If we currently have 2 IDB databases: IDB1: top level frame origin A with subframe origin A IDB2: top level frame origin B with subframe origin A IDB2 won't be deleted if user asks to delete all IDB databases for origin A. This is because we are managing IDB database deletion by their mainframe origin.
<rdar://problem/42164227>
Created attachment 345004 [details] Patch
Comment on attachment 345004 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=345004&action=review > Tools/TestWebKitAPI/Tests/WebKitCocoa/WebsiteDataStoreCustomPaths.mm:-136 > - auto pid = [webView _webProcessIdentifier]; > - if (pid) > - kill(pid, SIGKILL); This seems unrelated. You mentioned you think it's the cause of some flakiness and there's a bug for it. We should implement it over in that bug. It seems weird that we thought we needed to reset all processes but this change would stop doing that.
Created attachment 345101 [details] Patch for landing
Comment on attachment 345101 [details] Patch for landing Clearing flags on attachment: 345101 Committed r233853: <https://trac.webkit.org/changeset/233853>
All reviewed patches have been landed. Closing bug.
Jut that after this patch (In reply to WebKit Commit Bot from comment #5) > Comment on attachment 345101 [details] > Patch for landing > > Clearing flags on attachment: 345101 > > Committed r233853: <https://trac.webkit.org/changeset/233853> Just want to note that after this patch we are getting a different failures at a different line number. Not sure if this patch is dependent on https://bugs.webkit.org/show_bug.cgi?id=187066#c6 for a complete fix? TestWebKitAPI.WebKit.WebsiteDataStoreCustomPaths /Volumes/Data/slave/highsierra-release/build/Tools/TestWebKitAPI/Tests/WebKitCocoa/WebsiteDataStoreCustomPaths.mm:220 Value of: [[NSFileManager defaultManager] fileExistsAtPath:fileIDBPath.get().path] Actual: true Expected: false
(In reply to David Fenton (:realdawei) from comment #7) > Jut that after this patch (In reply to WebKit Commit Bot from comment #5) > > Comment on attachment 345101 [details] > > Patch for landing > > > > Clearing flags on attachment: 345101 > > > > Committed r233853: <https://trac.webkit.org/changeset/233853> > > Just want to note that after this patch we are getting a different failures > at a different line number. Not sure if this patch is dependent on > > https://bugs.webkit.org/show_bug.cgi?id=187066#c6 for a complete fix? > > > TestWebKitAPI.WebKit.WebsiteDataStoreCustomPaths > > > /Volumes/Data/slave/highsierra-release/build/Tools/TestWebKitAPI/Tests/ > WebKitCocoa/WebsiteDataStoreCustomPaths.mm:220 > Value of: [[NSFileManager defaultManager] > fileExistsAtPath:fileIDBPath.get().path] > Actual: true > Expected: false Yes. With 187066's patch, the test should pass.
(In reply to Sihui Liu from comment #8) > (In reply to David Fenton (:realdawei) from comment #7) > > Jut that after this patch (In reply to WebKit Commit Bot from comment #5) > > > Comment on attachment 345101 [details] > > > Patch for landing > > > > > > Clearing flags on attachment: 345101 > > > > > > Committed r233853: <https://trac.webkit.org/changeset/233853> > > > > Just want to note that after this patch we are getting a different failures > > at a different line number. Not sure if this patch is dependent on > > > > https://bugs.webkit.org/show_bug.cgi?id=187066#c6 for a complete fix? > > > > > > TestWebKitAPI.WebKit.WebsiteDataStoreCustomPaths > > > > > > /Volumes/Data/slave/highsierra-release/build/Tools/TestWebKitAPI/Tests/ > > WebKitCocoa/WebsiteDataStoreCustomPaths.mm:220 > > Value of: [[NSFileManager defaultManager] > > fileExistsAtPath:fileIDBPath.get().path] > > Actual: true > > Expected: false > > Yes. With 187066's patch, the test should pass. Super! thanks.