Intent to Extend Origin Trial: Subresource loading with Web Bundles

50 views

Skip to first unread message

Daisuke Enomoto

unread,

Aug 16, 2021, 12:11:18 PM8/16/21

to blink-dev, Hayato Ito, Tsuyoshi Horo, Kunihiko Sakamoto, Kinuko Yasuda, Kenji Baheux, Jeffrey Yasskin

There is not yet a specification integrating bundles into HTML.

Summary

Provides a new approach to load a large number of resources efficiently using a format that allows multiple resources to be bundled, e.g. Web Bundles.

Blink component

Blink>Loader>WebPackaging

TAG review

https://github.com/w3ctag/design-reviews/issues/616

TAG review status

Pending

Risks

Interoperability and Compatibility

No interoperability and compatibility risk has been identified for the prototype phase. This is purely a feature addition for the performance optimization for now. It is expected that a browser which doesn’t support this feature should load subresources from the network, as usual.

Gecko: No signal
WebKit: No signal
Web developers: No signals

Ergonomics

Activation

Developers needs to package their subresoruces beforehand in order to take advantage of this feature. We'll work with JS bundler ecosystems to provide a plugin to package subresources as Web Bundles.

Security

No security risk has been identified for the prototype phase.

Goals for experimentation

<Key change included in the Intent to Extend> This Intent to Extend includes a change that fixes a memory leak encountered in the implementaition during the original OT. The extention allows us to get sufficient data before shipping the feature. In addition, the original main goals stay unchanged: 1. Measure how this feature makes a subresource loading faster in real sites. 2. Measure how this feature improves Ad Serving. See WebBundles for Ad Serving (https://github.com/WICG/webpackage/issues/624) for details. 3. Collect feedback on the API surface so that we can discuss how this and alternative approaches like resource-bundles (https://github.com/WICG/resource-bundles) could potentially converge in the future.

Reason this experiment is being extended

We encountered a memory leak in the implementation during the OT, and the fix was merged to M92. It would be nice that we have a chance of a few more iterations so that we can fix our implementation in case we find something is wrong in M93. Therefore, we plan to extend the OT to M96 so that we get sufficient data before we ship the feature. Original I2E: https://groups.google.com/a/chromium.org/g/blink-dev/c/9CwkzaF_eQ4/m/kuR07FTTCAAJ

Ongoing technical constraints

None

Debuggability

We probably need DevTools debugging support to provide a good debugging experience in loading subresources from Web Bundles. Prototyping this feature will give us a better change of designing the right support.

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

Yes

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

Flag name

Requires code in //chrome?

False

Tracking bug

https://bugs.chromium.org/p/chromium/issues/detail?id=1082020

Launch bug

https://bugs.chromium.org/p/chromium/issues/detail?id=1133108

Estimated milestones

Updated Plan: M90 - M96

Original Plan: M90 - M92

Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5710618575241216

Links to previous Intent discussions

Intent to prototype: https://groups.google.com/a/chromium.org/g/blink-dev/c/wYn13HabRN0/m/L4y4iY1-AgAJ
Intent to Experiment: https://groups.google.com/a/chromium.org/g/blink-dev/c/9CwkzaF_eQ4/m/kuR07FTTCAAJ

----------

Thanks

Daisuke

Chris Harrelson