Last Comment Bug 591747 - [SeaMonkey 2.0] mochitest-other: "test_bug428405.xul | [SimpleTest/SimpleTest.js, window.onerror] An error occurred - gChromeWindow is null"
: [SeaMonkey 2.0] mochitest-other: "test_bug428405.xul | [SimpleTest/SimpleTest...
Status: VERIFIED FIXED
: regression, verified1.9.1
Product: Toolkit
Classification: Components
Component: Download Manager (show other bugs)
: 1.9.1 Branch
: x86 Mac OS X
: -- major (vote)
: ---
Assigned To: Justin Wood (:Callek)
:
Mentors:
http://mxr.mozilla.org/mozilla1.9.1/s...
Depends on:
Blocks: SmTestFail 544356
  Show dependency treegraph
 
Reported: 2010-08-29 03:15 PDT by Serge Gautherie (:sgautherie)
Modified: 2010-09-23 12:08 PDT (History)
5 users (show)
bugzillamozillaorg_serge_20140323: in‑testsuite+
See Also:
Crash Signature:
(edit)
QA Whiteboard:
Iteration: ---
Points: ---
Has Regression Range: ---
Has STR: ---
.14-fixed


Attachments

Description Serge Gautherie (:sgautherie) 2010-08-29 03:15:41 PDT
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey2.0/1283066967.1283067319.14360.gz
OS X 10.5 comm-1.9.1 test mochitest-other on 2010/08/29 00:29:27
{
6946 ERROR TEST-UNEXPECTED-FAIL | chrome://mochikit/content/chrome/widget/test/test_bug428405.xul | [SimpleTest/SimpleTest.js, window.onerror] An error occurred - gChromeWindow is null
}
Comment 2 Steven Michaud [:smichaud] (Retired) 2010-08-29 08:53:11 PDT
Does this test always fail?

Does it also fail for Firefox?

And why do you conclude this is a Cocoa Widgets bug?  It seems much
more likely to be some kind of cross-platform bug -- one that makes
nsIWindowMediator.getMostRecentWindow("navigator:browser") return
null.
Comment 3 Serge Gautherie (:sgautherie) 2010-08-29 11:18:03 PDT
(In reply to comment #2)

> Does this test always fail?

Yes, afaict (though I didn't check all builds).

> Does it also fail for Firefox?

No:
{
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox3.5/1282941921.1282942396.4363.gz&fulltext=1
OS X 10.5.2 mozilla-1.9.1 test mochitest-other on 2010/08/27 13:45:21

1398 INFO Running chrome://mochikit/content/chrome/widget/test/test_bug428405.xul...
1399 INFO TEST-PASS | chrome://mochikit/content/chrome/widget/test/test_bug428405.xul | undefined
1401 INFO Running chrome://mochikit/content/chrome/widget/test/test_bug444800.xul...
}
Though 'undefined' always looks odd to me...

> And why do you conclude this is a Cocoa Widgets bug?  It seems much

Sorry, that was just a leftover from when I initially looked at bug 428405.

> more likely to be some kind of cross-platform bug -- one that makes
> nsIWindowMediator.getMostRecentWindow("navigator:browser") return
> null.

Actually, I had a typo in the regression url, which confused me...
Correct regression timeframe:
http://hg.mozilla.org/releases/mozilla-1.9.1/pushloghtml?fromchange=28e2ed70bd32&tochange=4b76384e608e
Bug 544356 !

Neil, do we need to port the browser.js change ? Or ...
Comment 4 neil@parkwaycc.co.uk 2010-08-29 11:41:52 PDT
Don't we already have the equivalent of the browser.js code in isClosingLastBrowser() ?

Does the test for bug 544356 pass? That test changes the browser's windowtype and if it's not restored correctly then subsequent tests will fail.
Comment 5 Serge Gautherie (:sgautherie) 2010-08-29 14:11:10 PDT
(In reply to comment #4)

> Don't we already have the equivalent of the browser.js code in
> isClosingLastBrowser() ?

(
Yes, part of it at least:
http://hg.mozilla.org/releases/mozilla-1.9.1/annotate/4b76384e608e/browser/base/content/browser.js#l5762
http://hg.mozilla.org/releases/comm-1.9.1/annotate/7bd5a7cf4b6a/suite/browser/navigator.js#l2470
)

> Does the test for bug 544356 pass? That test changes the browser's windowtype
> and if it's not restored correctly then subsequent tests will fail.

http://tinderbox.mozilla.org/showlog.cgi?tree=SeaMonkey2.0&errorparser=unittest&logfile=1283066967.1283067319.14360.gz&buildtime=1283066967&buildname=OS%20X%2010.5%20comm-1.9.1%20test%20mochitest-other&fulltext=1
OS X 10.5 comm-1.9.1 test mochitest-other on 2010/08/29 00:29:27
{
6825 INFO Running chrome://mochikit/content/chrome/toolkit/mozapps/downloads/tests/chrome/test_close_on_last_window.xul...
6826 INFO TEST-KNOWN-FAIL | chrome://mochikit/content/chrome/toolkit/mozapps/downloads/tests/chrome/test_close_on_last_window.xul | skip test for toolkit download manager UI
}

http://mxr.mozilla.org/mozilla1.9.1/source/toolkit/mozapps/downloads/tests/chrome/test_close_on_last_window.xul
Indeed, it seems the test misses to set the window type back (or shouldn't change it that early) in this case.

NB: And the other skipped D.M. tests may need to be checked too...
Comment 6 Justin Wood (:Callek) 2010-08-29 14:34:04 PDT
didn't I/We fix this on trunk?
Comment 7 Serge Gautherie (:sgautherie) 2010-08-30 07:59:43 PDT
On trunk,
{
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1283167369.1283169450.9901.gz&fulltext=1
OS X 10.5 comm-central-trunk debug test mochitest-other on 2010/08/30 04:22:49

13219 INFO TEST-KNOWN-FAIL | chrome://mochikit/content/chrome/toolkit/mozapps/downloads/tests/chrome/test_close_on_last_window.xul | skip test for toolkit download manager UI

17388 INFO TEST-PASS | chrome://mochikit/content/chrome/widget/tests/test_bug428405.xul | undefined - true should equal true
}
The test itself hasn't changed, so something else must have...
Comment 8 Serge Gautherie (:sgautherie) 2010-08-30 08:05:00 PDT
(In reply to comment #7)
> The test itself hasn't changed, so something else must have...

Ah, that was about test_bug428405.xul...

***

(In reply to comment #6)
> didn't I/We fix this on trunk?

It seems m-1.9.1 actually misses
http://hg.mozilla.org/mozilla-central/rev/58d7aaf82913
(No Bug) Make test not modify the windowtype until certain we are actually running the test. Fixes some SeaMonkey orange. rs=Mossop
:-/
Comment 10 Serge Gautherie (:sgautherie) 2010-08-31 20:03:55 PDT
V.Fixed, per
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey2.0/1283307989.1283308336.2113.gz
OS X 10.5 comm-1.9.1 test mochitest-other on 2010/08/31 19:26:29
Comment 11 Al Billings [:abillings] 2010-09-23 12:08:23 PDT
Marking as verified for 1.9.1 based on Serge's comment 10.

Note You need to log in before you can comment on or make changes to this bug.