Thanks for the review Prasanta. However, I don't see a point to write a test case for isFileSystemRoot(), since, it is not going to fail on any (older/newer) java versions, and it was only introduced because of the fix for JDK-8175015.
Let me know if you think otherwise. Regards, Krishna From: Prasanta Sadhukhan Sent: Wednesday, January 3, 2018 11:27 AM To: Krishna Addepalli <krishna.addepa...@oracle.com>; swing-dev@openjdk.java.net Subject: Re: <Swing Dev> [10][11][JDK-8194044] Regression manual Test javax/swing/JFileChooser/8067660/FileChooserTest.java fails Fix looks fine. But I guess, it is possible to add a automated regression test to it utilising isFileSystemRoot(). Regards Prasanta On 1/2/2018 4:31 PM, Krishna Addepalli wrote: Hi All, Please review a fix for bug: Bug: JDK-8194044 : https://bugs.openjdk.java.net/browse/JDK-8194044 Webrev: HYPERLINK "http://cr.openjdk.java.net/%7Ekaddepalli/8194044/webrev00/"http://cr.openjdk.java.net/~kaddepalli/8194044/webrev00/ This was caused due to the fix for JDK-8175015, in which the line 446 in Win32ShellFolderManager2.java was changed from "getDrives()" to Win32ShellFolder2.listRoots(). While the earlier function returns an object of Win32ShellFolder2, the latter returns an array of Files. The condition on line 450: "return (sf.isFileSystem()&&sf.parent != null && sf.parent.equals(Win32ShellFolder2.listRoots())" was returning false because of the wrong object being passed. Earlier it was a Win32ShellFolder2 object, and the comparision was done properly, but with the changes, the equals fucnction was receiving a file array object, and hence it was immediately returning false, leading to the problem of empty strings being shown for Root drives. The fix is to replace "Win32ShellFolder2.listRoots()" with "getDrives()" function. With this fix, the regression is addressed, as well as the original JDK-8175015 which was a memory leak issue. Thanks, Krishna
