I agree that we should do this. :-) I think the current discussions (I'm still catching up on email, but e.g. https://docs.google.com/a/google.com/document/d/1W9GfYWj-YWRO6Run3r1P4WTOE0LGgPMvLVqmC1vpBsM/edit?usp=sharing) are mostly centered around Chrome making the decision to turn the screen off when the button is pressed, so it'll hopefully be straightforward to also trigger ChromeVox there (but I don't know anything about ChromeVox).

Moving conversation from https://codereview.chromium.org/2507733004/:

I was suggesting that maybe just the new TabletPowerButtonController class should announce "screen off" and "screen on" when it starts or stops forces the backlight off due to a power button being pressed on a convertible Chromebook that's in tablet mode. (Actually, I think that we also stop forcing it off in response to certain types of keyboard input.)

warx@ wrote:

"Yes, that would be the easy way. I would expect this user case: device with no
power connected and in locked screen-off state, then press-and-release power
button will announce screen on. But soon the screen will be off without alert,
if no user activity is seen. User may not get noticed that screen is already
off.

"I am curious about this kind of user experience. However, the clean way may be
true as I see  crbug.com/646388  implies tablet power button behavior to announce
alert right now."

---

But I don't think we currently announce when the display is dimmed or turned off due to user activity, and I'm not sure whether that's a useful thing to do anyway. Perhaps screen-off-for-inactivity is useful to announce, since that can mean that we're about to lock the screen in ~10 seconds (if "Require password to wake from sleep" is set).

Taking a step back, how often is it the case that ChromeVox would be used on a device in tablet mode that doesn't have any external input devices?

I think that "screen off" and "screen on" announcements are useful because they:

a) indicate that certain types of input, e.g. touchscreen, are going to be ignored or not, and
b) provide feedback about what happened when the power button is pressed.

I'm not sure that announcing "screen off" for screen-off-for-inactivity is useful in the same way, since that doesn't actually block any input, unlike turning the screen off with the power button.

Dominic, I'd love to hear your thoughts about how this should work. :-) I'm happy to fill in any missing details about how displays are turned off or on on Chromebooks.

I talked to dmazzoni@ and dtseng@ that let us first bind the alert with just the new added feature for convertible/tablet device, that is sending alert when SetBacklightsForcedOff(false/true) is called.

If later turns out that more situations/requirements needed for UX, let us filing new bug for those.

Thanks, that approach makes sense to me too.

https://codereview.chromium.org/2507733004/ is landed, not sure why it is not shown here. As per comment 6, let us mark this as fixed now for test verification.

So for the purpose of testing, for convertible/tablet devices, alert happens when pressing/releasing power button causing the screen off/on switch. It also happens on laptop mode with screen off, and there is key event or mouse event.

But it depends on tablet power button behavior switch enabled. The plan is to enable it on M-57.

verified on 58.0.3029.51