- From: Arthur Barstow <art.barstow@nokia.com>
- Date: Mon, 04 Apr 2011 14:09:03 -0400
- To: Anders Höckersten <andersh@opera.com>, "public-webevents@w3.org" <public-webevents@w3.org>
Ooops - I meant Anders (not Sangwhan). On Apr/4/2011 2:00 PM, ext Arthur Barstow wrote: > Hi All, > > Below, Matt proposes a solution. If anyone disagrees, please speak up; > otherwise, I think we should consider Matt's proposal as agreed (and > he can implement his proposal in the spec). > > Sangwhan - I also copied your response to this thread below. It > appears Matt's proposal is consistent with what you said. Please clarify. > > -Thanks, ArtB > > On Apr/2/2011 10:09 AM, ext Matt Brubeck wrote: >> On 03/28/2011 01:22 PM, Benjamin Poulain wrote: >>> From my experience on the issue with Qt on Windows 7, this mixing of >>> mouse event and touch event make it impossible to do anything useful >>> with the page. >>> >>> The best solution I had was to discard the mouse event generated by the >>> platform for the touch event. This way the touch and mouse on the same >>> computer work as expected. >> >> I agree. In order to support this technique, I think the Touch >> Events spec should at least specify: >> >> "If the user agent dispatches both touch events and mouse events in >> response to the same user action, then the 'touchstart' event must be >> dispatched before any mouse events for that action." >> >> This allows content to support both mouse and touch events, and >> switch between them dynamically in a clean way. (For example, this >> would allow our own test suite to detect non-touch-aware browsers.) >> It is also compatible with existing implementations. Any objections? > > On Mar/29/2011 4:39 AM, ext Anders Höckersten wrote: > > On Mon, 28 Mar 2011 18:44:07 +0200, Matt Brubeck > <mbrubeck@mozilla.com> wrote: > > > >> There is a related issue on browsers that support both touch and > mouse input - for example, Firefox 4 on Windows 7 tablets. > >> > >> On platforms like Windows, the operating system may translate touch > input into mouse movements. This means that, instead of one > "mousemove" event happening after the "touchend" event, mousemove > events and touchmove events are interleaved. (Maybe these mouse > events should be suppressed if preventDefault() is called on one or > more of the touch events?) > >> > >> [removed text Matt left in by mistake] > >> > >> I'm not yet clear enough about this problem to enter any issues in > Track, but I'd like to hear other thoughts. > > > > I can add that QT also has this behavior of sending both mouse > events and touch events, at least on some systems[1]. > > > > I'm leaning towards wanting to ignore system mouse events that are > generated for touch events, and let the browser handle emulation of > mouse events by itself. The reason is that I don't see why website > makers should have to worry about the specific touch/mouse behavior of > different OSes (beyond having to cater to both touch and mouse > separately, of course), but perhaps I'm missing something here? > > > > /Anders > > > > [1] > http://doc.trolltech.com/latest/qtouchevent.html#mouse-events-and-the-primary-touch-point > >
Received on Monday, 4 April 2011 18:09:47 UTC