Columbia Kermit/kermit.zip/newsgroups/misc.20000114-20000217/[e-mail] _Tue Jan 18 16:26:08 2000.msg

Internet Message Format | 2020-01-01 | 5KB

Return-Path: <[e-mail]>
Received: from newsmaster.cc.columbia.edu (newsmaster.cc.columbia.edu [128.59.59.30])
    by watsun.cc.columbia.edu (8.8.5/8.8.5) with ESMTP id QAA08195
    for <[e-mail]>; Tue, 18 Jan 2000 16:26:08 -0500 (EST)
Received: (from news@localhost)
    by newsmaster.cc.columbia.edu (8.8.5/8.8.5) id QAA28515
    for [e-mail]; Tue, 18 Jan 2000 16:24:39 -0500 (EST)
X-Authentication-Warning: newsmaster.cc.columbia.edu: news set sender to <news> using -f
From: [e-mail] (Frank da Cruz)
Subject: Case Study #11: C-Kermit Meets SSH
Date: 18 Jan 2000 21:24:36 GMT
Organization: Columbia University
Message-ID: <862lmk$rr1$[e-mail]>
To: [e-mail]


One of the most frequent feature requests for Kermit software in recent
years has been for SSH (Secure SHell) connections.  Although other security
methods including Kerberos, SRP, and SSL/TLS are available in C-Kermit 7.0,
we have not been able to add SSH, primarily for licensing and patent reasons.

But with the UNIX version of C-Kermit 7.0 you can make SSH connections
anyway.  This gives you C-Kermit's file transfer, character-set translation,
and scripting capabilities on an SSH connection.

The key is C-Kermit's new PTY command.  Pty (pronounced "pity") is the
common abbreviation for "pseudoterminal".  A pseudoterminal is a virtual
(simulated) device that has the characteristics of a real terminal; a Pty is
used by UNIX as your controlling terminal when you come in via Telnet,
Rlogin, SSH, or other "virtual" means, rather than through a serial port
with its "real" terminal driver.  The Pty driver mimics what the real
terminal driver does on a real terminal.

Ptys are also used by programs like 'expect' that run and interact with
other programs.  Ptys are used for this rather than simple standard i/o
redirection because (a) not all programs use standard i/o; (b) many programs
behave differently when their standard i/o is not a terminal; and (c)
redirected standard i/o has no associated terminal driver.

Unfortunately the application programming interface (API) for using Ptys is
not standardized across Unix versions (let alone between Unix and other
operating-system families), so each platform requires custom code.  We were
able to add Pty control to C-Kermit 7.0 for the following platforms:

 . 4.4BSD, including BSDI/OS, NetBSD, FreeBSD, OpenBSD, Mac OS X
 . DG/UX 5.4R4.11
 . Digital UNIX 3.2 and Tru64 UNIX 4.0
 . HP-UX 9.00 and later
 . IBM AIX 4.1 and later
 . IRIX 6.0 and later
 . Linux
 . NeXTSTEP 3.x
 . QNX 4.25
 . SCO OSR5.0.5
 . SCO Unixware 7
 . SINIX 5.42
 . Solaris 2.x and 7
 . SunOS 4.1.3

It can be added for others too but some degree of work would be required
in each case -- maybe a little, maybe a lot.

The Pty interface has many uses.  Today we'll talk about SSH since so many
people ask about it.  Let's begin by making an interactive SSH connection,
just as you would with the regular SSH client:

  ssh xyzcorp.com

To do this with C-Kermit, simply prefix the command above with "pty" and
give it at the C-Kermit> prompt:

  C-Kermit> pty ssh xyzcorp.com

You can escape back to C-Kermit and re-CONNECT in the normal way, but
remember that the SSH client is active too; Newline (Return or Enter key)
followed by tilde (~) is the SSH escape sequence, which you can read about
in "man ssh".

Once you have a terminal connection, can it be used to transfer files?  Yes,
but first you'll need disable control-character unprefixing because the Pty
driver can "do things" to control characters (and also to prevent sending
the SSH escape sequence in a packet).  Luckily, Kermit protocol is
well-suited to such interference.  Before initiating a file transfer, just
tell the file sender to:

  set prefixing all

That should be all you need (in fact, you don't have to prefix EVERY control
character, but the fine tuning is up to you).  Assuming you have fast Kermit
protocol settings (which C-Kermit 7.0 uses by default), performance is
limited by network and/or the underlying pty driver -- some are fast, some
aren't.

Scripting SSH connections is easy too.  As you might recall, a Kermit
script uses SET HOST instead of TELNET and it uses INPUT and OUTPUT
commands in place of CONNECT and your eyes and fingers.  The PTY command is
like the TELNET command; it makes the connection and then goes into CONNECT
(online interactive terminal) mode.  To open an SSH connection without
entering CONNECT mode, use:

  set host /pty ssh xyzcorp.com

and then write the rest of your script in the normal manner, except with
caution regarding the ssh client's escape sequence.  Note the new /PTY
switch on the SET HOST command; type "set host ?" at the C-Kermit> prompt
to see other new switches, and "help set host" to learn more about them.

For complete details on C-Kermit 7.0's Pty interface, read Section 2.7 of
ckermit2.txt.  For more about why SSH is not included directly in C-Kermit,
see:

  http://www.columbia.edu/kermit/ckfaq.html#ssh

If you're a programmer and want to add Pty support for a platform not listed
above, let me know.

Finally, here's an exercise for the interested reader: rewrite this article
under the new title, "C-Kermit Meets Tn3270".

- Frank