Specification has been updated

[KristofferStrube]

As shown on the Status page, the spec has been updated. Some things have been added, and some things have changed.

Changed

• type of echoCancellation on MediaTrackCapabilities to sequence<(boolean or DOMString)> from sequence<boolean>. ⚠️
• type of echoCancellation on MediaTrackConstraintSet to ConstrainBooleanOrDOMString from ConstrainBoolean. ⚠️
• type of echoCancellation on MediaTrackSettings to (boolean or DOMString) from boolean. ⚠️
• the event type of devicechange to DeviceChangeEvent from Event.

Some of these changes (marked with ⚠️) would result in binary-breaking changes as the types in the wrapper would change. So we need a major version for these changes. Currently, the wrapper is broken in Chrome if the camera returns one of the EchoCancellationModeEnum values in its MediaTrackCapabilities, which I have observed recently. So I would like to update it soon. I might even make a major release with these changes alone, before the added features are done, to stabilize it again. Another option is to temporarily make a fix for it by adding a custom converter for a bool[]? that would ignore types that were not a bool.

Added

• boolean backgroundBlur to MediaTrackSupportedConstraints.
• sequence<boolean> backgroundBlur to MediaTrackCapabilities.
• ConstrainBoolean backgroundBlur to MediaTrackConstraintSet.
• boolean backgroundBlur to MediaTrackSettings.
• EchoCancellationModeEnum type.
• DeviceChangeEvent type.
• DeviceChangeEventInit type.
• getUserMedia on Navigator, which is our MediaDevicesService.
• ConstrainBooleanOrDOMStringParameters which has properties that should be a union of a bool and EchoCancellationModeEnum.
• ConstrainBooleanOrDOMString which should be represented by a union type of bool, EchoCancellationMode, and ConstrainBooleanOrDOMStringParameters.

All of these can be added without issue.

Could I have caught it earlier?

Yes. First, I could have added more tests, but having them alone would not have helped if I did not run them periodically. I'm considering creating a job or similar process to run my tests periodically and alert me if they suddenly fail to parse. That could be expanded further in a subsequent issue.