Hi, Please help codereview the change for JDK-6233323/8144977
issue: https://bugs.openjdk.java.net/browse/JDK-6233323 https://bugs.openjdk.java.net/browse/JDK-8144977 webrev: http://cr.openjdk.java.net/~sherman/62333233_8144977/webrev This is an implementation issue from day-one. ZIP spec specifies the name of a directory entry must be ended with a slash, so a directory name "dir" in ZIP world should have the name "dir/". ZipFile.getEntry() works well with these type of "canonicalized" directory names. However, its implementation also tries to be "liberal" when taking name that doesn't end with "/". Where there is no such entry (file) found, it tries a second lookup with a "/" appended, and returns such entry if find one. The original implementation always to use the original passed in name (the one without the ending slash) as the entry name for the returned ZipEntry. This is fine if the entry is from the first round of lookup, in which the name stored in zip file matches the passed in name and to use the name directly obviously helps avoiding creating another String object from the binary name (byte[] in zip entry). But in case of the entry is from the second round lookup, the direct consequence is the isDirectory() method of the returned entry always returns false (ZipEntry.isDirectory() implementation simply checks if the name ends with / or not), even it is actually a directory entry. The fix is to simply return the name stored in the zip file. It appears this should be the desired and expected result. Thanks, Sherman
