From: "Cliff Woolley" <[EMAIL PROTECTED]> Sent: Monday, September 17, 2001 12:05 AM > It did up until about two weeks ago, but then that was because r->uri was > set to a completely bogus value (the string "INTERNALLY GENERATED > file-relative req" or something like that), rather than just "". =-) I > don't think r->uri can be actually NULL, though (somebody correct me if > I'm wrong). We haven't decided what should become of r->uri in a relative file lookup yet, have we? I'd suggest the undecorated (relative) file name itself, e.g. "file.foo". Clearly invalid (it's relative) yet protective (it hits all tests except <location >, and other applications interested in r->uri won't be overly confused). We can set an apr_pool_userdata element to flag that it is a legitimate exception, and filter those out again later in the request parsing. Bill