Glance mode widget: please allow popView() at top level (for parity with non-glance widget UX)

Just some general things

The problem with popping the top level view is the VM doesn't know what to go back to.  That's why you get the error, and it's valid.  Without glances, it's a watch face, but doesn't know which one.  With glances, it's the widget loop.

Are you using both onKey and onBack?

Trying to use onBack/KEY_ESC in a widget does not prevent a widget from timing out - from about 30 seconds on a va/venu to 2 minutes on other watches.  This happens if the widget was started with glance view or not, and doesn't occur in the sim - a widget never times out in the sim.

When doing a glance view, you don't return a delegate.  So just, 

	function getGlanceView() {
        return [ new WUWGView() ];
    }	

In your main view, you can tell if the widget was started from a glance or not

Sys.getDeviceSettings.isGlanceModeEnabled;

this impacts what you can do in the delegate for the main view.  For example, if started with a glance, you can use up/down.  Without glance mode, you can't.

Depending on the device, when displaying a glance, Ui.requestUpdate() may be honored or not.  This you can see in the sim, and is based on "liveUpdates" as set for the device.  It's false on a f6 but true on an f6 pro for example.

Thanks for the response. I think I understand how the glance view itself works and that's not what I'm worried about here. I also have a plan for handling "glance view with live update" vs. "glance view without live update".

I'm not really worried about widget timeouts here (I realize there's nothing that can be done here -- that's just a limitation that my users have to live with.)

I also understand why popping the top-level view of a widget is not allowed, but I disagree with the rationale, especially in the case of glance mode widgets (see below).

jim_m_58 said:

Are you using both onKey and onBack?

Actually, yes. Embarassingly, that's the problem in my code. onKey() is returning false but onBack is returning true. When I return false from both, then the widget exits as expected. Thanks!

However, I still think there's an issue here.

If the glance view is meant to replace the initial view of a non-glance mode widget, then we should be able to pop the initial view of a glance mode widget, which should return us to glance view.

My rationale is:

- Non glance widget: 1st view = "widget landing page", 2nd view = "full widget page"

^ in this case popping the 2nd view returns us to the first view

Now if I want a similar UX for glance mode widgets:

- Glance widget: Glance View = "mini widget landing page", 1st view = "full widget page"

^ in this case, I want to be able to pop the 1st view and return to the glance view (at will)

If I can't pop the first view of the glance widget, now I have to have an extra landing page for no reason. (Which is similar to the UX that glance-enabled users are seeing right now.)

IOW they see:

- Glance widget: Glance View = "mini widget landing page", 1st view = "widget landing page", 2nd view = "full widget page"

In general, I disagree with the rationale behind not being allowed to pop or exit the initial view of a widget under any circumstances. I've read the threads and I understand the argument, but the simple fact is that the user is allowed to press Back on the initial view and *something* happens. It seems to me that whatever happens in that case should also happen when the initial view is popped.

I guess I have a pretty esoteric use case where I selectively block the Back button in a widget, yet I also allow the user to "exit" the (full) widget in a different way (when the Back button is blocked by the widget.). But I don't see anything wrong with it in principle. It works fine in non-glance mode without crashing anything.

A similar stopwatch widget (which also uses Back to take laps and reset the stopwatch) simply doesn't allow the user to exit the widget without stopping and resetting the stopwatch. (Or letting the widget time out.) To me that's unacceptable, because I consider it a feature that the stopwatch widget can be exited while the stopwatch is running.

TL;DR in a non-glance widget we can do "special" things with the Back button (and exit the widget in other ways), but in a glance widget we really can't, unless we add a redundant landing view.

jim_m_58 said:

The problem with popping the top level view is the VM doesn't know what to go back to.  That's why you get the error, and it's valid.  Without glances, it's a watch face, but doesn't know which one.  With glances, it's the widget loop.

For a glance-mode widget, popping the top level should take you back to the glance view, since that's how you got to the top level in the first place, right?

Or it should just do whatever would happen if the user pressed Back. Why can't popView() just do the exact the same thing as a manual press of Back?

There was a thread about this a while back where Brian talked about why this was the case.  The problem is "who" handles the press of the back button - the VM or the firmware

Too bad there isn't support for synthetic button events at a sufficiently low level, then. Thanks for the explanation.