Embedded web clarifications

[RupertBenWiser]

Speaking as someone involved in the embedded web, I think it would be helpful to be clear where in a tech stack one becomes a user agent. For example, while it might be tempting to call a WebView a user agent, WebViews are often highly configurable libraries and it is not appropriate for them to override the decisions of the application developer. I think defining the user agent as the final compiled application that owns web browsing would be a helpful clarification; issue 7 already seems to assume this.

The scoping matters because it means that when we are discussing these duties, we are generally talking about extremely large app ecosystems (in the order of millions) that embed WebViews. The ecosystems section for instance feels very much to me like it is focused on browsers where they are a lot more generic in their offerings; social apps may be more “sticky” for users to switch away from. The honesty section could also be an example of where an app may be different; app developers may opt to provide a frictionless experience by autogranting permissions instead of prompting users on trusted websites.

This distinction may also help clear up why an app developer may be interested in using a component such as SFSafariViewController to embed web content as in that case, their application wouldn’t be the user agent.

[RupertBenWiser]

Cross posting our discussion on the WebView CG here: https://github.com/orgs/WebView-CG/discussions/7