Bug 143779

Summary: [Win] DRT does not seem to reset the 'UsesPageCache' setting between tests
Product: WebKit Reporter: Chris Dumez <cdumez>
Component: Tools / TestsAssignee: Chris Dumez <cdumez>
Status: RESOLVED FIXED    
Severity: Normal CC: ap, bfulgham
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on:    
Bug Blocks: 143463    
Attachments:
Description Flags
Patch
none
Patch none

Description Chris Dumez 2015-04-15 09:42:40 PDT
DRT does not seem to reset the 'UsesPageCache' settings between tests on Windows. This should be done in resetWebPreferencesToConsistentValues() in DumpRenderTree.cpp, similarly to what is done in DumpRenderTree.mm on Mac / WK1.
Without test, PageCache could stay enabled after page cache tests (They enable PageCache explicitly via 'testRunner.overridePreference("WebKitUsesPageCachePreferenceKey", 1)') and cause weird behaviors.
Comment 1 Chris Dumez 2015-04-15 09:45:06 PDT
Created attachment 250800 [details]
Patch
Comment 2 Chris Dumez 2015-04-15 09:47:07 PDT
Created attachment 250801 [details]
Patch
Comment 3 Brent Fulgham 2015-04-15 09:50:34 PDT
Comment on attachment 250801 [details]
Patch

r=me.
Comment 4 Chris Dumez 2015-04-15 09:54:59 PDT
Comment on attachment 250801 [details]
Patch

Clearing flags on attachment: 250801

Committed r182844: <http://trac.webkit.org/changeset/182844>
Comment 5 Chris Dumez 2015-04-15 09:55:03 PDT
All reviewed patches have been landed.  Closing bug.
Comment 6 Chris Dumez 2015-04-15 17:26:13 PDT
FYI, after this change. The number of new passes went from 28 to 32 on the Windows Debug bot. Also, the flakes went down from 9 to 1. Similar results on the Windows Release bot.

So it looks like the change was indeed useful :)
Comment 7 Brent Fulgham 2015-04-15 17:27:18 PDT
I'm also wondering if (In reply to comment #6)
> FYI, after this change. The number of new passes went from 28 to 32 on the
> Windows Debug bot. Also, the flakes went down from 9 to 1. Similar results
> on the Windows Release bot.
> 
> So it looks like the change was indeed useful :)

That's wonderful news! I wonder if I can re-enable a few of the tests I identified as messing up the page cache previously. All of those symptoms might have been caused by this bug you found.