TPAC Web Components Community Group Breakout Takeaways

[Westbrook]

The 13 September 2023, 09:30–10:30 Central European Summer Time session "Web Components API/Specs Report and Alignment" brought focus to a number of feature areas of interest to the WCCG:

• Declarative Shadow DOM
  Firefox is committed to shipping the API as spec'd
• Scoped Custom Element Registries
  Has a breakout later in the day where various related issues as gathered by @rniwa will be discussed and hopefully resolved
  • Google Doc with minutes here: https://docs.google.com/document/d/1e1anpmGiYyAHZYg7oRkLVsGrqYWTjFCKSqyjp8p9n4I/edit
  • Resolutions:
    • [scoped-registries] Element upgrade ordering #923 (comment)
    • [scoped-registries] How does customElements.upgrade() work? #1001 (comment)
• Cross ShadowRoot Aria
  Has a breakout later in the day where various related issues will be discussed and hopefully resolved
  • Google Doc with minutes: https://docs.google.com/document/d/1DG1eMnVDqQTJe4DdQ5c7fCTXnbneRypYHdSh5s8LsgA/edit#heading=h.qt8qvnu3tp3h
  • Standing issue:
    • WebKit would prefer an approach that allowed for type mapping in the context that there are multiple types of relationships but not more than one of any type of relationship.
• CSS Slot Content Detection
  Interest across implementors, and needs continued research and discussion with the CSSWG as part of [css-shadow-parts] Make ::slotted() a combinator w3c/csswg-drafts#7922 and [selectors] Pseudo-class to indicate when a slot has content w3c/csswg-drafts#6867
  • /slotted/ was later taken up but the CSSWG: TPAC Web Components Community Group Breakout Takeaways #1026 (comment)

Feature areas of interest to implementors:

• Declarative Custom Elements
  The WCCG to work with the community to begin gathering more information on what DCE means to them.
  • Is templating required to move forward in this area?
  • Without templating do we need heavy marketing into what DCE is or isn't?
  • Can we gather learnings from Mozilla about their experience with XBL?
• DOM Parts
  • Google Doc here: https://docs.google.com/document/d/1ZmKfLXGvgmoF0p_gZMngCic3pLv7IJyD3TIgBZVKP0g/edit
  • High-level note: DOM Parts curly syntax looks good, people are eager to see more in-depth code examples and performance implications from Google's prototype
  • Resolutions:
    • Have a quarterly meeting dedicated to this discussion Jan 2024 DOM Parts Quarterly Meeting #1027
• Customized Built-ins and Alternatives
  • Google doc here: https://docs.google.com/document/d/10hRHzZWwBuz0aJqBZ63alfzYBxKdhN8WINHizioNfok/edit
  • High-level summary: lots of discussions around different options, and some of the underlying problems / things users would like to do, mostly centered around has attribute
  • Resolution:
    • This needs more discussions and should be part of the quarterly discussions

Decisions

A quarterly virtual Face to Face will be held. We will look to facilitate a more structured discussion, a la OpenUI, or the CSSWG with actual resolutions to increase the success of those sessions. Work to include quality minuting, and broad visibility of these sessions should be included as well.

Badly drawn minutes

[03:52] <westbrook> Joey: Scoped Custom Element Registries still have a couple of open issues
[03:53] <westbrook> Ryosuke: main hold back are minor details on what happens with same name in two registries and then move between trees.
[03:57] <bennyp> Slot content question: would this strip whitespace by default?
[03:59] <bennyp> <x-l>  </xl> <- has whitespace (therefore has slotted content?) <x-l></x-l> <- no content at all
[03:59] <bennyp> this gotcha can affect authors who use html formatting tools
[03:59] <JustinFagnani> https://github.com/w3c/csswg-drafts/issues/7922
[04:00] == christianliebel [[email protected]] has joined #webcomponents
[04:01] <JustinFagnani> wicg/webcomponents issue for detecting filled slots: https://github.com/WICG/webcomponents/issues/936
[04:02] <westbrook> Ryosuke: raise /thing/ combinator at the CSS WG as a real thing or not.
[04:03] <westbrook> ryosuke: maybe we need this as a pseudo class
[04:04] <westbrook> justin: walks through the JS path the allows for content detection, requiring double render and blocks SSR
[04:06] <westbrook> ryosuke:  https://github.com/w3c/csswg-drafts/issues/6867 seems slightly more attainable than https://github.com/w3c/csswg-drafts/issues/7922 at least in terms of work
[04:06] <westbrook> justin: separating has from desendents is likely a smaller conversation
[04:06] <westbrook> ryosuke: agrees
[04:08] <westbrook> ryosuke: is CSSWG is at TPAC then we should try and get this into the convo there
[04:11] <JaredW_> Declarative Custom Elements ?
[04:12] <westbrook> Justin: what DCE means could be unclear
[04:13] <westbrook> Justin: CG could support community out reach to clarify what that means.
[04:13] <westbrook> Justin: thinks there is interest
[04:13] <westbrook> Justin: doesn't get to the reports because of "broken things", but collaborative interest is great to hear
[04:13] <westbrook> ryosuke: gathering use cases is a good next step
[04:14] <westbrook> ryosuke: XBL was some version of this, as CSS essentially declared a shadow root onto something
[04:15] <westbrook> ryosuke: Mozilla should have some feedback on the experience of shipping (and unshipping ) XBL
[04:16] <westbrook> ryosuke: buggest win may be the platform performance wins that a browser could bring
[04:16] <westbrook> *biggest
[04:16] <westbrook> Justin: long road to DCE starts with DOM parts
[04:17] <westbrook> Ryosuke: DCE includes some level of replacing something in your tree
[04:18] <westbrook> Ryosuke: DCE could take progress even without templating
[04:18] <JustinFagnani> https://github.com/WICG/webcomponents/issues/1009
[04:18] <westbrook> Justin: "non-linear" progress in this area could confuse consumers, but could be marketed correctly
[04:19] <westbrook> Justin: looking for clarity on whether DCEv1 actually needs templating.
[04:21] <westbrook> Ryosuke: happy to meet virtual, but feels that a process (like what's in CSSWG) is what's missing.
[04:23] <westbrook> Ryosuke: discussion happens but conclusions are not made due to _something_...
[04:24] <westbrook> Joey: OpenUI does this by having "resolutions" and blog posts after weekly meetings
[04:25] <westbrook> Justin: is there a WG that we work under? or can we just make proposals to the DOM specs?
[04:27] <westbrook> Would be good to see these things as a consumer of the CSSWG.
[04:27] <JaredW_> Maybe part of this is having a clearer process of (a) is there a proposal(s) for [X] spec, and (b) what is the current status? Can the CG surface those more regularly?
[04:27] <westbrook> James: the level is less important than the tight agreement of an implementor
[04:29] <westbrook> Justin: do we need a WG or a process?
[04:30] <westbrook> To group: Any one against a quarterly face to face?
[04:31] <westbrook> Ryosuke: everyone agrees we can!
[04:32] <westbrook> Ryosuke: do we want more structure?
[04:32] <westbrook> Ryosuke: _probably_
[04:32] <westbrook> Ryosuke: include resolutions in those meetings.

[LeaVerou]

I wrote up some of my notes from the customized built-ins breakout here: w3c/tpac2023-breakouts#44 (comment)