Re: [bug-inetutils] Present release goals

[Alfred M. Szmidt]

   The issues we need to attend to for the immediate release are:

      * The revision of the autoconf/automake settings
        must be completed. Presently the work in June
        broke the functionality for OpenBSD. (There is
        a "struct option" that leads to a failure.

        The acceptance test for this step is that
        a bootstrapped source tree must be movable
        via "rsync -CLa from/ host:/to/" to all of
        GNU/Linux, GNU/kFreeBSD, FreeBSD, and OpenBSD,
        and be configurable and compilable without
        errors. These are the four systems we can
        claim support for!

Good!  I'll work on it, if you can give me access to any such boxes
that would be grand.  I might give it a shot in qemu, but I'll
probobly just blindly try and fix this issue.

      * The problem with rpl_ioctl() and GNUlib for
        GNU/kFreeBSD [1] must be resolved. It is beyond
        my understanding why Simon is not attending to
        this problem, be it that he prefers GNU/Hurd.

I'll follow it up.   

      * tftpd failing for IPv4-mapped-to-IPv6 [2].
        (Ignore for the time being problems in OpenSolaris.)
        The solution must be checked on GNU/Linux,
        GNU/kFreeBSD, and FreeBSD, with a dual stacked
        superserver and "bindv6only=0".

I'll look over the tftp patch.

      * I have prepared a recalculated for the TFTP client,
        which got delayd as I discovered the preceding error.
        Should we include this?

Is it safe? If so, yes.

      * The test script for syslogd [3] must be adapted
        to automated testing. The script has been used
        manually by me, and all its features were instrumental
        to me in uncovering segfaults and verification of new
        code additions. It is best that someone else makes the
        adaptions to automatic testing.

Yes, we definitly should add more tests.

      * The problems claimed in "tests/tftd.sh" on the build
        engine Hydra of NixOS should probably be ignored.
        The mail conversations showed that their setup does
        not even include "/etc/services" in the right location,
        so they are severely breaking every sane rule of thumb.

Yup, I'm not going to bother with it.  We can use the GCC build farm
instead, they have a few weirdo setups.

      * Update "doc/inetutils.texi".

      * Write NEWS.

      * Test the utils again and again!

I'll do those, and try to write some new tests.


   Next release
   ============

   I believe these are sufficient to arrive at a worthy update.

      * Implement IPv6 support in "ftpd". At the moment I have done
        this in a preliminary patch set, which runs as expected on
        OpenBSD. Should the "EPSV ALL" be implemented, there is an
        old bug in the FTP client that must be attended to also.
        That malfunction is present in "netkit-ftp" in Debian,
        in "ftp" of OpenBSD, and "ftp" of FreeBSD, i.e., in all
        descendants of Netkit, so there is no reason for panic.

We can pick the easy solution, document this in the manual.

      * Support OpenSolaris. Besides iruserok() [4], the main
        obstacle for OpenSolaris is the Kerberos code. There
        is also a known problem in "libinetutils/tftpsubs.c",
        which must be debugged properly.

I'm looking for a Solaris box to test on.

      * Rework Kerberos support, separating Kerberos4 code
        from KerberosV code, and make proper implementation
        of Shishi code that works on all platforms we want
        to support.

Maybe Simon could do this? Simon, do you have time?  I'm not at all
familiar with Kerberos.

Mats, yet again, thanks for picking up the ball!