Open Bug 660224 Opened 11 years ago Updated 3 years ago

Intermittent css-ui-invalid/default-style/input-focus.html, css-ui-invalid/default-style/textarea-focus.html | image comparison (==), max difference: 83, number of differing pixels: 1724 or 2448

Categories

(Core :: Layout: Form Controls, defect, P5)

defect

Tracking

()

People

(Reporter: mounir, Unassigned)

References

Details

(Keywords: intermittent-failure, Whiteboard: [likely a real bug, unlikely a test flakyness][maybe an invalidation bug - see comment 268][test disabled on Windows][leave open])

Attachments

(1 file)

+++ This bug was initially created as a clone of Bug #626103 +++

These two failures are different from the select-*.html failures.
No longer blocks: 659708
No longer depends on: 626103
http://tinderbox.mozilla.org/showlog.cgi?log=Mozilla-Beta/1306503277.1306507153.19612.gz
Rev3 WINNT 6.1 mozilla-beta debug test reftest on 2011/05/27 06:34:37
s: talos-r3-w7-020
REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/test/build/reftest/tests/layout/reftests/css-ui-invalid/default-style/textarea-focus.html | image comparison (==)
Whiteboard: [orange]
Mounir, looking at the images in this reftest, it seems like one of the screenshots has been taken before the CSS UI invalid highlight has finished animating.  Do we really paint these highlights as animations?
(In reply to comment #3)
> Mounir, looking at the images in this reftest, it seems like one of the
> screenshots has been taken before the CSS UI invalid highlight has finished
> animating.  Do we really paint these highlights as animations?

These highlights are simple box-shadow. I don't think box-shadow should be animated but I don't know how they internally work. Though, you might mistaken this because the box-shadow is lighter when the element has a focus ring so what you might think is an unfinished animation is just because the box-shadow didn't change.
See: https://mxr.mozilla.org/mozilla-central/source/layout/style/forms.css#639
(In reply to comment #4)
> (In reply to comment #3)
> > Mounir, looking at the images in this reftest, it seems like one of the
> > screenshots has been taken before the CSS UI invalid highlight has finished
> > animating.  Do we really paint these highlights as animations?
> 
> These highlights are simple box-shadow. I don't think box-shadow should be
> animated but I don't know how they internally work. Though, you might
> mistaken this because the box-shadow is lighter when the element has a focus
> ring so what you might think is an unfinished animation is just because the
> box-shadow didn't change.
> See:
> https://mxr.mozilla.org/mozilla-central/source/layout/style/forms.css#639

Hmm, so the first image has a crisp red border, while the second image has a lighter red border (which is probably because of the 0.4 alpha value).  Does that mean that the textarea was not focused when the first page was loaded?
Depends on: 660226
No longer depends on: 660226
(In reply to comment #6)
> Does that mean that the textarea was not focused when the first page was loaded?

The caret is inside the textarea so it's at least "somewhat" focused.
FWIW, this bug seems to only happen on Windows so far...
Summary: Intermittent failure in textarea-focus.html input-focus.html → Intermittent css-ui-invalid/default-style/input-focus.html, css-ui-invalid/default-style/textarea-focus.html | image comparison (==)
Whiteboard: [orange]
Summary: Intermittent css-ui-invalid/default-style/input-focus.html, css-ui-invalid/default-style/textarea-focus.html | image comparison (==) → Intermittent css-ui-invalid/default-style/input-focus.html, css-ui-invalid/default-style/textarea-focus.html | image comparison (==), max difference: 83, number of differing pixels: 1724 or 2448
IMO, this is showing a bug. The random orange disappeared for a 4 months and just came back...
Would that be possible that bug 539356 fixed this issue?
(In reply to Robert O'Callahan (:roc) (Mozilla Corporation) from comment #268)
> Yes.

Any idea what could have regress the fix?
Marking this as depending on bug 808466. Hopefully, bug 808466 is the cause of this orange coming back.
Depends on: 808466
Whiteboard: [likely a real bug, unlikely a test flakyness]
(In reply to Mounir Lamouri (:mounir) from comment #266)
> IMO, this is showing a bug. The random orange disappeared for a 4 months and
> just came back...

(In reply to Mounir Lamouri (:mounir) from comment #282)
> Marking this as depending on bug 808466. Hopefully, bug 808466 is the cause
> of this orange coming back.

Bug 808466 didn't fix this. Any ideas?