Groups keyboard shortcuts have been updated
Dismiss
See shortcuts

Intent to ship: Customizable select

1,247 views
Skip to first unread message

Joey Arhar

unread,
Jan 24, 2025, 1:21:08 PMJan 24
to blink-dev

Contact emails

jar...@chromium.org, mas...@chromium.org, dan...@microsoft.com


Explainer

https://open-ui.org/components/select


Specification

https://github.com/whatwg/html/issues/9799 (see sub-PRs linked there)


Design docs

https://open-ui.org/components/select


Summary


Customizable <select> allows developers to take complete control of the rendering of <select>, including the ability to fully customize both the in-page “button” element as well as the popup picker, with arbitrary content within options. Developers can opt in to this new behavior with a simple CSS declaration that uses a new `base-select` value for the `appearance` property.


This new capability has been in development for a very long time, starting in late 2019 in earnest (with the original explainer from June 2020). Due to that early work, two other platform APIs were identified as prerequisites for customizable-<select>: the popover API and anchor positioning. Now, with both of those APIs landed in standards and browsers, the customizable-<select> is unblocked.


Once the customizable-<select> feature began to get discussed in standards (at OpenUI in 2020, and in WHATWG and CSSWG in mid-2023), it rapidly evolved. Not only did the names of the elements change (<selectmenu>, <selectlist>, then <select>, etc.) but the shape of the API changed considerably. It evolved from a very shadow-DOM centric API using things like the slot attribute to a more HTML-like API using new elements such as the <selectedcontent> element. The WHATWG/CSSWG/OpenUI joint task force worked through the question of how to opt-in to the new behavior, selecting among a new element, an HTML attribute, a CSS property, or something else, and arrived at a consensus of using the CSS appearance property. Many discussions were had (e.g. in OpenUI and WHATWG) about the content model and the allowed set of controls, to ensure the new control is accessible and rational, while still providing a very flexible, powerful API to developers. Overall, the standards process shaped this API into something that follows platform conventions, has natural naming and methods to achieve goals, and builds naturally upon recently added features such as popover, anchor positioning, interactivity:inert, and many others. In addition, the construction of the customizable-<select> API has been a huge catalyst to find other missing platform features and quirks, such as corner cases in the popover API.


Throughout this process, we’ve worked hard to reach out to the developer community, to ensure all common use cases are supported, there aren’t lingering compat issues, and the new customizable-<select> control is as accessible as possible. In some cases, e.g. the necessary changes to the HTML parser to allow arbitrary content, there is still some compat risk. We are working hard to increase our confidence that we can ship those changes via (separate) Finch testing and rollout. However, at this point, we are confident that we have reached a stable API shape with a low level of risk.


Blink component

Blink>Forms>Select


Search tags

select, custom select, controls, form controls, custom controls, custom form controls


TAG review

https://github.com/w3ctag/design-reviews/issues/1007


TAG review status

Issues resolved


Risks



Interoperability and Compatibility

Interop risk is low because we have been designing this feature closely with Apple and Mozilla during the standardization process over the last 15 months.


Changing the HTML parser for <select> elements is a prerequisite for this change and has some compat risks, which had a separate intent here: https://groups.google.com/a/chromium.org/g/blink-dev/c/5_9-Qkvlj2M/m/Q96A126vAAAJ


Gecko: No signal

https://github.com/mozilla/standards-positions/issues/1060


WebKit: No signal

https://github.com/WebKit/standards-positions/issues/386


Web developers: Very strongly positive

https://2024.stateofhtml.com/en-US/features/#reading_list


Other signals:


Ergonomics

None in particular.



Activation

Clear documentation will be required to help developers understand the opt in and how to style and replace various parts of the new select element. We already have some documentation here but will create another blog post for launching the feature:



Security

The picker for customizable select does not extend outside of the web contents like an appearance:auto select does, so there should not be any new capabilities exposed which would have security considerations.



WebView application risks

Does this intent deprecate or change behavior of existing APIs, such that it has potentially high risk for Android WebView-based applications?

None



Debuggability

The customizable select should already be fairly debuggable via existing DevTools features. The new select does add a few new pseudo-elements, which have been added to DevTools.



Will this feature be supported on all six Blink platforms (Windows, Mac, Linux, ChromeOS, Android, and Android WebView)?

Yes


Is this feature fully tested by web-platform-tests?

Yes

https://wpt.fyi/results/html/semantics/forms/the-select-element/customizable-select



Flag name on about://flags

None


Finch feature name

CustomizableSelect


Non-finch justification

None


Requires code in //chrome?

False


Tracking bug

https://crbug.com/40146374


Estimated milestones

134



Anticipated spec changes


Customizable select is currently in stage 2 in WHATWG. We have had spec PRs open for review since August of last year, and we have been repeatedly asking for reviews and feedback at WHATNOT meetings. However, we have not received stage 3 approval yet, which means that the other implementers have not yet signed off on the final design. These are the only remaining open issues that have been raised, and we don’t anticipate any significant changes based on the resolutions of these:


Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5737365999976448


This intent message was generated by Chrome Platform Status.


Mike Taylor

unread,
Jan 28, 2025, 10:17:15 AM (10 days ago) Jan 28
to Joey Arhar, blink-dev

Please request the various bits in your chromestatus entry, thanks.

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.
To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAK6btw%2BX5UTWoDCGNUCpEUHYh%2BhgddTgOerGHugFnBGbQnb-Rg%40mail.gmail.com.

Joey Arhar

unread,
Jan 28, 2025, 11:03:53 AM (10 days ago) Jan 28
to Mike Taylor, blink-dev
> Please request the various bits in your chromestatus entry, thanks.

Done

Mike Taylor

unread,
Jan 28, 2025, 11:37:10 AM (10 days ago) Jan 28
to Joey Arhar, blink-dev

On 1/28/25 11:03 AM, Joey Arhar wrote:

> Please request the various bits in your chromestatus entry, thanks.

Done
Thanks, Joey.


On Tue, Jan 28, 2025 at 7:17 AM Mike Taylor <mike...@chromium.org> wrote:

Please request the various bits in your chromestatus entry, thanks.

On 1/24/25 1:20 PM, Joey Arhar wrote:
There's an impressive amount of related spec work here, and not all of it landed. How should we think about this? Is everything on track to land, or are some issues/PRs non-blockers to the feature?
How should we think about the tests that we are failing?

Joey Arhar

unread,
Jan 28, 2025, 11:56:07 AM (10 days ago) Jan 28
to Mike Taylor, blink-dev
> There's an impressive amount of related spec work here, and not all of it landed. How should we think about this? Is everything on track to land, or are some issues/PRs non-blockers to the feature?

Yes they're on track to land once editorial review has completed, and no blocking concerns have been raised other than the two mentioned in the "Anticipated spec changes" section of the intent to ship. It's possible other vendors will raise new concerns but we haven't heard any.

> How should we think about the tests that we are failing?

Some of the tests are flaky, as you can see since some are passing in Edge (Windows) but not Chrome (Linux) or vice-versa. There are some open bugs about the flakiness that we are working on, but it's important to note that the feature itself works correctly.

Daniel Bratell

unread,
Jan 29, 2025, 10:57:43 AM (9 days ago) Jan 29
to Joey Arhar, Mike Taylor, blink-dev

I think this is a very exciting improvement to the web platform so thanks for all the work! That said, I would want to know more about the interoperability picture. Both Mozilla and WebKit have been scarily passive on their standard position issues, even if they have excited engineers working on the spec. When could a web developer expect this to work all across the web?

/Daniel

Alex Russell

unread,
Jan 29, 2025, 11:20:57 AM (9 days ago) Jan 29
to blink-dev, Daniel Bratell, blink-dev, Joey Arhar, Mike Taylor
I'm an enthusiastic LGTM1 on this; thank you so much for pushing this forward.

As part of my vote here, I want to explicitly set the marker that when this ships, incompatible changes will be looked at with extreme skepticism. This is our usual modus operandi when we approve an I2S, but given how far ahead we are here, I want to reiterate that this is pouring the concrete, and we won't green-light incompatible changes, no matter how convinced folks at WHATWG are late in the game.

Best,

Alex

To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscribe@chromium.org.
--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscribe@chromium.org.

Joey Arhar

unread,
Jan 29, 2025, 12:48:37 PM (9 days ago) Jan 29
to Alex Russell, blink-dev, Daniel Bratell, Mike Taylor
> Both Mozilla and WebKit have been scarily passive on their standard position issues, even if they have excited engineers working on the spec. When could a web developer expect this to work all across the web?

The other browser engine representatives have been consistently enthusiastic about solving the problem in this case, and have been quite involved in reviewing the design. This is evidenced by the fact that they've all agreed that the API is at stage 2 ("Consensus that the rough API shape defined in the draft specification is the right approach to solve the problem") and have almost agreed to stage 3. Given that, and that there is very strong developer demand for this feature, we see no reason why they would not implement it soon.

To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.
--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.

Daniel Bratell

unread,
Jan 29, 2025, 1:26:24 PM (9 days ago) Jan 29
to Joey Arhar, Alex Russell, blink-dev, Mike Taylor

Yes, I understand that it's not really fair to ask you about what other browser vendors will do, because they are they. I had kind of hoped for something like "they have an implementation behind a flag and should be right behind us", but I guess that is not the case.

Do you know (yes, again asking unfair questions) if they have done any work on the prerequisite features?

At least, if they have participated in the creation process it should not be something they can't implement, because as Alex says, once this ships, changes would be painful and maybe impossible. A nice use of this feature will still render reasonably well in older browsers, right?

/Daniel

(I tried to figure out when the <select> element was added to the web. 1995 or earlier it seems, so that is basically how long people have wanted more power over it.)

Joey Arhar

unread,
Jan 29, 2025, 4:38:44 PM (9 days ago) Jan 29
to Daniel Bratell, Alex Russell, blink-dev, Mike Taylor
> Do you know (yes, again asking unfair questions) if they have done any work on the prerequisite features?

Yes, firefox and safari have both shipped popover. Anchor positioning is proposed for interop 2025.

> A nice use of this feature will still render reasonably well in older browsers, right?

Yes. Older browsers will still render whatever text the author puts in their <option> elements in the browser native UI.

Daniel Bratell

unread,
Jan 30, 2025, 10:09:07 AM (8 days ago) Jan 30
to Joey Arhar, Alex Russell, blink-dev, Mike Taylor

LGTM2

Good luck with the launch! I hope the others follow quickly. (No official signals from them but they have had plenty of time and opportunities to object and with your answers I'm confident that this can become interoperable)

/Daniel

Vladimir Levin

unread,
Jan 30, 2025, 10:50:36 AM (8 days ago) Jan 30
to Daniel Bratell, Joey Arhar, Alex Russell, blink-dev, Mike Taylor
Reply all
Reply to author
Forward
0 new messages