Am 6/7/2011 15:43, schrieb Alex Riesen: > Use the same test and prerequisite as introduced in similar > fix in 650af7ae8bdf92bd92df2. > > Signed-off-by: Alex Riesen <raa.lkml@xxxxxxxxx> > --- > > Why didn't Johannes noticed it, I wonder. The failing test was > here all the time. Good catch! Here's the explanation: > -test_expect_success 'a file with the same (long) magic name exists' ' > - : >":(icase)ha" && ":foo" is the name of an alternate data stream on the current directory. Bash can successfully open and write such a "file". (But there is no way to list it, even though test -f detects its existence.) > - test_must_fail git add -n ":(icase)ha" && This 'git add' fails, either for the intended reason, or because regular directory listing functions do not detect the alternate data stream; I do not know. > - git add -n "./:(icase)ha" This obviously succeeds because a regular open() call can open the alternate data stream just like any file. Therefore, the test passes. This is on Windows XP with NTFS. -- Hannes -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html