Closed Bug 668849 Opened 13 years ago Closed 7 years ago

Sporadic TEST-UNEXPECTED-FAIL | reftests/text-overflow/selection.html | image comparison (==), max difference: 255, number of differing pixels: 310 (or 300)

Categories

(Core :: Layout: Block and Inline, defect, P5)

Product:
Core
Component:
Layout: Block and Inline
Platform:
x86
Windows XP
Type:
defect

Tracking

()

Status:
RESOLVED WONTFIX
Tracking Flags:
Tracking Status
firefox16 --- affected
firefox17 --- affected
firefox18 --- affected
firefox19 --- affected

People

(Reporter: dholbert, Unassigned)

References

(Depends on 1 open bug,
URL
)

Details

(Keywords: intermittent-failure) 

http://tinderbox.mozilla.org/showlog.cgi?log=Mozilla-Inbound/1309520044.1309523229.22895.gz
Rev3 WINNT 5.1 mozilla-inbound debug test reftest on 2011/07/01 04:34:04
s: talos-r3-xp-019

REFTEST TEST-UNEXPECTED-FAIL | file:///c:/talos-slave/test/build/reftest/tests/layout/reftests/text-overflow/selection.html | image comparison (==)

From looking at the failure in reftest-analyzer, it looks like the reference case has no red-wavy-underlines below the bottom 8 lines, whereas the testcase does (at least for parts of those lines).

This test landed yesterday, in bug 666669.
So, is there a way to wait for spell checking to be done in a reftest?
Assignee: nobody → matspal
Not off the top of my head.  We could add one...
Depends on: 669020
Pushed a wallpaper:
http://hg.mozilla.org/integration/mozilla-inbound/rev/dbad9fdb83e6

(In reply to comment #2)
> Not off the top of my head.  We could add one...

Filed bug 669020.
Whiteboard: [orange] → [orange][inbound]
Assignee: matspal → nobody
Whiteboard: [orange][inbound] → [orange]
And you get a random-if.
Assignee: nobody → philringnalda
Actually, we can afford a couple more seconds on reftest runs, let's see how waiting twice as long works before giving up and doing random-ifs.

https://hg.mozilla.org/integration/mozilla-inbound/rev/175bcd2c7912
Whiteboard: [orange] → [orange][leave open]
Fun question: is there a length of time after which debug WinXP does show spellcheck squiggles? Why do we believe there is?
philor, this is why this bug depends on bug 669020 ;-)
(In reply to Mats Palmgren [:mats] from comment #1274)
> philor, this is why this bug depends on bug 669020 ;-)

Well, my question was really "if I just keep increasing the setTimeout, will I eventually stop the failures?" I think, from asking try, that the answer is no - I got the same failure rates on WinXP with a 10 second, 60 second, and 200 second timeout, so I think that all bug 669020 will give you is intermittent "timed out waiting for reftest-wait to be removed."
OK, I guess what we really need is a deterministic method of triggering
spell checking to occur, as well as complete.

It would be unfortunate to have to disable this test, since it's the
only one that tests the rendering of the spell check underline clipped
by a text-overflow marker.  I won't object if you do though, with 1300+
stars that seems entirely motivated.
One of the many things I wish someone would make for me is a true random-if, so that random-if(20, ...) would run the test 20 times and unexpected-fail if none of them passed or unexpected-pass if none of them failed.

Marked as random-if on WinXP in https://hg.mozilla.org/integration/mozilla-inbound/rev/0bf91ba2ddba
Assignee: philringnalda → nobody
Whiteboard: [orange][leave open] → [leave open]
Summary: Sporadic TEST-UNEXPECTED-FAIL | reftests/text-overflow/selection.html | image comparison (==) → Sporadic TEST-UNEXPECTED-FAIL | reftests/text-overflow/selection.html | image comparison (==), max difference: 255, number of differing pixels: 310 (or 300)
Priority: -- → P5
WONTFIXing, since this is WinXP-specific and we're desupporting XP as noted in bug 1315083 comment 0.

(Aryx is removing the reftest.list annotation for this failure in bug 1334898.)
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → WONTFIX
Whiteboard: [leave open]
You need to log in before you can comment on or make changes to this bug.