Re: Several event types are too discrete to be useful for touchscreen input

On Wed, Nov 7, 2012 at 5:53 PM, James Craig <jcraig@apple.com> wrote:

>
> On Nov 5, 2012, at 6:06 PM, Rick Byers <rbyers@chromium.org> wrote:
> > On Mon, Nov 5, 2012 at 5:43 PM, James Craig <jcraig@apple.com> wrote:
> >> On Nov 5, 2012, at 12:44 PM, Rick Byers <rbyers@chromium.org> wrote:
> >>
> >>> Is there a good reason to prefer polar co-ordinates to cartesian
> co-ordinates?
> >>
> >> Not really. A lot of times the particular language used when creating
> actions gets changed by the time it makes it into the spec. Some of the
> discussion was leading towards directional events, but when we decided that
> wouldn't be enough, polar coords was suggested.
> >>
> >>> If the goal is to enable apps to use these events instead of
> mousewheel, it would help in adoption to have the co-ordinate system be the
> same (and seems like most real use cases would end up having to convert to
> delta-X and delta-Y values anyway).
> >>
> >> Sure that sounds reasonable, and as you note, probably easier on
> authors.
> >>
> > Great, thanks!
>
> Thinking about panrequest some more in the context of scrollrequest (in
> scroll views) or moverequest (e.g. an element on a layout canvas).
>
> There may be times when it is difficult for the UA to determine an x/y
> delta for more discrete events such as paging up or down. We could assume
> the offsetHeight or offsetWidth of the element, but I'm not sure that's a
> safe assumption. On scrollrequest, I'm leaning towards keeping the discrete
> enums, possibly with an optional x/y delta, or perhaps leaving the x/y
> delta variant as a separate wheel event. Cartesian deltas make sense for
> panning maps via panrequest, but probably don't cover enough for
> scrollrequest where you do have more discrete events like keypresses for
> home/end/pageup/pagedown/arrows/spacebar, in addition to the scrollwheel
> events.
>
> Thoughts?
>

If you want UAs to generate scrollrequest based on trackpad or touch screen
gestures, then it must allow the precise offsets to be specified.  A lot of
browser work goes into tuning the scroll curves for a particular trackpad
gesture (to get the scroll feeling just right), and those offsets show up
in mousewheel events today.

That said, it's not clear to me what the relationship will be between
mousewheel and scrollrequest.  If you're OK having UAs continue to use
mousewheel for most scrolls, and potentially also opt into scrollrequest
for discrete scrolling with semantics (moving one page, etc.) then it
should be fine.

James
>
>