Disposition of comments for the Device and Sensors Working Group

It seems the feedback I made in
http://lists.w3.org/Archives/Public/public-device-apis/2012Dec/0016.html
applies here too. (Also, IDL has no "string" type, but I agree with
Tab that it should be an enum.)

On Tue, Dec 18, 2012 at 1:27 PM, Anne van Kesteren <annevk@annevk.nl> wrote:
> [...]

I just noticed another thing. The default values for the event members
are not defined. You need to do that, otherwise (new
DeviceLightEvent("trololol").value) is not defined.

See http://dom.spec.whatwg.org/#defining-event-interfaces for advice
on defining new event interfaces.

Heya! I saw the LC announcement in public-webapps, and have some feedback.

* The introduction isn't very well written - it jumps straight into
technical details (and too much of them), without adequate explanation
of the purpose of the spec and the defined interfaces. I suggest
replacing it with something like:

This specification defines events that provide information about the
ambient light level, as measured by a device's light sensor. A
LightLevelEvent describes the light level as one of three simple
categories - "dim", "normal", and "bright" - while a DeviceLightEvent
provides a more granular answer by describing the light level in terms
of lux units.

* Chapter 5 should start with an intro that describes the event. It's
fine for this to be somewhat repetitive with the spec's own intro -
the point is to make it sensical for someone to jump straight into
this section. I suggest something like:

The DeviceLightEvent interface provides information about the ambient
light levels, as detected by the device's light detector, in terms of
lux units.

* Chapter 5 should have a note (for authors) that the precise lux
value reported by different devices in the same light may be
different, due to differences in detection method, sensor
construction, etc.

* 5.1 and 5.2.3 seem to be duplicate information. I suspect this is a
result of ReSpec, but this should be avoided if possible. Same for
6.1 and 6.2.3.

* Similar, Chapter 6 should have an intro. I suggest something like:

The LightLevelEvent interface provides information about the ambient
light levels, as detected by the device's light detector, in terms of
three general range: "dim", "normal", or "bright".

* The LightLevelEvent interface should use a WebIDL enum, rather than
describing the values solely in prose. Here's the IDL necessary:

enum LightLevel { "dim", "normal", "bright" };
[Constructor (DOMString type, optional LightLevelEventInit eventInitDict)]
interface LightLevelEvent : Event {
readonly attribute LightLevel? value;
};
dictionary LightLevelEventInit : EventInit {
LightLevel? value;
};


* In 6.2.1 (the description of the allowed values for the
LightLevelEvent.value attribute), "may" is used. Either "must" should
be used, or it should be made non-normative and switched to "can" or
something. I think the latter is appropriate, once the enum (above)
is adopted, as the WebIDL constitutes sufficient normative
requirements already.

* In 6.2.2 (the definition of how to fire the event), if the light
level can't be determined, the .value attribute of the event should be
null, not the empty string. I've already handled the nullability in
the WebIDL above.

* In 6.2.2, the prose is a little unclear about when to fire
lightlevel events. I recommend replacing the sentence starting with
"When the current light changes..." and the following note with
something like:

This specification does not define the exact lux ranges that the
LightLevel values map to, as devices with different sensitivities may
want to map them slightly differently. However, it recommends that
"dim" correspond to ambient light below 50lux, "normal" correspond to
light between 50lux and 10000lux, and "bright" correspond to light
above 10000lux. User agents must fire a light level event when the
current light level changes. User agents may fire a light level event
at any other time, for example if they have reason to believe that the
page does not have sufficiently fresh data.

~TJ

... from a reviewer's perspective, I would greatly prefer it if we
just put all sensor APIs in a single document. These are quite trivial
APIs and with the current organization most of the content is
boilerplate.)