W3C

– DRAFT –
MiniApps Ecosystem Community Group, MiniApps Working Group, High-Performance Baseline for Web Apps CG Joint Meeting

13 November 2025

Attendees

Present
bkardell, Bryan Ellis, chunming, Da Bin, Ding Wei, Dingwei3, erisu, jyasskin, Niklas, niklasmerz, Shubham, tako, Thomas, Thomas_Steiner, xiaoqian
Regrets
-
Chair
Martin
Scribe
xiaoqian, martin, Euclid

Meeting minutes

<martin> item+ Towards a High Performance Web

<martin> item+ Ant Group's Vision

<martin> item+ Status of Specs

<martin> item+ Re-Chartering

<martin> item Logistics & Welcome

Logistics & Welcome

martin: welcome everyone!
… we will have a presentation from Vic online about how to towards a high-perf web
… the second topic will be from Yu Sen on their perspective on the status of Miniapps and the vision towards AI & MiniApps

topic High Performance (Tencent)

[Presentation https://web.wujicode.cn/page/next_web_en]

Vic Yao's slides

vic: Web has a long history
… Most of the users spend their time on native apps, not the web
… Users keeps away of the web
… Many native apps still use web for their content
… Many web framework appeared (e.g., React Native)
… we call for a web evolution
… Move beyond desktop, attract more devices

[Seamless browser experience slide]
… Web loading is hard, one page at a time, blank screens...
… Native app smoother
… reconstruct the web infrastructure

[Learning from MiniApps slide]
… MiniApps only need a small package
… (2MB max)
… MiniApps provide low level API for developers
… Reduce JS size for better performance
… MiniApps introduce a new way for computing
… MiniApps do not allow dynamic JS execution.
… This mean we should compile ahead of execution
… Improve security, AOT increases enhancement, you control better the origin
… Less modifications, and little flexibility

[slide: Different UI code languages]
… HTML is a document with no separation of data and structure
… MiniApp's WXML (similar to Vue,js, Angular, MiniApps...)
… I don't think a developer would like it

[Tech selection summary slide]

[How good is the MiniApps solution]

vic: Touch-screen friendly and friendly for developers
… Bottleneck is JS
… Difficutlt to load in watches and other devices

[JS original sin - slide]
… JS huge memory comsumption

vic: React Native has been trying to solve this
… Comparison between JS and static Hermes (graph)
… Improvement, better performance

vic: WASM+WASI cross platfrorm, cross platform
… UI layer: we should abandon the traditional HTML elements
… we should define new elements
… We can simplify them
… We should define UI templates, and encapsulate as WASI

[slide - Logic Layer]

vic: Developer can write logic in JS
… my first choice would be replace JS with AssemblyScript and compile to WASM
… This would open the Web to other languages
… Rust, C++, Go, Python
… There are zero-cost solutions: WebAssembly Component Model + WebASsembly Interface Types
… This could help us support this on more devices
… Better performance int the end

[Hig Perf. Web Solution summary slide]
… WASM, threading, data-driven and streamlined CSS
… precompilation, and WebAssembly Component Model + WIT

jyasskin: What parts exists and what are needed from the different W3C WGs?

vic: Early idea. This is not new, this comes from other platforms/ecosystems
… this technology could be moved to the High Performance

jyasskin: This is not entirely true. We should talk more
… there are many things that can improve it
… OpenUI...

vic: totally agree, it'd be great if we can have APIs to interoperate with the DOM and HTML from WASM, so web developers can continue working on the UIs as what we want

bkardell: Jeffrey mentioned CSP. Service workers work to help
… We should keep talking and involve other groups

Slideset: https://web.wujicode.cn/page/next_web_en

bkardell: The web by evolution, not revolution. This has worked well
… PWA, and other technologies like Electron
… There we can find synergies, to make valuable progress.

vic: Yes, we tried several times, but it didn't work
… Today is different, not only desktop. There are many technologies are there is fragmentation
… Why Snapchat has implemented the cross platform projects?
… their apps are easy to run

<jyasskin> https://www.w3.org/2001/tag/doc/evergreen-web/#subsetting is relevant.

vic: Using specific platform-specific SDKs is easier for developers

jyasskin: It would be useful to create a list of the things that do not work on the web
… I would like to work together with the TAG to see how we can solve it
… It think it could be wrong to say that Mobile web is different
… it's dangerous to create subsets
… To bring everyone into the full capabilities

ask li

limin: WASM is excellent for high-computation
… I would like to understand, how often you need these requirements for high computation
… If you deal with code that doesn't change, it looks like an implementation issue, more than standards

vic: Traditional Web , 90% of web works well.
… q?
… nobody wants to use a mobile page on mobile phone
… it is not a subset. We can provide a new kernel
… one for traditional web and another one for the smart devices

jyasskin: We should be specific. Browsers are responsible, not JS itself

vic: It cost more power if you run JS
… Facebook built their app in HTML5
… but not well enough

martin: I like the idea of creating a list of things that are not working and lowering the performance

Vic: sure!

martin: will sync-up later!

MiniApp and Mobile Web in AI Era

ACTION: Vic to draft a list of the things missing in the Web standards

topic Ant Group's Vision (Yu Sen)

[Yu Sen presentation @@@to add prez]

[State of Alipay MiniApps]

yusen: An Group is focused on inclusion
… Payment platform and Digital Open Platform
… Scan a QR code, order, pay
… Cross platform, iPhone, Android, HarmonyOS
… MiniApps can run in multiple super-apps
… Miniapps also can run on various devices, inc. auto

[summary of Mobile tech and MiniApp tech.]
… MiniApps use the web (using a DSL)
… Developers use IDEs and tools

[MiniApp Runtime and tools for developers]
… MiniApps are similar to web
… How do we optimize rendering and execution?
… multiple threads?
… code linking and completion. Open source facilitates end to end testing
… AI, can GPU, NPU support?

[next section: Future]

yusen: AI supports developers with coding
… AI also support users. Rendering environment for chatting... Generative UIs
… LLMs to operate

yusen: Challenge in Chatting: multi-modal speech to text

[Challenge in Chatting slide]
… Developers must figure out what spec to follow

yusen: From AI perspective, do we need a subset of the web standards?

[Challenge in Operating]

yusen: How to extract all the information and let AI understand it
… We must introduce a new API for agentic web
… How automatic frameworks support automation of e2e testing

[Security and Privacy slide]

yusen: We need permissions (like sensors)
… Execution of dynamic content and scripts
… These require efforts

ask bkardell

bkardell: There are superapps, and miniapps
… Are they not standardized?

yusen: There are documents, but not describe accurate behavior

bkbardell: Any corresponding...

yusen: we are working on open source

bkbardell: Is there a WPT?

[xiaoqian translating]

yusen: Capability of Alipay, we run the same behavior across platforms

martin: this has been something we want to discuss in the working group
… they are very similar in API and the behavior

<tomayac> The analysis Martin was referencing: https://web.dev/explore/mini-apps

martin: most of them are following the document of the WeChat mini program
… but there are definitely differences, especially the access to the OS capacities
… we are going to talk about in a later session
… we are also working on a project to find the common set of elements, find a way to create them
… but it's simply not there yet

Wei_Ding: You sen talked about AI and the Web
… if you have any user requirements, also it's focusing on the MiniApp
… perhaps in the future we can consider bringing these ideas to the Web AI IG

<jyasskin> +1

act Dingwei

act next

Test suite for MiniApps: w3c/miniapp-tests

martin: talking about the lack of performance, do you have any performance issues with MiniApps?

yusen: we have been spending a lot of effort on performance, especially on the JS execution
… my major concern is that JS has a very large ecosystem
… the developers use JS
… we should adopt to the ecosystem
… we should also work harder to make sure the ecosystem can work with MiniApps
… especially on Mobile

bkardell: I'd like to connect you with a few other groups that might be interested in these topics
… OpenUI is an example
… they are working with the Developers and a few companies to create a few catalogs and components that are currently lacking in the Web
… we can working together
… we can also get review from the community, f.ex, get feedback from the a11y groups

<jyasskin> +1

bkardell: they can also help to train the AI components

<martin> +1 (I'm following the CG)

yusen: these are very helpful and fruitful suggestions
… companies will soon evolve their user interface

Shubham: You mentioned Taobo
… If I use Taobo in the native app or as miniapp, is there any difference?

[xiaoqian translating]

yusen: Some funtions are limited in the miniapp

Subham: Any limitation in terms of security (credentials)

yusen: There is full taobo miniapp in alipay, only minor functions
… Security is the same
… Both are protected

MiniApps Specs: Status

Martin: most of the current documents in the MiniApps are stable
… but we haven't started the implementation yet
… which is problematic

[martin introduces the specs in the group]

<Dingwei3> q?

limin: are mini apps widgets and quick apps web apps hosted in webview as well?

martin: yes, although in a slightly different way

[slides: main challenges for final developers]

[slide#10 Taro: Cross-Platform Framework]

[slide#11 Taro Sample Code]

<jyasskin> This looks worse than div soup...

bkardell: ^^
… why are they not use semantic tags

martin: you can, I removed some of the slides

bkardell: [suggestion on reuse the standards...]

Yusen: Taro is a totally React Framework
… there are many other MiniApps Frameworks in China

bkardell: they don't use DOM, so there will be accessibility issues

martin: they do include aria, so there is possibility to do it, but you need to do it manually

[slide#13 MiniApps Markup Elements]

martin: image in MiniApps is called image, instead of img sync with the web
… an example of differences

<bkardell> note the tabs answer on that slide, please don't use it

[slides#12 MiniApps APIs Gap]

bkardell: what is the zip file management about?

martin: something similar to gzip in packaging and loading

[coffee break]

[we'll start at 1100]

Mentioned the API gap analysis: https://github.com/w3c/miniapp-components/blob/main/docs/gap-analysis-api.md

Mentioned Web Vs MiniApps element analysis: https://github.com/w3c/miniapp-components/blob/main/docs/gap-analysis-elements.md

Non exhaustive, just informative

[martin shows a demo of MiniApps using pure web techs]

Martin: Demonstrating "Pure Web" SuperApp.

Martin: I was listening to experts from Widget yesterday. How complex to run several webviews simultaneously?

Martin:

1. Can we reuse something from PWA?

2. Isolated Web app?

3. High-performance engines?

Martin: Miniapp is popular in Asia. Conversation with Ryan: Can we reuse internationalization/security web standards?

Martin: Need support from vendors. We don't have concrete super app to implement these yet.

Martin: Only one miniapp on Windows? Right, Xiaoqian?

Xiaoqian: Not sure project abandoned or not. Not many interest for miniapp in Desktop.

Martin: Questions?

bkardell: What WebViews are using these superapps?

xiaoqian: OS-specific webviews
… When we launched the group, Miniapps was really successful
… 100.000s Miniapp existing

xiaoqian: We considered that the web could be an option for an homogeneous way to develop MiniApps
… One feature of MiniApp uses is the access to the OS native APIs

xiaoqian: We want to guarantee to continue with the One Web principle

martin: Not many powerful features

bkardell: some APIs you mentioned like step counter, etc...

martin: Perhaps can be implemented if needed

bkardell: At some layer it makes sense collaborate with MiniApps. The more we can collaborate, the better choices

niklas: Cordova, Apache, WebViews CG

xiaoqian: Improvements to WebView would be great news to MiniApps.

niklas: From other coversations with WVs providers...
… Number 1 priority is performance and stability
… Good sign for MiniApps and WebViews
… Nothing related to standards but the implementations
… In our topic is hard to develop the standards. MiniApps might face the same challenge

xiaoqian: we have tried hard in our group.
… Packaging was the more controversial

<tomayac> https://chromeos.dev/en/tutorials/getting-started-with-isolated-web-apps/2

tomayac: IWA could have a good solution for these challenges
… security, packaging, etc.

martin: what we want to achieve is similar
… sandbox in the browsers/ superapps
… we need to find a way to ensure the security

simon: basically IWA could like to solve all these same problems
… packaging
… please file issues, make sure it can happen in the platform

IWA == isolated web app

martin: if we are to create something new, why not work with the other vendors and make the most of the Web

simon: from a high level, it sounds similar

xiaoqian: What is the plan of IWA in W3C?
… Will you move this to specific WG?
… Now we are in WICG. Suggestions welcome

<Simon> WICG/isolated-web-apps

<Simon> WICG/direct-sockets

jyasskin: is it possible to standardize IWA in MiniApps WG

<Simon> WICG/web-smart-card

<Simon> (above a couple of examples on launched APIs that went through standardization)

jyasskin: I'm not talking across groups. It will be good for the group, to mention explicitly that you want to solve this challenge
… you have concrete things to be standardized in the WG

xiaoqian: We want to close the gap with the Web.

We should avoid duplicates

jyasskin: Recommendation to have a detailed list of relations/collaborations in the charter

xiaoqian: Thomas did a great job analyzing MiniApps. Any comments?

tomayac: Some isolated proposals could fit in IWA works
… IWA and MiniApps could join forces. If IWA features can improve security, it could be also moved to the web

niklas: Could IWA could move to the broad web. It could be also interesting for WebViews
… Interesting as well for Cordova and similar uses
… WebViews should be part of the discussion

xiaoqian: about subseting the web
… Any idea?

nikals: Subsetting the web is risky
… sometime WebViews subset the standards for several reasons
… In our CG we identify the feature gap across the features implemented in practice in WebViews

martin: we organised a breakout session to design a lighter webview
… mini app is one of the use cases

bkardell: Be careful about the term "subsetting"
… The web is designed to be adapted to the devices (if you don't have camera, you can still use it)

bkardell: If we want to subset the web removing ARIA.. a total no.

martin: documenting a guidance for developers might be helpful
… it's different idea to creating a new browser engine with only this subset

niklas: Naming is hard. Subset is hard to understand
… profiles, perhaps... Baseline

<Dingwei3> +1 niklas

niklas: Baseline for MiniApps could fit, but also confusing

bkardell: try to match the current web proposals

martin: we will start working on the new charters, we might need to find a new scope, how we work with other groups

yusen: 95% behavior is similar across different vendors

<Zakim> jyasskin, you wanted to address templating

jyasskin: Web Components CG are defining templating systems
… I can find the link

yusen: okay

niklas: These differences in the implementations...

<bkardell> jyasskin: https://github.com/justinfagnani/dom-templating-api-proposal/blob/main/EXPLAINER.md

niklas: We have an ongoing project based on CompactData to understand the differences among the WebView implementations
… We have a similar situation. But we are understanding the reality of the implementations.
… First, getting the information, later to solve the challenge
… My approach, and offering collaboration

yusen: Good suggestion.

<bkardell> https://github.com/justinfagnani/dom-templating-api-proposal/blob/main/EXPLAINER.md#related-issues-and-proposals

yusen: We have an internal compactdata to show difference of our miniapps

yusen: metrics about SOs, usage, etc.

<niklas> This was WebView has today: https://caniwebview.com/baseline/

niklas: find here the data and the visualization

yusen: We also benefit from the database

dingwei: Any resolution?

ACTION: work on a list of performance issues for Mobile

<jyasskin> In addition to justinfagnani's templating proposal, there are some more in the Web Components group: https://github.com/WICG/webcomponents/blob/gh-pages/proposals/Declarative-Custom-Elements-Strawman.md#2-creating-a-shadow-tree-without-scripts and the very old https://github.com/WICG/webcomponents/blob/gh-pages/proposals/Template-Instantiation.md look relevant.

bkardell: You mentioned extensions for automobile

do these exist? what does it mean "extensions"

yusen: Alipay and WeChat provide a framework on Linux ecosysten AMPE
… This Framework may run on IoT or screen device, but not watches

yusen: In China, companies have requiremetns to extend their business on all devices. THis is important for us

bkardell: People are interested in WebViews, not always in MiniApps... Could it be interesting for others (like embeddings).
… Standardization could be the common ground. Not focused on MiniApps

<martin> +1

dingwei: New name?

bkardell: I wouldn't hurt
… rebranding could support.
… to attract more participation

<Simon> agree (could possibly also help to include IWAs then)

Niklas: I agree. Also the idea of a new logo
… MiniApp might prevent people to join, because they don't relate
… something like "embedding" the web. Not great but better

bkardell: Agreed

[discussion on naming...]

@@: something like hybrid embedded

martin: thank you for joining the meeting and all the good suggestions!

adjourned meeting! Thanks

rrsagent make minutes

rrsagent make minutes

Summary of action items

  1. Vic to draft a list of the things missing in the Web standards
  2. work on a list of performance issues for Mobile
Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Warning: ‘s/@@@to add slides/https://web.wujicode.cn/page/next_web_en’ interpreted as replacing ‘@@@to add slides’ by ‘https://web.wujicode.cn/page/next_web_en’

Succeeded: s/@@@to add slides/https://web.wujicode.cn/page/next_web_en

Succeeded: s/@@@/totally agree, it'd be great if we can have APIs to interoperate with the DOM and HTML from WASM, so web developers can continue working on the UIs as what we want

Succeeded: s/to idea/the idea

Succeeded: s/WPT/WPT?

Succeeded: s/xiaoxian/xiaoqian

Succeeded: s/jyasskin/bkardell/

Succeeded: s/jyassin/bkardell/

Succeeded: s/... what is the use case?/do these exist? what does it mean "extensions"

Maybe present: @@, [slide, [slides, bkbardell, dingwei, limin, martin, nikals, simon, Subham, tomayac, vic, Wei_Ding, yusen

All speakers: @@, [slide, [slides, bkardell, bkbardell, dingwei, jyasskin, limin, martin, nikals, niklas, Shubham, simon, Subham, tomayac, vic, Wei_Ding, Xiaoqian, yusen

Active on IRC: bkardell, chunming, Dingwei3, erisu, Euclid, jyasskin, martin, niklas, niklasmerz, Simon, tako, tomayac, xiaoqian