[w3c/manifest] Provide a way to find out the result of the install prompt without prompt() (#578)

There are currently two ways to find out whether the user installed your app or not:

- The `appinstalled` event. (Only gives you a positive answer, doesn't tell you if they cancelled the install prompt.)
- The `beforeinstallprompt` event: call `prompt()`. The result of `prompt` tells you whether the user accepted.

Chrome's implementation has an additional `userChoice` attribute (on the `BeforeInstallPromptEvent`) that lets you passively detect prompt acceptance/denial without calling `prompt()`.

There's one problem with using `prompt()` instead of a passive `userChoice`: it may produce slightly different prompting behaviour.

As the spec currently stands, calling `prompt()` in the `beforeinstallprompt` event handler is a no-op; it doesn't change the behaviour at all (which defaults to prompting), and therefore, `prompt()` in the event handler is equivalent to `userChoice`. However, #576 calls for a change to allow user agents to not prompt by default, but show a prompt if `prompt()` is called. This means developers using `prompt()` as a substitute for `userChoice` would suddenly get non-default behaviour. There would be no way for a site to passively detect prompt acceptance/denial without inadvertently forcing the prompt to be shown.

Some possible approaches here (presented without opinion):
- "We aren't going to action #576, and therefore `prompt()` is just as good as `userChoice`".
- "This use case is not important enough; if you care enough about collecting data, then you can decide for yourself when the prompt should be shown and not rely on the default behaviour. Or use `appinstalled` to collect data."
- "We do want a way to let developers passively collect this data. Let's add Chrome's `userChoice`."
- "We do want a way to let developers passively collect this data. But `userChoice` (being an attribute promise) is weird; let's design something else."

Thoughts?

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/w3c/manifest/issues/578