Re: bin/43751: ls -L returns wrong exit code

["Valeriy E. Ushakov" <uwe%stderr.spb.ru@localhost>]

The following reply was made to PR bin/43751; it has been noted by GNATS.

From: "Valeriy E. Ushakov" <uwe%stderr.spb.ru@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: bin/43751: ls -L returns wrong exit code
Date: Fri, 13 Aug 2010 06:42:36 +0400

 On Thu, Aug 12, 2010 at 18:17:29 -0700, john heasley wrote:
 
 > Thu, Aug 12, 2010 at 08:50:04PM +0000, Valeriy E. Ushakov:
 > 
 > Crap, you are right.  my 4.99 test had the FQPN, as you noted.
 > 
 > >  Basically, we use -L only to select FTS_LOGICAL instead of
 > >  FTS_PHYSICAL.  printing code doesn't know whether -L was specified.
 > >  For links with non-existent targets fts_children(3) returns the link
 > >  info (as it would for FTS_PHYSICAL) and printing code just happily
 > >  prints it.
 > 
 > Does it need to?  something like this is a quick fix.
 > 
 > Index: ls.c
 > ===================================================================
 > RCS file: /cvsroot/src/bin/ls/ls.c,v
 > retrieving revision 1.67
 > diff -d -u -d -u -r1.67 ls.c
 > --- ls.c     8 Jul 2010 20:43:34 -0000       1.67
 > +++ ls.c     13 Aug 2010 00:52:59 -0000
 > @@ -389,6 +389,7 @@
 >  static void
 >  traverse(int argc, char *argv[], int options)
 >  {
 > +    char path[MAXPATHLEN + 1];
 >      FTS *ftsp;
 >      FTSENT *p, *chp;
 >      int ch_options, error;
 > @@ -442,6 +443,13 @@
 >                      if (!f_recursive && chp != NULL)
 >                              (void)fts_set(ftsp, p, FTS_SKIP);
 >                      break;
 > +            case FTS_SLNONE:
 > +                    if (options & FTS_LOGICAL) {
 > +                        readlink(p->fts_path, path, sizeof(path) - 1);
 > +                        errno = ENOENT;
 > +                        warn("cannot access %s", path, p->fts_name);
 > +                        rval = EXIT_FAILURE;
 > +                    }
 >              }
 >      error = errno;
 >      (void)fts_close(ftsp);
 
 Shouldn't this check be in display()?
 
 You don't have to readlink(), the error is reported for the link name,
 not the target name.
 
 -uwe