Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

(engine-restrictions) Lift restrictions on addEventListener's options param for host element #1824

Open
ravijayaramappa opened this issue Apr 13, 2020 · 16 comments
Labels
BUG P3 Up for grabs Issues that are relatively small, self-contained, and ready for implementation

Comments

@ravijayaramappa
Copy link
Contributor

ravijayaramappa commented Apr 13, 2020

Description

Adding an event listener on the custom element with a third options argument logs an error.

This restriction should be on the component instance and not on the host element. From the outside world, a DOM element should behave per spec when addEventListener is invoked with the options param, regardless of whether the element is a standard html element or a native custom element or an LWC host element.

Steps to Reproduce

https://playground.lwcjs.org/projects/d1fzv_d_h/3/edit

import { LightningElement, track } from 'lwc';

export default class App extends LightningElement {
    renderedCallback() {
        this.addEventListener('foo', () => {}, { passive: true });
    }
}

Reference: https://developer.mozilla.org/en-US/docs/Web/API/EventTarget/addEventListener

@caridy
Copy link
Contributor

caridy commented Apr 13, 2020

FYI, this is probably just moving the following block:

addEventListener: generateDataDescriptor({
value(
this: HTMLElement,
type: string,
listener: EventListener,
options?: boolean | AddEventListenerOptions
) {
const vmBeingRendered = getVMBeingRendered();
assert.invariant(
!isInvokingRender,
`${vmBeingRendered}.render() method has side effects on the state of ${toString(
this
)} by adding an event listener for "${type}".`
);
assert.invariant(
!isUpdatingTemplate,
`Updating the template of ${vmBeingRendered} has side effects on the state of ${toString(
elm
)} by adding an event listener for "${type}".`
);
// TODO [#420]: this is triggered when the component author attempts to add a listener
// programmatically into a lighting element node
if (!isUndefined(options)) {
logError(
'The `addEventListener` method in `LightningElement` does not support any options.',
getAssociatedVMIfPresent(this)
);
}
// Typescript does not like it when you treat the `arguments` object as an array
// @ts-ignore type-mismatch
return originalAddEventListener.apply(this, arguments);
},
}),

to

const descriptors: PropertyDescriptorMap = {

@git2gus
Copy link

git2gus bot commented Apr 14, 2020

This issue has been linked to a new work item: W-7446526

@ekashida
Copy link
Member

I'd have to dig in a little more but giving users the ability to subscribe to the capture phase of focusin might require us to refactor our delegatesFocus implementation. It might be fine as long as we always run first, not sure...

@ekashida
Copy link
Member

ekashida commented Jul 6, 2020

Looks like we only restrict this for the host element:
https://playground.lwcjs.org/projects/LY4Z_tvOv/1/edit

@ekashida
Copy link
Member

ekashida commented Mar 28, 2021

When introducing support for options, we'll also need to update the existing logic to prevent multiple listeners for the same event since we currently only consider the identity of the listener function.

Some points to consider when determining whether a listener is "identical":

  1. Only the configuration parameters of the option object is considered and the values are read once when binding (it doesn't matter if the values change over time). This means that passing in true and { capture: true } are equivalent.

  2. IE11 doesn't support option objects, it just cares whether the third argument is truthy or not. We'd need to detect browser behavior to support this since different option objects should be considered equivalent.

@bk-dev
Copy link

bk-dev commented Apr 20, 2021

Hello, there. I'm also seeing this behavior in our project. I am using LWCs as web components in my app and would like to use the third optional parameter when binding event listeners to them. When running in development mode the LWC runtime throws an error for each usage. @pmdartus

@nolanlawson
Copy link
Contributor

Updated repro using StackBlitz for posterity

FWIW it seems to log an error, not throw an error.

Screen Shot 2023-04-07 at 9 00 25 AM

@fafnirical
Copy link

I just ran into this issue when trying to use LWC in Storybook 7, which requires lit-html/lit v2. It doesn't seem to be an issue with lit-html v1, though it still logs the error (but doesn't throw it). Have there been any updates since 2021?

@nolanlawson
Copy link
Contributor

@fafnirical Is the issue just that it's logging an error? Or is there an actual functional difference?

@fafnirical
Copy link

fafnirical commented Apr 10, 2023

@fafnirical Is the issue just that it's logging an error? Or is there an actual functional difference?

@nolanlawson There's an actual functional difference. In addition to the logged error (LWC error]: The `addEventListener` method in `LightningElement` does not support any options.) I see when rendered by both lit-html v1 and v2, lit-html v2 causes LWC to throw the following error in Storybook:

vendors~main.iframe.bundle.js:1475 Uncaught (in promise) TypeError: Invalid second argument for Element.addEventListener() in [object HTMLElement] for event "changesegment". Expected an EventListener but received [object Object].
    at HTMLBridgeElement.addCustomElementEventListener (vendors~main.iframe.bundle.js:1475:13)
    at HTMLBridgeElement.patchedAddEventListener$1 [as addEventListener] (vendors~main.iframe.bundle.js:4318:42)
    at HTMLBridgeElement.value [as addEventListener] (engine-dom.js:1422:41)
    at L._$AI (vendors~main.iframe.bundle.js:72132:84)
    at S.v (vendors~main.iframe.bundle.js:71920:112)
    at M.g (vendors~main.iframe.bundle.js:71991:13)
    at M._$AI (vendors~main.iframe.bundle.js:71964:191)
    at c.setValue (vendors~main.iframe.bundle.js:71166:122)
    at _callee$ (vendors~main.iframe.bundle.js:71624:65)
    at tryCatch (vendors~main.iframe.bundle.js:71540:1357)

The source of the error is https://github.com/salesforce/lwc/blob/v2.43.0/packages/@lwc/synthetic-shadow/src/faux-shadow/events.ts#L210-L214

And the code for the story:

/// myLwcComponent.stories.ts
import { html } from 'lit-html';

const handleEvent = (e: CustomEvent) => console.log(e);

export const example = () => html`<my-lwc-component @changesegment=${handleEvent}></my-lwc-component>`;

@nolanlawson
Copy link
Contributor

I see, so this is a synthetic shadow issue. Synthetic shadow is deprecated, and we had not planned to make any changes to it. But I can understand if you cannot remove synthetic shadow from your storybook testing.

@nolanlawson
Copy link
Contributor

Ah, I also just noticed that this error is only thrown in non-prod mode. Maybe we should tie this into #3245 then.

if (process.env.NODE_ENV !== 'production') {
if (!isFunction(listener)) {
throw new TypeError(
`Invalid second argument for Element.addEventListener() in ${toString(
this
)} for event "${type}". Expected an EventListener but received ${listener}.`
);
}
}

@nolanlawson
Copy link
Contributor

@fafnirical Actually, I don't understand how you're hitting this error. Your error message is:

Invalid second argument for Element.addEventListener() in [object HTMLElement] for event 
"changesegment". Expected an EventListener but received [object Object].

How is the second argument to addEventListener an object and not a function? Per MDN, the second argument should always be a function.

@nolanlawson
Copy link
Contributor

Nevermind, I get it. The second argument can be an object with a handleEvent method. Looks like that is exactly what Lit is using.

@nolanlawson nolanlawson added the Up for grabs Issues that are relatively small, self-contained, and ready for implementation label Aug 11, 2023
@jmsjtu
Copy link
Member

jmsjtu commented May 7, 2024

This issue only logs a warning, it doesn't throw an error. Adding to our trust backlog.

@jmsjtu jmsjtu added BUG P3 and removed BUG P3 labels May 7, 2024
Copy link

git2gus bot commented May 7, 2024

This issue has been linked to a new work item: W-15704155

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
BUG P3 Up for grabs Issues that are relatively small, self-contained, and ready for implementation
Projects
None yet
Development

Successfully merging a pull request may close this issue.

7 participants