971729 - [System] navigator.vibrate() is available even when the device has no vibrator

Status: NEW

(Core :: DOM: Device Interfaces, defect, P3)

Description

[Marcus Cavanaugh [:mcav]]

The IDL docs for navigator.vibrate()[1] state that "If the device does not have a vibrator, this function does nothing." Presently, this works as described.

However, because the function is always present, even if the hardware does not have a vibration motor, there is no way to detect whether or not the device supports vibration. Even Firefox Nightly (just the plain-old browser) does a no-op instead of nulling the function.

Use-cases:

- Clock wants to hide the "Vibrate" option for alarms if the hardware does not support vibration. (bug 964770)

- Apps may want to offer alternate functionality (e.g. visual feedback for input or games) if the system does not support vibration.

I asked if this was expected on b2g-dev; Jonas Sicking's response was "We should probably make the function not exist if the hardware doesn't support vibration."

Some places in Gaia check for the presence of `navigator.vibrate` first (as they should); some do not. So this would require updating any unqualified references in Gaia, unless we find some other way to indicate whether or not the device actually supports vibration.

[1]: https://mxr.mozilla.org/mozilla-central/source/dom/interfaces/base/nsIDOMNavigator.idl?rev=b405f493e834#80

Comment 1

[Boris Zbarsky [:bzbarsky]]

This is pretty easy to hook up if we have a way to test whether the hardware supports vibration.

Comment 2

[Robert Kaiser]

I thought most APIs we have are set to have a property/object that exists but evaluates to false in an |if| statement, i.e. I'd expect something like |if (navigator.vibrate) { }| to work for feature detection - but not sure if that works when it's supposed to be a function.

Comment 3

[(no longer active)]

(In reply to Robert Kaiser (:[e-mail]) from comment #2)
> I thought most APIs we have are set to have a property/object that exists
> but evaluates to false in an |if| statement, i.e. I'd expect something like
> |if (navigator.vibrate) { }| to work for feature detection - but not sure if
> that works when it's supposed to be a function.

That check will work fine if there is no name "vibrate" in navigator.

Comment 4

[(no longer active)]

(In reply to Boris Zbarsky [:bz] from comment #1)
> This is pretty easy to hook up if we have a way to test whether the hardware
> supports vibration.

Marcus, do you know if we have an existing way to determine this?

Comment 5

[Marcus Cavanaugh [:mcav]]

Not that I know of, but I'm completely unfamiliar with how the hardware vibrate. Someone lower-level in the stack might know, but based on what I see[1] it seems like that doesn't presently exist.

[1]: https://github.com/dhuseby/gecko/blob/master/hal/Hal.h