Intent to Extend Experiment: Web Install API
Lia Hiscock
Contact emails
liahi...@microsoft.com, krist...@microsoft.com
Explainer
Specification
https://github.com/w3c/manifest/pull/1175
Design docs
https://docs.google.com/document/d/12nSXJLm8mW0gWZ_yjlXfrV8r9gwJliVt4WVa-209-KA/edit?tab=t.0
Summary
Allows a website to install a web app. The API provides 3 signatures, with 0, 1, and 2 parameters, respectively. When invoked, the website installs either itself, or another site from a different origin, as a web app (depending on the provided parameters). All 3 signatures will be experimented with in parallel.
*Terminology - A site installing itself is a "current document install". A site installing another site is a "background document install".
Blink component
Web Feature ID
Search tags
TAG review
https://github.com/w3ctag/design-reviews/issues/1051 (Although this is technically the early design review, we went through several rounds of feedback off of this that influenced the shape of the API significantly)
TAG review status
Issues open
Origin Trial Name
WebAppInstallation
Goals for experimentation
We are looking to answer the following questions about the current API surface with this experiment to aid in further developments of the feature. We’ve added 2 new UKMs and 2 UMAs, in addition to preexisting web app/permissions metrics, to help in answering these questions during the experiment.
1 - Is there evidence of abuse or annoyance in the API's prompt surfaces (permission and installation)? metric(s) - App installation rate per origin; Permissions metrics, such as rate of prompt denial, or permission disablement in site settings.
2 - Is there a spike in uninstallations, that would indicate user regret, or lack of UX clarity? metric(s) - Amount of apps uninstalled within 1 hour of installation.
3 - Do developers understand the manifest id requirements of the API? metric(s) - Rate of DataErrors returned to the renderer.
4 - The API has 3 signatures, and can do 2 types of installations, as well as app launches. Are developers understanding and effectively using all signatures? metric(s) - JS use counter data; success rate of both current and background installs; CTR of app launch dialog.
Chromium Trial Name
WebAppInstallation
Origin Trial documentation link
WebFeature UseCounter name
WebDXFeature::kNavigatorInstall
Risks
Interoperability and Compatibility
Interop - Low risk. This is a new API that installs web apps, which are supported by other browsers.
The no argument version of the API (installs the current document) has enough support to be merged into the W3C web app manifest spec - https://github.com/w3c/manifest/pull/1175. It has also been reviewed favorably by TAG - https://github.com/w3ctag/design-reviews/issues/1051. While the argument versions of the API (installing a document in the background) are still under discussion in WICG, we have no reason to believe there will be interop risk with these additions.
Compatibility - Low risk. This is a new API that requires explicit developer action to use.
Gecko: No signal (https://github.com/mozilla/standards-positions/issues/1179)
WebKit: No signal (https://github.com/WebKit/standards-positions/issues/463)
Web developers: Positive (https://github.com/w3ctag/ethical-web-principles/issues/120#issuecomment-2285348765)
https://github.com/w3ctag/ethical-web-principles/issues/120#issuecomment-2285431557
Other signals: pwastore.io - https://www.reddit.com/r/PWA/comments/1o1excp/comment/niit2jh/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
Ergonomics
This could be used in conjunction with the navigator.getInstalledRelatedApps API, which tells a developer if any related web apps are installed for their site, before attempting to install with navigator.install. There is overlap between navigator.install and the BeforeInstallPrompt event. navigator.install is more ergonomic, and we think developers will prefer its declarative install. See this thread - https://github.com/MicrosoftEdge/MSEdgeExplainers/issues/1055
Activation
No activation risks. It should be relatively easy for developers to take advantage of this feature immediately, as-is. The API was designed with ergonomics in mind, and we have four places with instructions for developers (two test sites, the explainer itself, and a Developer Trial README)
Security
See security considerations for both current document and background document functionality -
current document - https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/main/WebInstall/explainer-current-doc.md#accessibility-privacy-and-security-considerations
background document - https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/main/WebInstall/explainer-background-doc.md#accessibility-privacy-and-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?
N/A
Reason this experiment is being extended
We are requesting an extension so that navigator.install can run in parallel with the HTML Install Element origin trial (approved to start in 153), which is the new, declarative entry point to the same installation capability. We have metrics in place to enable direct comparison of the entry points, and we now need the community to tell us which they prefer.
Ongoing technical constraints
None
Debuggability
Existing DevTools support for promise-based JS APIs. No new DevTools support is needed.
Will this feature be supported on all six Blink platforms (Windows, Mac, Linux, ChromeOS, Android, and Android WebView)?
No
Windows, Mac, Linux, and ChromeOS will be shipped first. Android will be supported later, due to significant technical deviation in the web app ecosystem -
https://issues.chromium.org/issues/424497410.
As of now, no plan to support Android WebView.
Is this feature fully tested by web-platform-tests?
No.
Web app installs are not supported in WPT, but we are adding tests to cover Blink-specific precondition failures (e.g. malformed input, illegal contexts, etc)
DevTrial instructions
https://microsoftedge.github.io/Demos/pwa-web-install-api
Flag name on about://flags
web-app-installation-api
Finch feature name
WebAppInstallation
Requires code in //chrome?
False
Tracking bug
https://issues.chromium.org/issues/333795265
Measurement
We have a JavaScript UseCounter and WebDX feature - https://chromestatus.com/metrics/webfeature/timeline/popularity/375 We will also implement chromium UMAs and possibly UKMs for the API service results.
Estimated milestones
|
Origin trial desktop first |
143 |
|
Origin trial desktop last |
148 |
|
Origin trial extension 1 end milestone |
153 |
|
Origin trial extension 2 end milestone |
153 |
|
DevTrial on desktop |
139 |
Link to entry on the Chrome Platform Status
https://chromestatus.com/feature/5183481574850560?gate=5595120835624960
Links to previous Intent discussions
Intent to Prototype:
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/DB9PR83MB05184F25E5FEE4595055FE34CC042%40DB9PR83MB0518.EURPRD83.prod.outlook.com
Ready for Trial: https://groups.google.com/a/chromium.org/g/blink-dev/c/ame_t7wN2cU
Intent to Experiment: https://groups.google.com/a/chromium.org/g/blink-dev/c/KCpVX8F6yh8/m/Fbs3p4PdAwAJ
This intent message was generated by Chrome Platform Status.