="${bar}"></div>`;
// <div foo="false"></div>
const bar = {};
html`<div foo="${bar}"></div>`;
// <div foo="[object Object]"></div>
The boolean attribute binding will either set the attribute to the empty string or remove the attribute.
const bar = 'something';
html`<div ?foo="${bar}"></div>`;
// <div foo></div>
const bar = 99;
html`<div ?foo="${bar}"></div>`;
// <div foo></div>
const bar = undefined;
html`<div ?foo="${bar}"></div>`;
// <div></div>
const bar = null;
html`<div ?foo="${bar}"></div>`;
// <div></div>
const bar = true;
html`<div ?foo="${bar}"></div>`;
// <div foo></div>
const bar = false;
html`<div ?foo="${bar}"></div>`;
// <div></div>
const bar = {};
html`<div ?foo="${bar}"></div>`;
// <div foo></div>
The defined attribute binding will set the attribute via setAttribute
verbatim
if the value is non-nullish. Otherwise, it will remove the attribute.
const bar = 'something';
html`<div ??foo="${bar}"></div>`;
// <div foo="something"></div>
const bar = 99;
html`<div ??foo="${bar}"></div>`;
// <div foo="99"></div>
const bar = undefined;
html`<div ??foo="${bar}"></div>`;
// <div></div>
const bar = null;
html`<div ??foo="${bar}"></div>`;
// <div></div>
const bar = true;
html`<div ??foo="${bar}"></div>`;
// <div foo="true"></div>
const bar = false;
html`<div ??foo="${bar}"></div>`;
// <div foo="false"></div>
const bar = {};
html`<div ??foo="${bar}"></div>`;
// <div foo="[object Object]"></div>
The basic property binding binds the value to the target element, verbatim.
const bar = 'something';
html`<div .foo="${bar}"></div>`;
// <div></div>
// el.foo = bar;
The content binding does different things based on the value type passed in.
The most basic content binding sets the value as text content.
const bar = 'something';
html`<div>${bar}</div>`;
// <div>something</div>
const bar = 99;
html`<div>${bar}</div>`;
// <div>99</div>
const bar = undefined;
html`<div>${bar}</div>`;
// <div></div>
const bar = null;
html`<div>${bar}</div>`;
// <div></div>
const bar = true;
html`<div>${bar}</div>`;
// <div>true</div>
const bar = false;
html`<div>${bar}</div>`;
// <div>false</div>
const bar = {};
html`<div>${bar}</div>`;
// <div>[object Object]</div>
When the content being bound is itself a template result, you get composition.
const bar = html`<span>something</span>`;
html`<div>${bar}</div>`;
// <div><span>something</span></div>
When the content being bound is an array of template results, you get a list.
const bar = [
html`<span>one</span>`,
html`<span>two</span>`,
];
html`<div>${bar}</div>`;
// <div><span>one</span><span>two</span></div>
// … but, you typically don’t have a static array, you map it idiomatically.
const terms = ['one', 'two'];
const bar = terms.map(term => html`<span>${item}</span>`);
html`<div>${bar}</div>`;
// <div><span>one</span><span>two</span></div>
When the content being bound is an array of key-value map entries (where the
key
is a unique string within the list and the value
is a template result),
you get also list. But, this value will come with some special behavior on top
of the basic array content binding. In particular, it keeps track of each
child node based on the given key
you declare. This enables the template
engine to move child nodes under certain circumstances (versus having to
constantly destroy and recreate). And that shuffling behavior enables authors to
animate DOM nodes across such transitions.
// Note that you can shuffle the deck without destroying / creating DOM.
const deck = [
{ id: 'hearts-one', text: '\u26651' },
// …
{ id: 'clubs-ace', text: '\u2663A' },
];
const items = deck;
const identify = item => item.id;
const template = item => html`<span>${item.text}</span>`;
const bar = items.map(item => [identify(item), template(item)]);
html`<div>${bar}</div>`;
// <div><span>♥1</span>…<span>♣A</span></div>
Following is a working example using lit-html:
// base-element.js
import XElement from 'https://deno.land/x/element/x-element.js';
import { html, render as litRender, svg } from 'https://unpkg.com/[email protected]/lit-html.js';
import { repeat } from 'https://unpkg.com/[email protected]/directives/repeat.js';
export default class BaseElement extends XElement {
static get templateEngine() {
const render = (container, template) => litRender(template, container);
return { render, html, repeat };
}
}
Use it in your elements like this:
// my-custom-element.js
import BaseElement from './base-element.js';
class MyCustomElement extends BaseElement {
static get properties() {
return {
items: {
type: Array,
},
};
}
static template(html, { repeat }) {
return ({ items }) => {
return html`
<div id="container">
${repeat(items, item => item.id, item => html`
<div id="${item.id}">${item.label}</div>
`)}
</div>
`;
};
}
}
customElements.define('my-custom-element', MyCustomElement);
A more complete implementation that incorporates all of the Lit directives can be viewed here.
Because native custom elements are now part of the browser specification it is important to distinguish x-element
from other popular JavaScript frameworks. The manner in which custom elements are defined is framework-agnostic. Here’s more explanation:
- We can register a new custom element
my-custom-element
within the current page context using a native browser API:customElements.define('my-custom-element', MyCustomElement);
- If the features of our custom element are really basic, we could do this easily without any libraries. As your features become more complex some common concerns and conveniences start to emerge (in our case these items became the
x-element
project). - Regardless of the manner in which the element has been defined, the current page context now guarantees a relationship between the new tag
<my-custom-element>
and the classMyCustomElement
. This concept is critical to understand because this normalization liberates developers from the need to choose a single framework (or framework version) to define their features. - Note that it is possible to create a DOM node named
my-custom-element
before the custom element has been defined viacustomElements.define('my-custom-element', MyCustomElement)
. This can be done using declarative HTML like<my-custom-element></my-custom-element>
or with imperative API calls likeconst el = document.createElement('my-custom-element')
. However at this stage themy-custom-element
DOM node is functionally equivalent to aspan
. - When
my-custom-element
is eventually defined within the page context all instances of that element are instantly “upgraded” using theMyCustomElement
class. This is the second important concept: DOM composition is independent from custom element definition. This decoupling enables composable feature developers to have flexibility when selecting a DOM template engine. Because child nodes withinmy-custom-element
can be fully encapsulated using the Shadow DOM creating and managing them becomes an implementation detail that template engines have no awareness of.
Consider the following illustration…
Node composition looks like:
+-- BODY -------------------------------------+
| |
| +-- DIV #root ----------------------+ |
| | | |
| | +-- DIV #component -------+ | |
| | | | | |
| | | DIV #light | | |
| | | | | |
| | +-------------------------+ | |
| | | |
| | +-- MY-CUSTOM-ELEMENT ----+ | |
| | | | | |
| | | DIV #shadow | | |
| | | | | |
| | +-------------------------+ | |
| | | |
| +-----------------------------------+ |
| |
+---------------------------------------------+
The declarative Light DOM representation looks like:
<body>
<div id="root">
<div id="component">
<div id="light"></div>
</div>
<my-custom-element></my-custom-element>
<my-app>
</body>
We can generate these nodes any way we prefer while leveraging my-custom-element
. In this scenario we will use React
and ReactDOM
to accomplish this:
function Example() {
return (
<>
<div id="component">
<div id="light" />
</div>
<my-custom-element />
</>
)
};
const root = document.getElementById('root');
const reactRoot = ReactDOM.createRoot(root);
reactRoot.render(Example);
A working example can be found (here)[../demo/react/]
Because my-custom-element
has no bound properties, the above example works as expected. ReactDOM
will generate and attach <my-custom-element>
to your root just like any other native element. However React 18 and all prior versions contain compatibility issues with custom elements, due to a variety of past design decisions that were deliberated at length here. In short, React’s original property binding and event management system predates the custom element specification. Addressing the incompatibility causes breaking changes to the framework which needed careful consideration.
Fortunately the React team recently announced support for custom elements in its next major version, React 19.
Features distributed as custom elements are framework and library agnostic. Thus, custom elements can integrate with any modern framework. By using native ShadowDOM encapsulation developers can choose the manner in which they manage the DOM while avoiding the risk of vendor lock-in.
Key concepts repeated:
- Custom elements are not a framework (native feature)
- Custom elements provide DOM, JS and CSS encapsulation (native feature)
- Developers can choose a framework to manage the DOM within their custom element
- Developers can choose a framework to manage the DOM that leverages their custom elements
- Developers can work with custom elements without using any framework at all (native feature)
- Developers can mix and match frameworks within the same page context
- It’s all good