Web Codecs API

[chrisn]

The M&E IG call on 7th July 2020 (minutes) discussed the Web Codecs API.

Please submit issues to feed back directly on the API. You are also welcome to share use cases by replying to this issue.

As Chris mentioned in the call, the Blink source code currently contains the latest interface IDL files.

This issue tracks a media-related issue/proposal/work item/discussion

[chcunningham]

As mentioned when we discussed WebCodecs during the July 7th meeting, Chrome is planning to launch an origin trial for the API in M86, scheduled for stable release in early October. If you would like to participate in this origin trial, please reach out to hbengali@google.com and chcunningham@google.com so we can learn more about your use case and help validate that the API surfaces you intend to use will be ready in the origin trial time frame.

[chcunningham]

As Chris mentioned in the call, the Blink source code currently contains the latest interface IDL files.

I'll be uploading a spec PR with the latest in the coming week. For now, here's a google doc with the basic structure. A few known issues:

• Encoder not yet written. Expect same as Decode, just reversed.
• No BYOB mechanism yet. We need something like a read-only ArrayBuffer to make this secure.
• GPU:CPU copies for video frame read back. We'd like to use a GPUBuffer for frames decoded to gpu memory, but this interface is still WIP in WebGPU spec.

[tidoust]

A status update on the API. Since last comment on this issue (July 2020), WebCodecs has ben adopted as a normative deliverable by the Media Working Group and published as a Working Draft.

The API is under active development in the Media Working Group, with lots of discussions in the spec's repository:

• Draft spec: https://www.w3.org/TR/webcodecs/ (published version always reflects the contents of the latest Editor's Draft)
• Issue tracker: https://github.com/w3c/webcodecs/issues/

Some of the issues being discussed are on "minor" technical details, other touch on core API design considerations. For instance, there is an ongoing call for consensus on whether WebCodecs should be exposed in Window environments (or only in workers for the time being).