[Webkit-unassigned] [Bug 145626] New: Non-deterministic Inspector Pause and Resume events makes tests flaky
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Wed Jun 3 18:12:38 PDT 2015
https://bugs.webkit.org/show_bug.cgi?id=145626
Bug ID: 145626
Summary: Non-deterministic Inspector Pause and Resume events
makes tests flaky
Classification: Unclassified
Product: WebKit
Version: 528+ (Nightly build)
Hardware: Unspecified
OS: Unspecified
Status: NEW
Severity: Normal
Priority: P2
Component: Web Inspector
Assignee: webkit-unassigned at lists.webkit.org
Reporter: mark.lam at apple.com
CC: graouts at webkit.org, joepeck at webkit.org,
jonowells at apple.com, mattbaker at apple.com,
nvasilyev at apple.com, timothy at apple.com,
webkit-bug-importer at group.apple.com
DebuggerManager.debuggerDidResume() defers the dispatch of the Resumed event. DebuggerManager.debuggerDidPause() can eat that Resumed event as well as the next Paused Event if debuggerDidPause() happens to be called before debuggerDidResume()'s delayedWork() fires. As a result, we can cannot write tests that can depend on a consistent order of Pause and Resume events.
I was advised to use the CallFramesDidChange event instead. The problem with this is that the number of CallFramesDidChange events firing also depends on the race between debuggerDidPause() and debuggerDidResume()'s delayedWork(). For example, the regression test of https://bugs.webkit.org/show_bug.cgi?id=145525 (which is currently written to assume that some Paused and Resume events will be eaten) runs fine on WK1 using DumpRenderTree. When we run it on WK2 using WKRT which runs a lot slower, the events don't get eaten. As a result, the test fails.
--
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20150604/70b77b92/attachment.html>
More information about the webkit-unassigned
mailing list