Closed Bug 600785 Opened 11 years ago Closed 11 years ago

TM: intermittent failure in browser_bug471404.js | Test timed out after JavaScript Error: "dirSvc.unregisterProvider is not a function"


(Toolkit :: Crash Reporting, defect)

Not set



Tracking Status
blocking2.0 --- -


(Reporter: orangereporter, Assigned: dbaron)



(Keywords: intermittent-failure)


(1 file)

Not immediately obvious to me whether this is the test doing something it shouldn't, or jseng doing something it shouldn't, but sometime today browser_bug471404.js started intermittently failing like
Rev3 WINNT 6.1 tracemonkey opt test mochitest-other on 2010/09/29 17:35:51
s: talos-r3-w7-033

TEST-INFO | chrome://mochikit/content/browser/toolkit/crashreporter/test/browser/browser_bug471404.js | Console message: [JavaScript Error: "dirSvc.unregisterProvider is not a function" {file: "chrome://mochikit/content/browser/toolkit/crashreporter/test/browser//aboutcrashes_utils.js" line: 60}]
TEST-UNEXPECTED-FAIL | chrome://mochikit/content/browser/toolkit/crashreporter/test/browser/browser_bug471404.js | Test timed out

where is just

let dirSvc = Cc[";1"]
Rev3 WINNT 6.1 tracemonkey opt test mochitest-other on 2010/09/29 20:43:55
s: talos-r3-w7-036
Rev3 MacOSX Leopard 10.5.8 tracemonkey opt test mochitest-other on 2010/09/29 22:04:35
s: talos-r3-leopard-044
blocking2.0: --- → ?
Rev3 WINNT 6.1 tracemonkey opt test mochitest-other on 2010/09/29 16:55:58
s: talos-r3-w7-045
Rev3 MacOSX Snow Leopard 10.6.2 tracemonkey debug test mochitest-other on 2010/09/29 17:25:35
s: talos-r3-snow-012
Blocks: 600883
No longer blocks: 600883
Assignee: general → dbaron
Component: JavaScript Engine → Breakpad Integration
Product: Core → Toolkit
QA Contact: general → breakpad.integration
Attached patch patchSplinter Review
Pretty sure this is a bug in the test, which is racing with GC of the wrapper.
Attachment #481861 - Flags: review?(ted.mielczarek)
Comment on attachment 481861 [details] [diff] [review]

Kind of awful that we have to think about wrappers like this. :-/ Is there no sane way to make sure that the wrapper hangs around? I can see us making this mistake fairly often, since it violates my expectations.
Attachment #481861 - Flags: review?(ted.mielczarek) → review+
You can make the wrapper hang around by storing it in a global variable or property of something else; that has other dangers, though.
Perhaps the better answer is that if the object has classinfo, you don't have to mess with QI at all.
Closed: 11 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla2.0b8
blocking2.0: ? → -
Target Milestone: mozilla2.0b8 → mozilla2.0b7
Whiteboard: [orange]
You need to log in before you can comment on or make changes to this bug.