Closed Bug 845095 Opened 7 years ago Closed Last year

Intermittent Mac states/test_tree.xul | Test timed out. followed by states/test_visibility.html failures

Categories

(Core :: Disability Access APIs, defect, P5)

x86_64
macOS
defect

Tracking

()

RESOLVED INCOMPLETE

People

(Reporter: philor, Unassigned)

Details

(Keywords: helpwanted, intermittent-failure, Whiteboard: [test disabled on OS X 10.6][leave open][see comment #40, comment #90])

Attachments

(3 files)

Now that we're running a11y on Mac, it turns out that 10.6 debug really doesn't enjoy running it very much. As quickly as it failed twice, and as rarely as inbound has managed to run it so far, I think it's going to be an explosion of orange and we'll need to disable the test pretty quickly.

https://tbpl.mozilla.org/php/getParsedLog.php?id=20079029&tree=Mozilla-Inbound
Rev4 MacOSX Snow Leopard 10.6 mozilla-inbound debug test mochitest-other on 2013-02-25 13:26:25 PST for push aa7ee3a0eede
slave: talos-r4-snow-059

https://tbpl.mozilla.org/php/getParsedLog.php?id=20079947&tree=Mozilla-Inbound
Rev4 MacOSX Snow Leopard 10.6 mozilla-inbound debug test mochitest-other on 2013-02-25 13:48:32 PST for push 71a93fff7311
slave: talos-r4-snow-077

14:01:38     INFO -  7638 INFO TEST-PASS | chrome://mochitests/content/a11y/accessible/states/test_tree.xul | Checked element cannot be state mixed!
14:01:38     INFO -  7639 ERROR TEST-UNEXPECTED-FAIL | chrome://mochitests/content/a11y/accessible/states/test_tree.xul | Test timed out.
...
14:01:42     INFO -  7643 ERROR TEST-UNEXPECTED-FAIL | chrome://mochitests/content/a11y/accessible/states/test_visibility.html | [SimpleTest.finish()] No checks actually run. (You need to call ok(), is(), or similar functions at least once.  Make sure you use SimpleTest.waitForExplicitFinish() if you need it.)
14:01:42     INFO -  7668 ERROR TEST-UNEXPECTED-FAIL | chrome://mochitests/content/a11y/accessible/states/test_visibility.html | [SimpleTest.finish()] this test already called finish!
14:01:42     INFO -  7669 ERROR TEST-UNEXPECTED-FAIL | chrome://mochitests/content/a11y/accessible/states/test_visibility.html | called finish() multiple times
And since every single one of the failures has been on one of those two builds which have failed every time so far, I'm starting to wonder if it was just a need to clobber (or, looked at from the other side, busted build dependencies).
Comment on attachment 718534 [details] [diff] [review]
Disable mochitest states/test_tree.xul on Mac.

Disabling in Mac for now
Attachment #718534 - Flags: review?(dbolter)
Comment on attachment 718534 [details] [diff] [review]
Disable mochitest states/test_tree.xul on Mac.

Review of attachment 718534 [details] [diff] [review]:
-----------------------------------------------------------------

::: accessible/tests/mochitest/states/test_tree.xul
@@ +95,5 @@
> +    if (!MAC) {
> +      SimpleTest.waitForExplicitFinish();
> +      addA11yLoadEvent(doTest);
> +    } else {
> +      todo(false, "Re-enable on Mac after fixing bug 845095 - intermittent orange");

I forget... do you need to call SimpleTest.finish() here?
Comment on attachment 718534 [details] [diff] [review]
Disable mochitest states/test_tree.xul on Mac.

Review of attachment 718534 [details] [diff] [review]:
-----------------------------------------------------------------

r=me as I trust you to sort that out.
Attachment #718534 - Flags: review?(dbolter) → review+
So, Alex, you don't want to just disable the test?
Attachment #718534 - Flags: checkin+
(In reply to Hubert Figuiere [:hub] from comment #32)
> So, Alex, you don't want to just disable the test?

No, I enabled looging to get a chance to understand what's going on here

(In reply to Hubert Figuiere [:hub] from comment #33)
> https://hg.mozilla.org/integration/mozilla-inbound/rev/8cb9d6981978

could you please back out this?
Attachment #718534 - Flags: checkin+
it seems the test is just unpatient:

09:58:45     INFO -  7772 INFO TEST-START | chrome://mochitests/content/a11y/accessible/states/test_tree.xul

10:04:43     INFO -  *****
10:04:43     INFO -  EQ matched: reorder
10:04:43     INFO -  *****
10:04:43     INFO -  7813 ERROR TEST-UNEXPECTED-FAIL | chrome://mochitests/content/a11y/accessible/states/test_tree.xul | Test timed out.

so we handled last expected event and I guess test finishing was scheduled but the test suite decided we run out of time.

Also the log contains things like:

10:04:41     INFO -  JavaScript error: , line 0: too much recursion
10:04:41     INFO -  JavaScript error: chrome://mochikit/content/tests/SimpleTest/SimpleTest.js, line 1101: too much recursion
10:04:41     INFO -  JavaScript error: chrome://global/content/bindings/tree.xml, line 951: too much recursion

perhaps that drives test suite crazy

Ideas?
Attached patch small cleanupSplinter Review
Attachment #721577 - Flags: review?(marco.zehe)
(In reply to alexander :surkov from comment #40)
> Also the log contains things like:
> 
> 10:04:41     INFO -  JavaScript error: , line 0: too much recursion
> 10:04:41     INFO -  JavaScript error:
> chrome://mochikit/content/tests/SimpleTest/SimpleTest.js, line 1101: too
> much recursion
> 10:04:41     INFO -  JavaScript error:
> chrome://global/content/bindings/tree.xml, line 951: too much recursion
> 
> perhaps that drives test suite crazy
> 
> Ideas?

Sorry I'm not sure about that, 301 the js team?
Attachment #718534 - Flags: checkin-
Comment on attachment 721577 [details] [diff] [review]
small cleanup

Review of attachment 721577 [details] [diff] [review]:
-----------------------------------------------------------------

Does that fix the "recursion" problem? The missing "return" is also mindboggling.

I'd say go for it anyway.
Attachment #721577 - Flags: review+
Comment on attachment 721577 [details] [diff] [review]
small cleanup

Review of attachment 721577 [details] [diff] [review]:
-----------------------------------------------------------------

::: accessible/tests/mochitest/states/test_tree.xul
@@ -76,2 @@
>        {
> -        "tree processor for " + prettyName(aTreeID);

This should have returned undefined before... ?
(In reply to Hubert Figuiere [:hub] from comment #43)

> Does that fix the "recursion" problem? The missing "return" is also
> mindboggling.

no

(In reply to David Bolter [:davidb] from comment #44)

> >        {
> > -        "tree processor for " + prettyName(aTreeID);
> 
> This should have returned undefined before... ?

(In reply to David Bolter [:davidb] from comment #45)
> s/should have/probably

yes
Keywords: helpwanted
Whiteboard: [orange][leave open] → [orange][leave open][see comment #40]
disable mochitest for now: https://hg.mozilla.org/integration/mozilla-inbound/rev/45c65a344578

we need help from someone who knows how mochitest suite works
(In reply to alexander :surkov from comment #89)
> disable mochitest for now:
> https://hg.mozilla.org/integration/mozilla-inbound/rev/45c65a344578
> 
> we need help from someone who knows how mochitest suite works

it only fails on 10.6 so we could renable it on 10.7  and 10.8 and then decide we have better things to do than figure out why its busted on 10.6 and wait a year or   two for 10.6 to go away.
Whiteboard: [orange][leave open][see comment #40] → [orange][leave open][see comment #40, comment #90]
Comment on attachment 726221 [details] [diff] [review]
Disable test only on MacOS 10.6. r=

Since we decided that a11y on Mac really only worked on 10.7, we are quite safe to disable on 10.6 without many risk.

Still worth investigating fixing though.
Attachment #726221 - Flags: review?(dbolter)
Comment on attachment 726221 [details] [diff] [review]
Disable test only on MacOS 10.6. r=

Review of attachment 726221 [details] [diff] [review]:
-----------------------------------------------------------------

r=me looks ok to me.
Attachment #726221 - Flags: review?(dbolter) → review+
Whiteboard: [orange][leave open][see comment #40, comment #90] → [test disabled on OS X 10.6][leave open][see comment #40, comment #90]
Pushed by cbook@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/2233ba87144b
Enable media dormancy everywhere for MP4 and Media Source playback - r=jya
Bulk assigning P3 to all open intermittent bugs without a priority set in Firefox components per bug 1298978.
Priority: -- → P3
Priority: P3 → P5
Resolving incomplete because we no longer support OS X 10.6. The test checks for 10.6 will be removed in bug 1336712.
Status: NEW → RESOLVED
Closed: Last year
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.