RE: Request for Review: single-touch tests; deadline Sept 24

Firefox on Nokia N9: 273 passes, 1 fail:

- "document.createTouch exists and creates a Touch object with requested
properties"; "assert_equals: touch.clientX is
touch.pageX-window.pageXOffset."; "expected 15 got 0".

 

Opera on Nokia N8: 233 passes, 3 fails:

- "document.createTouchList exists and correctly creates a TouchList from a
single Touch"; "assert_equals touchList.length is 1"; "expected 1 but got
0".

- "TouchList contains an identifiedTouch property"; "assert_true" "expected
true but got false"

- "document.createTouchList exists and correctly creates a TouchList from a
Touch array"; "assert_equals touchList.length is 1"; "expected 1 but got 0".

 

A couple of cosmetic errors in the test files.

- line 28, there is no variable called "name" (probably a copy-and-paste
error)

- line 72, ditto

 

Also, have we decided whether we want tests that validate "instanceof
TouchList"? Currently they are not there, and we were aware that WebKit
browsers fail such tests. See
http://lists.w3.org/Archives/Public/public-webevents/2012JulSep/0037.html.

 

Regards, Cathy.

 

 

From: ext Rick Byers [mailto:rbyers@google.com] 
Sent: Friday, October 05, 2012 3:23 PM
To: Barstow Art (Nokia-CIC/Boston)
Cc: public-webevents@w3.org
Subject: Re: Request for Review: single-touch tests; deadline Sept 24

 

On Fri, Oct 5, 2012 at 2:59 PM, Arthur Barstow <art.barstow@nokia.com>
wrote:

On 10/1/12 9:28 PM, ext Rick Byers wrote:

With the following two changes, and Art's guidance that these tests should
focus on testing the spec, not interoperability between implementations, I'm
now happy with the state of single-touch.html

On the multiple test results for the same problem issue, how about a simple
pragmatic trade-off: collapse all failures about identifiedTouch into one.
This addresses my main use case (wanting to be able to glance at the
failures and make sure they're all understood) without adding much
complexity.  I've committed a proposed change for that here:
https://github.com/AFBarstow/WebEvents/commit/eef7fbdb9e08bc5c901d0678d2f7ae
35b35f7071.  Seem OK?

 

Your two changes look OK to me so I copied the latest file on GH to Hg and
put it in a new dir called Moz-Nok-Goog since it has contributions from
Matt, Rick and Art:

<http://w3c-test.org/webevents/tests/touch-events-v1/submissions/Moz-Nok-Goo
g/single-touch.html>

 

FYI: shortened to http://goo.gl/ffIYx to make it easier to enter on mobile 

 

I would appreciate it folks would please run this version and report
results.

 

Latest Chrome desktop  (23.0.1271.19): 251 passes, 2 fails - both expected
due to known WebKit bugs":

document.createTouchList exists and correctly creates a TouchList from a
single Touch" ; Actual: "assert_equals: touchList.length is 1, expected 1
but got 0" ;
TouchList contans an identifiedTouch property" ; Actual: "assert_true:
expected true got false" ;

 

iOS 6 (Safari/8536.25): 251 pass, 2 fail - almost the same as the above

except that the createTouchList has a slightly different failure mode:
"assert_equals: touchList.item(0) is input touch expected object "[object
Touch]" but got null.  Looks like the same underlying webkit issue though.

 

I also found that my client/page check fails if I'm zoomed at all:

"touch.clientX is touch.pageX - window.pageXOffset expected 15 but got -201

I'll look into fixing that (presumably a units mismatch again).

 

Chrome for Android (18.0.1025453): 251 pass, 2 fail - same as above

 

Android browser (Android 4.1.1): 251 pass, 2 fail - same as above

 

On my Nokia N9 I got 233 Passes and 3 Fails:

* 2 are Expected: "document.createTouchList exists and correctly creates a
TouchList from a single Touch" ; Actual: "assert_equals: touchList.length is
1, expected 1 bug 0" ;

* 1 Expected: "TouchList contans an identifiedTouch property" ; Actual:
"assert_true: expected tru got false" ; [I think this is a browser bug that
was reported earlier]

Re multi-touch testing, next week I hope to take a deep dive but if someone
can beat me to it, that would be very much appreciated!

 

I will try to take a look at some point next week too.

 

-Thanks, Art

 

And for the screenX/Y failures on mobile devices: given the above and the
complexity of the various scale factors applied in different scenarios
(browser zoom, vs. device scale, etc), I don't think it's worth trying to
validate these values.  I've removed them:
https://github.com/AFBarstow/WebEvents/commit/86d50fc931a97f408234e937c4d45d
3fbb03ff6d

Rick