On Sat, Sep 29, 2001 at 09:26:47AM -0700, Justin Erenkrantz wrote: > On Sat, Sep 29, 2001 at 12:01:33PM -0400, Cliff Woolley wrote: > > On 29 Sep 2001 [EMAIL PROTECTED] wrote: > > > > > if (strncmp(rnew->filename, fdir, fdirlen) == 0 > > > && rnew->filename[fdirlen] > > > - && ap_strchr_c(rnew->filename + fdirlen, '/') == NULL) > > > + && ap_strchr_c(rnew->filename + fdirlen, '/') == NULL > > > + && strlen(r->uri) != 0) > > > > Might I suggest "&& r->uri[0] != '\0'" or some such? You don't really > > need the string's length, just whether it's zero or not. Constant time > > instead of linear is always nice. =-) > > Ah, yes, that'd be better. But, I have a feeling OtherBill will be > reverting it, so I won't muck anymore with it. If he says it is > the right thing to do, then we can change it to that... -- justin >
I think ap_sub_req_lookup_filename is still broken. (Or at least acting differently than it used to) If a full path is given as the first argument, a 403 is returned. This is not caught by the perl test framework since all the tests that use ap_sub_req_lookup_file use a relative path. I'll try to look into this tomorrow. -Ryan
