Re: Event handlers - Pointer Devices

 > On the main (or more useful) point perhaps, I think the distinction
 > between a device's internal features, its peripherals as attached
 > features, and more abstracted (even remote) resources is still very 
 > between Webapps and DAP.
 > Bryan Sullivan | AT&T

Perhaps the "System Information and Events API" section of the DAP 
charter should be two separate items. It seems to me that Events are the 
cause of the confusion here; and they are complex. The charter summary 
doesn't mention that most of the protocol APIs could trigger events.

For instance:

Many Cameras will be supporting simple image recognition of 2d barcodes; 
onbarcode() would, for instance, be an event. such an event would likely 
be in the DAP charter, not in the WebApps charter (am I wrong?). 
Additional events would encompass the current device APIs ( 
oncamerastart, ontaskadded, onnetworkunreachable, etc ).

The DAP really seems quite restricted to PIMs, which is surprising. 
Where do measurement accessories such as accelerometer, thermometer, 
compass, and other measurement services fit?

 > From: Doug Schepers
 > DAP is not focused on user-input devices, more on things like cameras,
 > calendars, etc. available on the local device.
 > I am concerned that it may not be fruitful, and may be
 > counterproductive, for us to start speculating on this list about IPR.
 > Let's keep the discussion on technical matters, please.
 > However, use cases and requirements would be appropriate to discuss, and
 > this should help frame a successful outcome for this spec.
 > [1]

Well, I'd like to speculate a little: touch is not patented but "gesture 
recognition" is. Working within that framework gives us a firm scope: 
"onpressurechange" is a touch event, "onPsychicRaysOfIntent" is 
something that can be left up to a mechanism detailed by the Device API 
group. Gestures may well belong to a persons personal device profile.

I've seen requests for "ondeviceshake", "onpinch", but I fear that 
implementations might run afoul of existing IP. There are also better 
names, that target their typical functionality: "onpinch" is often used 
to trigger a zoom command; "onzoom" (or a borrowing from SVG), would 
allow for the same functionality, without suggesting a gesture 
recognition system.

I realize that it'd be far easier to write some applications if these 
high level APIs were part of every implementation, but I think they are 
too risky. What is a "pinch", what threshold should a "shake" have? 
These are details to be left to individual implementations. Let's 
sidestep the issue early.


Received on Thursday, 4 March 2010 18:32:43 UTC