Closed Bug 1411490 Opened 8 years ago Closed 7 years ago

[Win7] Intermittent editor/reftests/xul/number-3.xul == editor/reftests/xul/number-ref.xul | image comparison, max difference: 255, number of differing pixels: 56

Categories

(Core :: Layout, defect, P5)

defect

Tracking

()

RESOLVED FIXED

People

(Reporter: intermittent-bug-filer, Unassigned)

References

Details

(Keywords: intermittent-failure, Whiteboard: [stockwell fixed:other])

Attachments

(2 files, 1 obsolete file)

There are 32 failures over the last 7 days. Most of them occur on Windows 7 (pgo, debug, opt) and windows7-32-stylo-disabled (opt) Here is a recent log example: https://treeherder.mozilla.org/logviewer.html#?repo=mozilla-central&job_id=144044657&lineNumber=2742 :jet can you please take a look? Thank you!
Flags: needinfo?(bugs)
Whiteboard: [stockwell needswork]
(In reply to Stefan Hindli [:stefan_hindli] from comment #3) > There are 32 failures over the last 7 days. > Most of them occur on Windows 7 (pgo, debug, opt) and > windows7-32-stylo-disabled (opt) > > Here is a recent log example: > https://treeherder.mozilla.org/logviewer.html#?repo=mozilla- > central&job_id=144044657&lineNumber=2742 > > :jet can you please take a look? > > Thank you! This test should be a 100% fail on Windows: https://searchfox.org/mozilla-central/source/editor/reftests/xul/reftest.list#21 Masayuki: should these turn into fails-if(winWidget)? fails-if(windowsDefaultTheme&&/^Windows\x20NT\x20(5\.[12]|6\.[012]|10\.0)/.test(http.oscpu))
Flags: needinfo?(bugs) → needinfo?(masayuki)
Attached image Diff between result and reference (obsolete) —
Sorry for the delay to reply. The "0" is positioned differently. Could be sub-pixel positioning issue on Windows?
Flags: needinfo?(masayuki) → needinfo?(jfkthame)
Oops, the previous image was tiff.
Attachment #8929943 - Attachment is obsolete: true
The tests here are annotated with fails-if(windowsDefaultTheme...), and have been ever since they first landed in bug but it appears from checking the reftest log that the windowsDefaultTheme flag is not always being set as expected. When I look at the log from a passing run, I see that it has \"windowsDefaultTheme\":true, whereas the failing log has \"windowsDefaultTheme\":false. Do we really expect some of our reftest jobs to occasionally run under different theme conditions than others? Or is the problem that the windowsDefaultTheme flag is a bit unreliable? I don't really know anything about how this works in the reftest harness, but bug 1050229 suggests Markus may be somewhat familiar with it.... passing needinfo? along to him.
Flags: needinfo?(jfkthame) → needinfo?(mstange)
In the last 7 days there are 31 failures. They occur on Windows 7. A recent log example: https://treeherder.mozilla.org/logviewer.html#?repo=mozilla-inbound&job_id=147429434&lineNumber=2727 :jet, can you please take a look?
Flags: needinfo?(bugs)
Whiteboard: [stockwell unknown] → [stockwell needswork]
(In reply to Narcis Beleuzu [:NarcisB] from comment #12) > In the last 7 days there are 31 failures. They occur on Windows 7. > A recent log example: > https://treeherder.mozilla.org/logviewer.html#?repo=mozilla- > inbound&job_id=147429434&lineNumber=2727 > > :jet, can you please take a look? Let's disable these for Win32 and move on.
Flags: needinfo?(mstange)
Flags: needinfo?(bugs)
Attached patch disable the tests on windows — — Splinter Review
Attachment #8932875 - Flags: review?(gbrown)
Comment on attachment 8932875 [details] [diff] [review] disable the tests on windows Review of attachment 8932875 [details] [diff] [review]: ----------------------------------------------------------------- I don't think this will be effective: I checked several recent failure logs, and all had windowsDefaultTheme=false. I think we need to follow-up on comment 10 to better understand windowsDefaultTheme, or disable on all win32.
Attachment #8932875 - Flags: review?(gbrown) → review-
See Also: → 1417345
ok, this is a great observations Geoff and I see that all failures are windowsDefaultTheme:False and passing instances are True. We set this value here: https://searchfox.org/mozilla-central/source/layout/tools/reftest/manifest.jsm#576 sandbox.windowsDefaultTheme = g.containingWindow.matchMedia("(-moz-windows-default-theme)").matches; while we don't run stylo-disabled, historically there has been only one instance of this failure on stylo-disabled which did have the theme- either way there is either a bug in: 1) the environment we are running in 2) how we query for moz-windows-default-theme I see bug 1343049 and when these failures were greatly reduced a fix was landed (Nov 23rd). This indicates that it is related to #1 and that #1 isn't fixed yet.
Depends on: 1343049
There have been 37 failures in the last week This is occurring only on the Windows 7 platform. As for the build type, I see 20 occurrences on opt, 16 on pgo and 1 on debug. Here is a link from a recent log file: https://treeherder.mozilla.org/logviewer.html#?repo=mozilla-inbound&job_id=150953315&lineNumber=2761 And a link to the reftest-analyzer: https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://queue.taskcluster.net/v1/task/ST3H4OToSZu4EsO53LtxUA/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1 Please also take a look at Comment 16
Flags: needinfo?(bugs)
Whiteboard: [stockwell unknown] → [stockwell needswork]
(In reply to Tiberius Oros[:tiberius_oros] from comment #18) > Please also take a look at Comment 16 Comment 16 says that this bug requires that bug 1343049 gets fixed.
Flags: needinfo?(bugs)
Hi :Jet : Any update on this bug?
Flags: needinfo?(bugs)
Summary: There have been 37 failures in the last week. All of the failures are on Windows 7 and on pgo and opt build types. :jet, please take a look at this after Bug 1343049 is fixed. Thank you.
Whiteboard: [stockwell unknown] → [stockwell needswork]
(In reply to Tiberius Oros[:tiberius_oros] from comment #27) > :jet, please take a look at this after Bug 1343049 is fixed. Thank you. I'm fairly certain that you will come find me if this bug still happens after bug 1343049 is fixed :)
Flags: needinfo?(bugs)
This was a Win7 incorrect-theme issue; but the failures noted in comment 35 are all from Linux. So they are actually something quite different, and should have been filed as a separate bug.
Blocks: 1437463
Summary: Intermittent editor/reftests/xul/number-3.xul == editor/reftests/xul/number-ref.xul | image comparison, max difference: 255, number of differing pixels: 56 → [Win7] Intermittent editor/reftests/xul/number-3.xul == editor/reftests/xul/number-ref.xul | image comparison, max difference: 255, number of differing pixels: 56
this was either fixed by run-by-manifests or changes made to the image.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
Whiteboard: [stockwell unknown] → [stockwell fixed:other]
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: