Currently the timing around applyConstraints() and getSettings() when there are no capabilities (like screen capture) looks like: - applyConstraints(newConstraints) - Select desiredDimensions from newConstraints (one or both of [width, height] may be 0) - Reconfigure capture with desiredDimensions (if this fails, reject applyConstraints() and early exit) - (settings.width, settings.height) = desiredDimensions - resolve applyConstraints() - frame comes in with actualDimensions - (settings.width, settings.height) = actualDimensions When applyConstraints resolves we need to guarantee that the settings are already correct.