Matthew Burgess wrote: > The unibyte-bracket-expr test does 'cp $in /t' which, if the tests are > being run as root, will cause '/t' to be created (and obviously, if > the tests aren't being run as root, that command should fail). The > attached patch simply removes that copy command, as that didn't affect > the status of the test on my machine (x86 Linux) and I couldn't see > any other reference to '/t'.
Thank you. How embarrassing. That's a testing artifact I should never have committed.
