Closed Bug 579414 Opened 14 years ago Closed 14 years ago

[win7] REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/mozilla-central-win7-opt-u-reftest/build/reftest/tests/layout/reftests/generated-content/dynamic-attr-01.html

Categories

(Core :: Layout, defect)

Product:
Core
Component:
Layout
Platform:
x86
Windows 7
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED

People

(Reporter: dholbert, Assigned: jfkthame)

References

Details

(Keywords: intermittent-failure)

Attachments

(1 file)

reftest failure log (with image data)
15.29 KB, text/plain
Details 
REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/mozilla-central-win7-opt-u-reftest/build/reftest/tests/layout/reftests/generated-content/dynamic-attr-01.html |
Reftest failure log attached.

This failed in 3 out of the last 5 Win7 Ro cycles:

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1279291981.1279293605.11808.gz
Rev3 WINNT 6.1 mozilla-central opt test reftest on 2010/07/16 07:53:01
s: talos-r3-w7-024

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1279292778.1279294573.15563.gz
Rev3 WINNT 6.1 mozilla-central opt test reftest on 2010/07/16 08:06:18
s: talos-r3-w7-029

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1279298344.1279299709.4021.gz
Rev3 WINNT 6.1 mozilla-central opt test reftest on 2010/07/16 09:39:04
s: talos-r3-w7-032

So far, whenever this fails, the test "layout/reftests/bugs/289480.html#top" also fails. (I'll be filing a bug on that one next)
Blocks: 438871
Whiteboard: [orange]
(In reply to comment #0)
> So far, whenever this fails, the test "layout/reftests/bugs/289480.html#top"
> also fails. (I'll be filing a bug on that one next)

Ah, that one seems to be bug 578114.

And as dbaron suggests in bug 578114 comment 1, that bug & this bug may both be due to the root problem described in bug 578116.
Summary: sporadic on win7: REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/mozilla-central-win7-opt-u-reftest/build/reftest/tests/layout/reftests/generated-content/dynamic-attr-01.html → [win7][d2d] REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/mozilla-central-win7-opt-u-reftest/build/reftest/tests/layout/reftests/generated-content/dynamic-attr-01.html
Blocks: d2d-reftest
(In reply to comment #0)
> This failed in 3 out of the last 5 Win7 Ro cycles:

Sorry, make that 4 out of the last 5.  (the latest failure is comment 2)
The testcase & reference both look like this:
>  before after
...and we seem to be failing because the very leftmost and very rightmost 1px of grayish antialiasing are clipped in the testcase.  (clipped from the upper-left serif of the initial 'b' character and the upper-right serif of the final 'r' character)
Sorry, I mistakenly assumed that the windows 7 boxes that currently show up on Tinderbox had d2d enabled.  They do not.  (So, comment 1 is incorrect, as the bugs referenced there are d2d-specific)

This appears to be failing > 50% of the time right now... :-/  If we don't fix it or mark it as random-on-win7 (can we do that?) soon, then this is on track to be one of the most frequent randomoranges we have...
No longer blocks: d2d-reftest
Summary: [win7][d2d] REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/mozilla-central-win7-opt-u-reftest/build/reftest/tests/layout/reftests/generated-content/dynamic-attr-01.html → [win7] REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/mozilla-central-win7-opt-u-reftest/build/reftest/tests/layout/reftests/generated-content/dynamic-attr-01.html
(In reply to comment #1)
> (In reply to comment #0)
> > So far, whenever this fails, the test "layout/reftests/bugs/289480.html#top"
> > also fails. (I'll be filing a bug on that one next)
> 
> Ah, that one seems to be bug 578114.

As noted in comment 10, 578114 is actually D2D-only, so it's a *different* underlying issue from the "289480.html#top" failure that we're seeing coinciding with this bug's failures. I filed Bug 579548 for these.
(In reply to comment #5)
> The testcase & reference both look like this:
> >  before after
> ...and we seem to be failing because the very leftmost and very rightmost 1px
> of grayish antialiasing are clipped in the testcase.  (clipped from the
> upper-left serif of the initial 'b' character and the upper-right serif of the
> final 'r' character)

At jfkthame's suggestion, I pushed a possible-fix to make this test use a sans-serif font (since the random failure is from clipping the tips of the serifs):
  http://hg.mozilla.org/mozilla-central/rev/0688052c607f

Hopefully that fixes this, but I'm leaving it open for now just in case.
Darn -- that last log was a failure from after the attempted fix in comment 13.  It's only got 1px of diference -- the blue pixel on the rightmost tip of the 'r'. (The 'b' matches now, but not the 'r')

Can't back out the failed fix right now, but will do so soon.
Another attempt to fix this:
http://hg.mozilla.org/mozilla-central/rev/f6ea1d942fe2

This changes the font to sans-serif (as in comment #13), and also adds a final "!" to the text, which should resolve the remaining problem described in comment #16.
Marking this as fixed; please re-open if it does happen again.
Assignee: nobody → jfkthame
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → FIXED
Whiteboard: [orange]
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: