Re: [bug-inetutils] [Platform-testers] inetutils pre-release

[Mats Erik Andersson]

torsdag den 17 november 2011 klockan 23:20 skrev Mats Erik Andersson detta:
> måndag den 14 november 2011 klockan 11:21 skrev Simon Josefsson detta:
> > Bruno Haible <address@hidden> writes:
> > > In other words, you need to add $(LIBREADLINE) to the value of LDADD
> > > in telnet/Makefile.am.
> > >
> > > Likewise in telnetd/Makefile.am.
> > 
> > Thanks for analyzing, I agree and pushed this tiny change in your name.
> > 
> > > Quite odd situation, but IMO the readline macro should not set 
> > > HAVE_LIBREADLINE
> > > if the include files are missing, or the code in ftp/cmds.c should be 
> > > testing
> > >   HAVE_LIBREADLINE && HAVE_READLINE_READLINE_H
> > > instead of just
> > >   HAVE_LIBREADLINE
> > 
> > Yes.  However, wouldn't an even better solution be to use
> > AC_LIB_LINKFLAGS here instead of the current check for readline?  That
> > should be more reliable.
> 
> Related to this is the fact that <term.h> must be prepended by <curses.h>
> in all standards. Also <stropts.h> is needed with HAVE_STREAMSPTY.
> 
> The commit 0a83fb1 attends to these matters and it achieves correct
> building of telnet and telnetd for
> 
>   GNU/Linux, GNU/kFreeBSD, GNU/OpenSolaris, OpenBSD, and OpenIndiana.

Quick report on GNU/OpenSolaris (snv_134f) and OpenIndiana (oi_151a).

Our Telnet server is functional on both, except that the environmental
variable TERM is corrupt on both these systems:

   $ env
   TERM=TERM=xterm
and
   $ env
   TERM=TERM=linux

It does not matter if the client be IU, OpenBSD, or Netkit, in every
case the tokenization is not splitting at the equality sign. My guess
is that there is a defect in "/bin/login" as found in SUNW.

Once the above setting is repaired using

   $ export TERM=xterm

then scrolling as well as Bash's terminal functionality return, so we
now know that the committed patch indeed achieves the intended func-
tionality (in the case of non-Kerberos support).

Best regards,
  Mats