RE: Need test results against Jan 2013 Touch Events LC

Hi Art,

Thanks for this reminder. I intended to follow up and test on additional
devices but have since forgotten...

TL;DR version: Chrome on Android passes all tests. Mobile Safari yields some
inconsistent results. Firefox and Opera on Android both fail because of the
API change. Additionally, Firefox on Android still fails on the clientXY
check.

With Mobile Safari on iOS 6.1.2 (i.e. latest version):
* create-touch-touchlist - either 70/70 passes or 69 passes + 1 fail, where
the fail is
- document.createTouch exists and create a Touch object with requested
propertie | assert_equals: touch.clientY is touch.pageY-window.pageYOffset.
| expected 20 but got -74
---> the fail occurs if there was any zooming or scrolling done on the
previous page before navigating to the test page (e.g. if I was at [TS] and
zoomed in to see the text clearly before tapping on the link to
create-touch-touchlist.html, or if I was already in
create-touch-touchlist.html, scrolled down to see the results, and refreshed
the page). If there was no scrolling or zooming done on the previous page,
all tests pass.
---> this is something that was observed previously, but we haven't been
able to determine whether it was an implementation error.

* multi-touch-interactions - 78/78 passes, or 69/69 passes, or 68 passes + 1
fail, where the fail is
- touchend #2: change in touches.length is valid | assert_equals: change in
touches.length equals changedTouches.length | expected 1 but got 2
--->this is something that was observed previously ([1], look for "A notable
multi-touch specific observation"), and looks like an implementation error
to me.
- also depending on timing in lifting all three fingers simultaneously, I
got 78/78 passes or 69/69 passes (3 touchend events vs 3 touchend events).  

* multi-touch-interfaces - 396/396 passes

* single-touch - 227/227 passes

With Chrome on Android (v18.0.1025469):
* create-touch-touchlist - 70/70 passes

* multi-touch-interactions - 78/78 passes

* multi-touch-interfaces - 413/413 passes

* single-touch - 227/227 passes

===Yay! We have a winner!===


With Firefox on Android (v19.0):
* create-touch-touchlist - 51 passes + 2 fails
- document.createTouch exists and creates a Touch object with requested
properties | assert_equals: touch.clientY is touch.pageY-window.pageYOffset.
| expected 15 but got 0
---> note that unlike Mobile Safari, this fail always occurs regardless of
scrolling/zooming status of previous page.
---> this is something that was observed previously, but we haven't been
able to determine whether it was an implementation error.
- document.createTouchList exists and correctly creates a TouchList from two
Touch objects | assert_equals: touchList.length is 2 | expected 2 but got 1
---> This arises from the API change and is something we expect to be
updated in the next release.

* multi-touch-interactions - 78/78 passes

* multi-touch-interfaces - 413/413 passes

* single-touch - 227/227 passes

With Opera on Android (v12.10)
* create-touch-touchlist - 52 passes + 1 fail, where the fail is
- document.createTouchList exists and correctly creates a TouchList from two
Touch objects | assert_equals: touchList.length is 2 | expected 2 but got 1
---> This arises from the API change and is something we expect to be
updated in the next release.

* multi-touch-interactions - 61/61 passes

* multi-touch-interfaces - 362/362 passes

* single-touch - 227/227 passes


- Cathy.


[1]
http://lists.w3.org/Archives/Public/public-webevents/2012OctDec/0027.html
[TS] <http://w3c-test.org/webevents/tests/touch-events-v1/approved/>


> -----Original Message-----
> From: Barstow Art (Nokia-CIC/Boston)
> Sent: Friday, February 22, 2013 8:20 AM
> To: public-webevents@w3.org
> Subject: Need test results against Jan 2013 Touch Events LC
> 
> Hi All,
> 
> As you can see from Cathy's e-mail below, she updated the tests to reflect
> the January Touch Events LC [LC]. Thanks Cathy!
> 
> I ran the four approved test files [TS] on a Nokia N9 using the native
WebKit-
> based browser and Opera Mobile (12.00.8) and I got identical results:
> 
> * create-touch-touchlist - 17 Pass; 2 Fails -
> document.createTouch(touch1) returns length of 0 instead of 1 and
> document.createTouch(touch1,touch2) returns length of 0 instead of 2
> 
> * multi-touch-interactions - 78/78 Pass
> 
> * multi-touch-interfaces - 396/396 Pass
> 
> * single-touch - 227/227 Pass
> 
> Based on Sangwhan's feedback regarding the API change to
> document.createTouch(), I presume the failure above on Opera is to be
> expected. However, given this API change was supposed to align with
> WebKit implementations, I would have expected the Nokia browser to pass
> this test (perhaps that implementation predates WebKit's latest API?).
> Rick - do you have any comments about this?
> 
> Anyhow, we need data from other implementations, preferably
> implementations that code to [LC] (and not the previous CR).
> 
> Olli, Matt, Boris - can one of you please run these four files ([TS]) on
your
> updated version of FF?
> 
> Sangwhan - can you please please provide data for Opera?
> 
> I would also be interested in data from latest iOS and Chrome.
> 
> -Thanks, Art
> 
> [TS] <http://w3c-test.org/webevents/tests/touch-events-v1/approved/>
> [LC] <http://www.w3.org/TR/2013/WD-touch-events-20130124/>
> 
> 
> On 1/22/13 2:44 PM, ext Cathy.Chan@nokia.com wrote:
> > Hi all,
> >
> > The TEv1 test cases have been updated to reflect the latest spec
changes.
> >
> > Removed TouchList.identifiedTouch(), plus fixing some typos.
> > https://dvcs.w3.org/hg/webevents/rev/19c6f3d0eb8a
> >
> > Use variadic form of Document.createTouchList().
> > https://dvcs.w3.org/hg/webevents/rev/67530967ba81
> >
> > Updated single-touch test file to remove test cases that are now
> > covered in create-touch-touchlist test file.
> > https://dvcs.w3.org/hg/webevents/rev/7cd6d8c1e3bc
> >
> > With these changes, Safari on iOS passes all tests except the
> > clientX/Y test. Firefox on N9 also fails the clientX/Y test (in
> > addition to the modified createTouchList tests) but in a different
> > manner. Since I currently have limited access to other devices, I
> > would appreciate it if other folks can test on other device/UA
combinations
> and report back. Thanks.
> >
> > Regards, Cathy.
> >
> 

Received on Friday, 22 February 2013 21:09:53 UTC