Introduce enumerated attribute reflection extended attributes #11455

lukewarlow · 2025-07-15T10:09:54Z

Introduce enumerated attribute reflection IDL

[ReflectEnum=identifier] can be used alongside [Reflect] on DOMString and DOMString? attributes to identify the IDL enum which holds the known value keywords.

Additionally, 3 new extended attributes are allowed on enums:

[InvalidValueDefault] and [InvalidValueDefault=string] sets the value to be used when the content attribute has an invalid value.
[MissingValueDefault] and [MissingValueDefault=string] sets the value to be used when the content attribute is missing.
[EmptyValueDefault] and [EmptyValueDefault=string] sets the value to be used when the content attribute is the empty string.
Implementation bugs are filed:
- Chromium: …
- Gecko: …
- WebKit: …
- Servo: ...
- Ladybird: ...
- JSDOM: ...
The top of this comment includes a clear commit message to use.

(See WHATWG Working Mode: Changes for more details.)

/common-dom-interfaces.html ( diff )
/dom.html ( diff )
/embedded-content.html ( diff )
/form-control-infrastructure.html ( diff )
/form-elements.html ( diff )
/forms.html ( diff )
/iframe-embed-object.html ( diff )
/image-maps.html ( diff )
/images.html ( diff )
/infrastructure.html ( diff )
/input.html ( diff )
/interaction.html ( diff )
/media.html ( diff )
/popover.html ( diff )
/scripting.html ( diff )
/semantics.html ( diff )
/tables.html ( diff )
/text-level-semantics.html ( diff )
/urls-and-fetching.html ( diff )

lukewarlow · 2025-07-15T10:11:23Z

The second commit here is the interesting part for this specific PR, if you have feedback on the underlying stuff comment on #11450

lukewarlow · 2025-07-16T12:17:36Z

Interestingly some of these attributes have their keywords defined in external specs. one of them (refererpolicy) has an enum definition, so potentially we could update that spec to simply add the [InvalidValueDefault, MissingValueDefaut] attributes to the enum. And refer to it from within the [ReflectEnum].

lukewarlow · 2025-07-16T15:20:40Z

source

+  [SameObject, PutForwards=<span data-x="dom-DOMTokenList-value">value</span>, <span data-x="xattr-Reflect">Reflect</span>] readonly attribute <span>DOMTokenList</span> <dfn attribute for="HTMLLinkElement" data-x="dom-link-sizes">sizes</dfn>;
+  [<span>CEReactions</span>, <span data-x="xattr-Reflect">Reflect</span>="<code data-x="attr-link-imagesrcset">imagesrcset</code>"] attribute USVString <dfn attribute for="HTMLLinkElement" data-x="dom-link-imageSrcset">imageSrcset</dfn>;
+  [<span>CEReactions</span>, <span data-x="xattr-Reflect">Reflect</span>="<code data-x="attr-link-imagesizes">imagesizes</code>"] attribute DOMString <dfn attribute for="HTMLLinkElement" data-x="dom-link-imageSizes">imageSizes</dfn>;
+  [<span>CEReactions</span>, <span data-x="xattr-Reflect">Reflect</span>="<span data-x="attr-link-referrerpolicy">referrerpolicy</span>", <span data-x="xattr-ReflectEnum">ReflectEnum</span>=<span data-x="enum-referrer-policy">ReferrerPolicy</span>] attribute DOMString <dfn attribute for="HTMLLinkElement" data-x="dom-link-referrerPolicy">referrerPolicy</dfn>;


The ReferrerPolicy enum in the webappsec referrer policy spec needs [InvalidValueDefault, MissingValueDefault] added to it for this to be complete.

lukewarlow · 2025-07-16T15:21:45Z

source

-  [<span>CEReactions</span>] attribute DOMString <span data-x="dom-link-rel">rel</span>;
+  [<span>CEReactions</span>, <span data-x="xattr-Reflect">Reflect</span>] attribute USVString <dfn attribute for="HTMLLinkElement" data-x="dom-link-href">href</dfn>;
+  [<span>CEReactions</span>, <span data-x="xattr-Reflect">Reflect</span>="<span data-x="attr-link-crossorigin">crossorigin</span>", <span data-x="xattr-ReflectEnum">ReflectEnum</span>=<span>CORSSettingsKeywords</span>] attribute DOMString? <dfn attribute for="HTMLLinkElement" data-x="dom-link-crossOrigin">crossOrigin</dfn>;
+  [<span>CEReactions</span>, <span data-x="xattr-Reflect">Reflect</span>] attribute DOMString <dfn attribute for="HTMLLinkElement" data-x="dom-link-rel">rel</dfn>;
  [<span>CEReactions</span>] attribute DOMString <span data-x="dom-link-as">as</span>;


This could use ReflectEnum but the current keyword definitions are in the fetch spec. We could define the enum in HTML but then we might end up out of sync, or we need to define the enum in the fetch spec.

- [ReflectEnum=identifier] can be used alongside [Reflect] on DOMString and DOMString? attributes to identify the IDL enum which holds the known value keywords. Additionally, 3 new extended attributes are allowed on enums: - [InvalidValueDefault] and [InvalidValueDefault=string] sets the value to be used when the content attribute has an invalid value. - [MissingValueDefault] and [MissingValueDefault=string] sets the value to be used when the content attribute is missing. - [EmptyValueDefault] and [EmptyValueDefault=string] sets the value to be used when the content attribute is the empty string.

lukewarlow · 2025-07-30T18:10:33Z

source

+  attribute">extended attributes</span> must either take no arguments, or take a string. They must
+  not appear on anything other than an enum.</p>
+
+  <p>TODO: Define how [InvalidValueDefault], [MissingValueDefault], and [EmptyValueDefault] actually


cc @annevk and @domenic I'm curious how you think I should go about this.

I think the idea is solid it's mostly a question of how to thread it editorially?

I wonder if it's easier if you put them on the enum values. Does IDL then give you the ability to inspect them as go over the enum values? That will make it possible to inspect them from the relevant places in prose I'd think.

Going on the value only works when you specify a keyword, but sometimes you simply want the empty string or null value (depending on DOMString vs DOMString?).

But are they really part of the enum at that point?

They're part of the reflection behaviour of the IDL attribute, which is ultimately what we want to codify in IDL. But yes strictly they're not one of the keywords.

4. If attributeDefinition indicates it is an [enumerated attribute](https://html.spec.whatwg.org/multipage/common-microsyntaxes.html#enumerated-attribute) and the [reflected IDL attribute](https://html.spec.whatwg.org/multipage/common-dom-interfaces.html#reflected-idl-attribute) is defined to be [limited to only known values](https://html.spec.whatwg.org/multipage/common-dom-interfaces.html#limited-to-only-known-values): 4.1 If contentAttributeValue does not correspond to any state of attributeDefinition (e.g., it is null and there is no [missing value default](https://html.spec.whatwg.org/multipage/common-microsyntaxes.html#missing-value-default)), or if it is in a state of attributeDefinition with no associated keyword value, then return the empty string.

I guess you could change this slightly such that the lack of InvalidValueDefault etc results in the "" or null being returned?

annevk · 2025-07-31T14:46:33Z

source

@@ -7548,6 +7549,12 @@ a.setAttribute('href', 'https://example.com/'); // change the content attribute

  <h4>CORS settings attributes</h4>

+  <div w-nodev><pre><code class="idl">[<span data-x="xattr-InvalidValueDefault">InvalidValueDefault</span>="<span data-x="attr-crossorigin-anonymous-keyword">anonymous</span>", <span data-x="xattr-MissingValueDefault">MissingValueDefault</span>]


It would be nicer if these attributes go directly on the keywords below so we don't have to repeat the keywords. I realize it's novel IDL syntax, but I think that's syntax we want to have in engines anyway.

annevk · 2025-07-31T14:48:19Z

source

@@ -7548,6 +7549,12 @@ a.setAttribute('href', 'https://example.com/'); // change the content attribute

  <h4>CORS settings attributes</h4>

+  <div w-nodev><pre><code class="idl">[<span data-x="xattr-InvalidValueDefault">InvalidValueDefault</span>="<span data-x="attr-crossorigin-anonymous-keyword">anonymous</span>", <span data-x="xattr-MissingValueDefault">MissingValueDefault</span>]
+enum <dfn enum>CORSSettingsKeywords</dfn> {
+  "<dfn attr-value for="audio/crossorigin,video/crossorigin,img/crossorigin,link/crossorigin,script/crossorigin" data-x="attr-crossorigin-anonymous-keyword">anonymous</dfn>",


Should these be the dfns? I think this will also get quite confusing if we expose them both as attr-value and enum values. Not sure how to resolve that.

annevk · 2025-07-31T14:49:58Z

source

+  <p>The <code data-x="xattr-ReflectEnum">[ReflectEnum]</code> <span>extended attribute</span> must
+  take an identifier. It must only be used on attributes with a type of <code
+  data-x="idl-DOMString">DOMString</code>. Additionally, it must only appear alongside <code
+  data-x="xattr-Reflect">[Reflect]</code>. The provided identifier must be that of an IDL enum,


ReflectEnum should be mutually exclusive with Reflect, no?

Reflect triggers the reflection and takes the attribute name string. ReflectEnum limits to only known values and takes the enum identifier. So they are used together. (ReflectEnum is more like ReflectDefault in that sense)

Here's an idea: make Reflect without argument implied if ReflectEnum, ReflectDefault, or ReflectRange are present. Because in the vast majority of (all?) cases it's just noise.

That's a good idea.

#11518 - PR to do that here.

annevk · 2025-07-31T14:51:31Z

source

+  attribute">extended attributes</span> must either take no arguments, or take a string. They must
+  not appear on anything other than an enum.</p>
+
+  <p>TODO: Define how [InvalidValueDefault], [MissingValueDefault], and [EmptyValueDefault] actually


I wonder if it's easier if you put them on the enum values. Does IDL then give you the ability to inspect them as go over the enum values? That will make it possible to inspect them from the relevant places in prose I'd think.

lukewarlow · 2025-07-31T15:34:40Z

source

@@ -13936,6 +13976,13 @@ Transport Protocol">HTTP&lt;/abbr> today.&lt;/p></code></pre> <!-- DO NOT REWRAP

  <h5>The <code data-x="attr-dir">dir</code> attribute</h5>

+  <div w-nodev><pre><code class="idl">[<span data-x="xattr-InvalidValueDefault">InvalidValueDefault</span>, <span data-x="xattr-MissingValueDefault">MissingValueDefault</span>]


Cases like this are where the idea of putting them on the values doesn't work.

lukewarlow mentioned this pull request Jul 15, 2025

Introduce attribute reflection extended attributes #11450

Merged

6 tasks

lukewarlow force-pushed the reflect-enum-idl branch from 4387065 to 0ede3bc Compare July 16, 2025 15:17

lukewarlow commented Jul 16, 2025

View reviewed changes

lukewarlow force-pushed the reflect-enum-idl branch 2 times, most recently from a651540 to 33e6fc8 Compare July 16, 2025 18:28

annevk mentioned this pull request Jul 30, 2025

Formalize content attribute reflection #3238

Open

lukewarlow force-pushed the reflect-enum-idl branch from 33e6fc8 to 67c3c25 Compare July 30, 2025 18:06

lukewarlow commented Jul 30, 2025

View reviewed changes

rahimabdi mentioned this pull request Jul 30, 2025

Update ARIA IDL spec to use new WebIDL reflection semantics w3c/aria#2586

Open

lukewarlow mentioned this pull request Jul 31, 2025

Support ElementInternals via IDL extended attribute reflection #11516

Open

annevk reviewed Jul 31, 2025

View reviewed changes

lukewarlow commented Jul 31, 2025

View reviewed changes

lukewarlow mentioned this pull request Jul 31, 2025

Imply [Reflect] when [ReflectRange] or [ReflectDefault] are used. #11518

Open

6 tasks

annevk mentioned this pull request Aug 4, 2025

Upcoming WHATNOT meeting on 2025-07-31 #11506

Closed

		@@ -7548,6 +7549,12 @@ a.setAttribute('href', 'https://example.com/'); // change the content attribute

		<h4>CORS settings attributes</h4>

		<div w-nodev><pre><code class="idl">[<span data-x="xattr-InvalidValueDefault">InvalidValueDefault</span>="<span data-x="attr-crossorigin-anonymous-keyword">anonymous</span>", <span data-x="xattr-MissingValueDefault">MissingValueDefault</span>]

		@@ -13936,6 +13976,13 @@ Transport Protocol">HTTP</abbr> today.</p></code></pre> <!-- DO NOT REWRAP

		<h5>The <code data-x="attr-dir">dir</code> attribute</h5>

		<div w-nodev><pre><code class="idl">[<span data-x="xattr-InvalidValueDefault">InvalidValueDefault</span>, <span data-x="xattr-MissingValueDefault">MissingValueDefault</span>]

Introduce enumerated attribute reflection extended attributes #11455

Are you sure you want to change the base?

Introduce enumerated attribute reflection extended attributes #11455

Conversation

lukewarlow commented Jul 15, 2025 • edited by pr-preview bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lukewarlow commented Jul 15, 2025

Uh oh!

lukewarlow commented Jul 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

lukewarlow commented Jul 15, 2025 •

edited by pr-preview bot

Loading

lukewarlow commented Jul 16, 2025 •

edited

Loading