728942 - 12 reftests pass on native and fail on xul on m-c due to cursor in the test image snapshot

Status: RESOLVED INVALID

(Firefox for Android Graveyard :: General, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

There are 12 tests which fail (unexpected-pass) on android xul (and are marked that way in the reftest.list file) but run correctly (expected fail != ) on android native:
http://mxr.mozilla.org/mozilla-central/source/layout/reftests/editor/

native:
these tests are different due to the gradient color at the top of the textarea element.  

xul:
the input element has the exact same color scheme for the test and reference image.


here is an example test image:
http://mxr.mozilla.org/mozilla-central/source/layout/reftests/editor/spellcheck-textarea-attr.html?force=1

here is the reference image for all of the tests:
http://mxr.mozilla.org/mozilla-central/source/layout/reftests/editor/spellcheck-textarea-ref.html?force=1

I don't know if we can set a preference, but hacking the reftest.list is not a good way to resolve this.

Comment 1

[Joel Maher ( :jmaher ) (UTC -8)]

Attachment: fix manifest to support xul tests for android and browserIsRemote (1.0)

this patch fixes the bogus browser.tabs.remote pref on native android and it keys off that for the manifest to ignore these tests.  

Jesse: you have asked in the past that we stop just doing blanket fails-if(Android), this is why I am asking for your input here.

Comment 2

[Jesse Ruderman]

Comment on attachment 601987 [details] [diff] [review]
fix manifest to support xul tests for android and browserIsRemote (1.0)

This patch is an improvement in that it enables tests in some instances where they were previously disabled :)

Please file a bug for the underlying problem that causes these tests to fail in remote Android tests.  (I can't tell from this bug report what the underlying problem is; in the future, when filing bugs about reftest failures, please include links to the images produced.)  Please include that bug number in the manifest.

Depending on the scope and longevity of the underlying bug, it might be best to introduce a new reftest variable for its presence rather than repeating "Android&&browserIsRemote" and the bug number in reftest manifests.

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

seems that a m-c update fixed this.