Internet Engineering Task Force                            Frank da Cruz
INTERNET-DRAFT                                         Jeffrey E. Altman
<draft-columbia-kermit-service-02.txt>               Columbia University
                                                         25 January 2000
                                                   Expires: 25 July 2000


                        INTERNET KERMIT SERVICE

                                DRAFT 02


STATUS OF THIS MEMO

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC2026.  Internet-Drafts are working
   documents of the Internet Engineering Task Force (IETF), its areas,
   and its working groups.  Note that other groups may also distribute
   working documents as Internet-Drafts.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference mate-
   rial or to cite them other than as "work in progress."

   The list of current Internet-Drafts can be accessed at
   http://www.ietf.org/ietf/1id-abstracts.txt

   The list of Internet-Draft Shadow Directories can be accessed at
   http://www.ietf.org/shadow.html.

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC 2119.


ABSTRACT

This document describes a new file transfer service for the Internet
based on Telnet Protocol for option negotiation and Kermit Protocol for
file transfer and management.  The Internet Kermit Service provides
access to both authenticated and anonymous users.  The use of Kermit
protocol over a Telnet connection provides several advantages over FTP,
including easy traversal of firewalls, transfers over multiple
transports, and security via a combination of supported Telnet
authentication and encryption option negotiations, plus significant
functional benefits.  While this document describes a new service for
the Internet, the clients for this service already exist on most
platforms in the form of Telnet clients that support the Kermit file
transfer protocol.  These clients are available not only from Columbia
University's Kermit Project but also numerous third parties.


CONTENTS

  0. PREFACE
  1. INTRODUCTION
  2. BACKGROUND
  2.1. History
  2.2. Motivation
  3. THE INTERNET KERMIT SERVICE MODEL
  3.1. Server-Side Kermit Server
  3.2. Client-Side Kermit Server
  3.3. Loosely Coupled Operation
  4. SECURITY CONSIDERATIONS
  4.1. AUTHENTICATION
  4.1.1. Telnet Authentication
  4.1.2. Plaintext Authentication via Kermit REMOTE LOGIN
  4.1.3. Plaintext Authentication via Command Prompt
  4.1.4. Anonymous Login
  4.2. ENCRYPTION (PRIVACY)
  4.2.1  Telnet Encryption
  4.2.2  Telnet Start_TLS
  5. SERVICES
  5.1. Features for System Administrators
  5.2. Features for Users
  5.3. User Interface
  6. REFERENCES
  7. AUTHORS' ADDRESS


0. PREFACE

This Internet-Draft is meant to be used as reference material for the
Kermit/Telnet Birds of a Feather session which will be held at the 43rd
IETF meeting in Orlando, Florida, USA during the week of 7 December
1998.  It is hoped that this document will provide necessary background
material for serious discussion of this proposed service.  Future
revisions of this draft will become the standards document after the
formation of an appropriate working group.


1. INTRODUCTION

The Internet Kermit Service is intended to:

 1. Provide direct access to Kermit file transfer and management
    services without requiring the user to first login to a shell
    account;

 2. Provide Kermit file transfer and management services to anonymous
    users;

 3. Provide services to all Telnet clients that support Kermit file
    transfer protocol via a simple, predictable, scriptable, and
    well-documented textual interface;

 4. Provide direct and tightly-coupled access to a Kermit server
    when requested via the Telnet Kermit Option [TKO].

This draft assumes knowledge of Transmission Control Protocol, the
Telnet Protocol [TEL], the Kermit File Transfer Protocol [KER,PRF],
Telnet Kermit Option [TKO], and the commands and features of Kermit
software [CKB,CMG,K95].

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [BCP].

Definitions:

Kermit server
  A software program that is ready to accept and act upon commands in
  the form of well-defined Kermit packets [KER].

Kermit client
  A software program that receives requests through its user interface
  from a human user (or a script or other source) and translates them
  to command packets, which it sends to a Kermit server, thus
  initiating a Kermit protocol transaction such as the transfer of one
  or more files.


2. BACKGROUND

2.1. History

"Kermit" is the name of an extensible platform- and medium-independent
file transfer and management protocol [KER,PRF] and of a suite of
communications software programs that implement it and integrate it with
other communications functions [CMG,CKB,K95].

The Kermit protocol was first developed at Columbia University in New
York City in 1981 for transferring files without errors between diverse
types of computers over potentially hostile communication links.  Since
1981, the Kermit Project at Columbia University has expanded the
protocol, developed communications software that implements it upon key
platforms, and worked with volunteer programmers at other sites adapting
Kermit protocol to other platforms or communication methods.  The Kermit
Project also serves as the central point of Kermit software development,
support, information, and distribution throughout the world.

Kermit software is now available for nearly every computer and operating
system in existence.  The major features of the most popular Kermit
programs are:

 . Connection establishment and maintenance for a variety of connection
   methods including direct serial, dialup, TCP/IP, X.25, DECnet, and
   NETBIOS.

 . Terminal emulation.

 . Error-free transfer of both text and binary files, individually or
   in groups.

 . Character-set translation during both terminal emulation and
   text-mode file transfer -- a unique feature of Kermit software.

 . Remote file management through the client/server protocol.

 . A powerful and portable scripting language allowing complete
   automation of any task that can be performed manually.

Kermit's command and script language is consistent across all platforms
and communication methods, thus offering a unified method for
accomplishing a wide range of communication tasks manually or under
script control.

A single Kermit program combines the functions of many different
programs such as uucp, cu, tip, telnet, rlogin, ftp, iconv, and expect:
it is a Telnet and Rlogin client that can also transfer files; it is a
file transfer program that can also convert character sets; it is a
dialout program that can use dialing directories and understands country
codes and area codes; it is fully scriptable; it offers both
client/server and interactive modes of operation.  In its desktop
versions (particularly for DOS, Windows, and OS/2) it offers all the
features of communications software that are usually lacking from
Internet client software (key mapping, colors, scrollback, mouse
functions, printer control, etc)

Kermit software is widely used throughout the academic, government, and
corporate spheres, both in the USA and internationally.

In addition to the Kermit software developed and/or distributed by the
Kermit Project at Columbia University, hundreds of other software
products -- commercial, shareware, and freeware -- also include some
level of support for the Kermit protocol.  Thus there are hundreds,
perhaps thousands, of independent and interoperable Kermit protocol
implementations based upon the open Kermit protocol specification [KER].

The Internet has formed the primary mechanism by which users and
developers of Kermit software have collaborated to produce feature and
command sets that continually evolve to meet their needs as technology
changes.

2.2. Motivation.

Kermit protocol and software makes connections from one computer to
another and transfers data between them.  Countless people "live" in
Kermit all day long; as a customizable Telnet or Rlogin (or serial
communication) client with a wide selection of terminal emulations and
convenience features, it is their window onto the Internet.

Others use it in more creative ways, including some that involve key
parts of the Internet, e.g. in batch or cron jobs that update news or
Web servers or fetch email, or to monitor routers, terminal servers, and
hubs and dial pagers when faults are detected.  It is used by vendors of
telecommunications equipment for remote diagnosis, patching, and
updates.  Telecom managers often use Kermit scripts to configure PBXs,
muxes, routers, or terminal servers.  In the world of commerce, Kermit
is widely used for financial transactions, EDI, medical claim
submission, and so forth.  It is used with mobile barcode readers in
warehousing and inventory applications.  It is found in US Postal
Service sorting and scanning equipment.  It connects many of the
logistics and supply systems throughout the military.  It is found in
fast-food restaurant cash registers, milling and die-cutting machines,
textile looms and cutters, printing presses, and medical diagnostic
equipment.  It was the communications backbone of the 1994 Brazilian
national election -- the largest in history.

And yet there has never been a strong, explicit connection of Kermit
with the Internet.  In the early years, Kermit acted as a kind of
do-it-yourself network, enabling ordinary users to make connections that
were not already there, and for some years was the predominant method of
connecting a personal computer to the ARPAnet (e.g. by dialing a TAC).

Nowadays, however, with so many of the world's computers on the
Internet, the role of Kermit software and protocol is changing.  Kermit
users on the network would like to have the features, functions, and
interface they are accustomed to -- especially the automation features
-- available for use in settings where presently only tools like FTP are
available -- and even more so in situations where standard software like
FTP can't be used.

An Internet Kermit Service can fill this role, and augment the data
transfer power and flexibity of other Internet applications such as Web
browsers:

 . Like FTP, Kermit provides a service that can be accessed from many
   different platforms with a consistent set of commands, but unlike
   FTP, these commands include programming constructions such as
   variables, arrays, looping and selection mechanisms, and local and
   remote procedure calls.

 . Like FTP, Kermit provides both text- and binary-mode data transfer,
   as well as file management capabilities.  But Kermit also offers
   numerous features lacking from FTP, such as character-set
   translation, flexible file selection mechanisms, attribute
   preservation, and so on (see Section 5.3 for a longer list).

 . Unlike standard FTP, Kermit can transfer data through multiple
   firewalls, proxies, and network address translators (NATs) on a
   single port.

 . Unlike FTP, Kermit can transfer data across a combination of
   transports (e.g. dial-up to a terminal server and thence to an
   Internet host).

 . Authentication and data transfer can take place over secure
   connections (mutually authenticated and encrypted) using established
   Telnet authentication and encryption options.

 . Unlike traditional Kermit use over Telnet, anonymous access is
   possible, and the considerable overhead of the intervening Telnet
   server and pseudoterminal service is eliminated.

Until now the primary obstacles to an Internet Kermit Service have been:

 . Issues of authentication, privacy, and anonymous access.  These have
   been addressed in our implementation, as described Section 4 of this
   document.

 . Issues of coordination and control.  A Kermit software program can be
   in any of several "modes": at its command prompt or menu, awaiting
   commands from the user; in terminal mode, in which the user's
   keystrokes are sent to the remote computer or service; or in protocol
   mode, in which two Kermit programs communicate via well-defined
   Kermit packets [KER].  Commands or operations valid in one mode do
   not necessarily work in another.  Until now, it has been the user's
   responsibility to switch modes at one or both ends of the connection
   as needed.  A companion draft [TKO] to this one specifies a mechanism
   to closely couple the client and server via Telnet protocol
   negotiations, allowing each to know the other's state and to switch
   to the appropriate mode automatically so a valid and useful
   relationship obtains at all times.

 . Lack of a standard TCP port.  The "registered" port 1649 was assigned
   by IANA for this purpose (27 September 1995) and is named "Kermit".
   (renamed from "Inspect".)


3. THE INTERNET KERMIT SERVICE MODEL

The Internet Kermit Service (IKS) uses a standard Telnet [TEL]
connection, in which all Telnet rules apply.  Unlike FTP, which requires
additional TCP connections, IKS uses a single channel for both signaling
and data transfer.  The connection is multiplexed via (a) Telnet
options, and (b) Kermit protocol messages.  This allows existing Telnet
clients that also support the Kermit protocol, whether or not they
support the Telnet Kermit Option [TKO], to use the IKS and take
advantage of all relevant Telnet options including authentication and
encryption.

The system Internet services daemon (e.g. inetd) waits for a connection
on the Kermit socket (1649) and then starts the IKS on the new
connection.  The IKS performs the familiar Telnet negotiations including
the Telnet Kermit option.  Unlike a standard Telnet server, the IKS does
not support the ability to present the user with an interactive system
shell.  The Kermit socket is used only for file transfer and management
functions provided by Kermit file transfer protocol and the Kermit
script language.

Once the connection is established, the Telnet Kermit Option is
negotiated in both directions.  The results determine which of the
following configurations is used by the Telnet client and Server:

 . Server-side Kermit Server (SKS)
 . Client-side Kermit Server (CKS)
 . No Kermit Server (NKS)

Different procedures and functions apply to each configuration.  The
configuration may be changed at any time by Telnet Kermit Option
subnegotiations, which assure that the Telnet client and server are
always in compatible states.

The three configurations are described in the following sections.

3.1. Server-Side Kermit Server

In the Server-Side Kermit Server (SKS) configuration, the Telnet server
is the Kermit server and the Telnet client is the Kermit client.  This
configuration is used when both Telnet client and IKS support the Telnet
Kermit Option and the IKS sends WILL KERMIT to the Telnet client and
receives DO KERMIT from the Telnet client [TKO].

In this case, the IKS immediately starts a Kermit server and reports
this to the Telnet client with a Telnet KERMIT START-SERVER
subnegotiation.

The SKS configuration is appropriate when the user wishes to interact
only with the Telnet client's commands or menus.

If authentication was not performed with one of the Telnet
Authentication Option protocols, the Kermit server rejects all Kermit
protocol operations (except REMOTE LOGIN, REMOTE HELP, REMOTE EXIT, BYE,
or FINISH -- that is, the ones that request help, that log in, that
close the connection, or that change the status of the connection)
until:

 . A Kermit REMOTE LOGIN command successfully authenticates the user;

 . The login retry limit is reached;

 . A Kermit BYE or REMOTE EXIT command is received, which closes
   the connection;

 . A Kermit FINISH command or a Telnet KERMIT REQ-KERMIT-STOP
   subnegotiation is received to request the IKS exit from Kermit
   server mode.  At this point, the IKS can either exit and close
   the connection or issue an interactive login prompt, depending
   on how it was started or configured by the system administrator.

Once the user is authenticated:

 . The Telnet client configures itself for Kermit client/server
   operation, with itself as the Kermit client, communicating with the
   server only by Kermit packets, and optionally adjusting its menus or
   commands to eliminate functions (such as terminal emulation) that
   make no sense in this context.

 . The relationship persists until the Telnet client and IKS agree to
   terminate the Kermit server via Kermit protocol commands (BYE,
   FINISH, or REMOTE EXIT), or by Telnet Kermit Option subnegotiation,
   or by closing the connection.

3.2. Client-Side Kermit Server

In the Client-Side Kermit Server (CKS) configuration, the Telnet server
is the Kermit client, and the Telnet client is the Kermit server.  This
configuration is used when the IKS has sent WONT KERMIT or SB KERMIT
STOP-SERVER, and the Telnet Client has sent WILL KERMIT and SB KERMIT
START-SERVER, indicating that it is prepared to accept and process
Kermit protocol packets.

In the CKS configuration, the Telnet client assumes the role of Kermit
server by virtue of its ability to recognize and process Kermit protocol
packets in its terminal emulator.  Thus the Telnet client must not send
WILL KERMIT or the KERMIT START-SERVER subnegotiation unless its
terminal emulator is capable of recognizing Kermit packets.

If the IKS is at top command level (as opposed to executing a script),
or when it reaches top level after finishing a script, it issues its
interactive command prompt.

At the point, the user may type commands or send scripted commands to
the IKS command prompt.  When a data-transfer command (such as SEND) is
issued by the user at the IKS prompt, a Kermit packet is transmitted and
recognized by the Telnet client, causing it to automatically perform the
requested action (e.g. receive a file), and then resume its previous
mode (terminal emulation or script execution) when the data transfer is
complete.

Thus, in the CKS configuration, data transfers are initiated by the IKS
rather than by the Telnet client.  This configuration is useful when the
user prefers the command interface or repertoire of the server to that
of the client.

If the IKS sends a Telnet KERMIT START-SERVER subnegotiation, the
relationship switches automatically to Server-Side Kermit Server
(Section 3.1), in which the Telnet client is the Kermit client and the
Telnet server is the Kermit server.

If the Telnet client sends a KERMIT STOP-SERVER subnegotiation, the
connection switches to No Kermit Server (Section 3.3) and the IKS issues
its command prompt.  At this point, neither side is a Kermit server, and
both sides may optionally disable Kermit protocol commands.  Subsequent
user action can designate one side or the other as the Kermit server, as
desired.

3.3. No Kermit Server

If both Telnet client and IKS send WONT KERMIT or SB KERMIT STOP-SERVER,
or if the Kermit client and server are connected across multiple hosts
or transports, thus precluding end-to-end Telnet negotiation, a Kermit
server is not known to be available.  In the KERMIT STOP-SERVER case,
the Kermit partners can later switch back to SKS or CKS, but in the
other two cases, there is no such signaling and loose coupling
characterizes the entire session.

In the No Kermit Server (NKS) configuration, the IKS presents a command
prompt to the Telnet client.  As in the Client-Side Kermit Server
configuration, plain-text commands are issued to the IKS.

In the loosely coupled NKS configuration, the Telnet client does not
know the state of the Telnet server, and so can not automatically adjust
its commands and menus to present only valid choices, or automatically
change its state to complement the server's; it is the user's
responsibility to assure that the "mode" (command prompt, terminal
emulation, server command wait) of each Kermit partner is appropriate
for each action.  Thus an Internet Kermit Server appears as an ordinary
remote Kermit program to any Telnet client that does not implement the
Telnet Kermit Option, or in which this feature is disabled or can not be
used.

The NKS configuration allows successful manual operation of the IKS
through Telnet clients that do not support the Telnet Kermit Option.
The Telnet client might or might not support Kermit "autodownload" and
"autoupload"; if it does not, then the user is forced to manually issue
command on both sides of the connection in the traditional and familiar
manner [CKB,CMG,K95].

4. SECURITY CONSIDERATIONS

4.1. AUTHENTICATION

Authentication is provided via one or more of the following methods:

 . The Telnet AUTHENTICATION option;

 . The Telnet START_TLS option;

 . Plaintext userid/password verification.

4.1.1. Telnet Authentication option

The use of one of the many Telnet authentication option methods removes
the need to transmit passwords in plaintext across public networks.  In
addition, the exchange of user authentication information often provides
a shared secret that can be used with the Telnet Encryption Option
protocols to encrypt the connection in one or both directions.

Telnet authentication may also be used in conjunction with the Telnet
START_TLS option to negotiate end user identity over the encrypted and
host authenticated TLS channel.

The IKS currently supports Kerberos 4, Kerberos 5, Secure Remote Password
and Microsoft NTLM authentication methods via the Telnet AUTH option.

4.1.2. Telnet over TLS option

The Telnet START_TLS option provides for the negotiation and establishment
of a TLS version 1 session after the initial telnet connection.  The TLS
connection provides host to client authentication via the use and X.509
certificate chains.  TLS also supports optional client to host
authentication using host verified X.509 certificates which may be used
to authenticate a userid provided by the client or be mapped to a userid
based upon properties of the certificate.

4.1.3. Plaintext Authentication via Kermit REMOTE LOGIN

In the Server-Side Kermit Server configuration, if the client is not yet
authenticated, the client must log in using a REMOTE LOGIN command, in
which a Kermit packet containing user ID and password in clear text is
sent from the Telnet client to the Telnet server, which then calls upon
local mechanisms to authenticate the user.  Any packets other than login
(or REMOTE HELP, REMOTE EXIT, FINISH, or BYE) packets are rejected
(returned with an error message) until the user is authenticated.  If
the number of unsuccessful login attempts exceeds the limit, the
connection is closed.  Many Kermit client programs support this login
method already.

This method should be avoided whenever possible.  If plaintext passwords
are used, they should only be sent after the Telnet START-TLS option has
been negotiated. (see 4.2.2)  Otherwise, passwords are open to packet
sniffing.

4.1.4. Plaintext Authentication via Command Prompt

In the Client-Side Kermit Server and No Kermit Server configurations,
the server presents the user with a plain-text interactive interface
that begins with the server issuing "Username:" and "Password:" prompts,
just as if the user were logging in to a multiuser timesharing system
such as VMS or UNIX.  When a password is not required an empty response
can be given.  Invalid username-password combinations result in a new
series of prompts up to the login retry limit, and then disconnection.

This method should be avoided whenever possible.  If plaintext passwords
are used, they should only be sent after the Telnet START-TLS option has
been negotiated. (see 4.2.2)  Otherwise, passwords are open to packet
sniffing.

4.1.5. Anonymous Login

When the username is "anonymous" or "ftp", the IKS behaves like an
anonymous ftp server, in a manner appropriate to the underlying
platform.  In UNIX, for example, access is restricted to a designated
area of the file system.  A password might or might not be required,
according to the preference of the site administrator.

If privacy is desired the Telnet START-TLS option should be used.  (see
4.2.2)

4.2.  ENCRYPTION (PRIVACY)

As the Internet becomes ever more public and susceptible to
eavesdropping, it becomes increasingly necessary to provide methods for
private access to services.  Telnet provides two such mechansims:

 . Telnet Encryption option
 . Telnet START-TLS option

4.2.1.  Telnet Encryption option

The Telnet Encryption option, although it has never achieved RFC status,
has been used for years in conjunction with the Telnet Auth option in
Telnet clients and servers that support Kerberos 4, Kerberos 5, Secure
Remote Password, and others.  The IKS currently supports the following
encryption methods under the Telnet Encryption option:

 .  cast128_ofb64
 .  cast5_40_ofb64
 .  des_ofb64
 .  cast128_cfb64
 .  cast5_40_cfb64
 .  des_cfb64

4.2.2. Telnet over TLS option

Transport Layer Security (TLS), the successor to Secure Sockets Layer (SSL),
provides methods to implement Server authentication, Client authentication,
and Transport Layer encryption.  Unlike Telnet Encryption, Start-TLS does
require the use of Telnet Authentication in order to provide a private
channel.  This means that it can be used in conjunction with plaintext
passwords and anonymous connections.

5. SERVICES

The Internet Kermit Service includes features for both users and system
administrators.  The IKS is incorporated into  the 7.0 release of
Columbia University's C-Kermit software, which is the "master" Kermit
software program in terms of features and command language.  An overview
of C-Kermit can be found at:

  http://www.columbia.edu/kermit/ckermit.html
  http://www.kermit-project.org/ckermit.html

When C-Kermit is employed as an Internet Kermit Service, it may offer
all its functions to "real" users (those who are authenticated as
specific users), and a safe subset of its functions to anonymous users.

The Internet Kermit Service resembles an FTP server in that it performs
its own authentication and uses a well-defined protocol to communicate
with its client, but differs from the FTP server by also offering (at
the system manager's discretion) an interactive user interface to the
Telnet client when it is in terminal mode.  It also differs from FTP in
restricting all protocol messages and data transfer to a single socket
connection.

An IKS has been deployed at Columbia University for worldwide public access
to the Kermit FTP site:

  telnet://kermit.columbia.edu:1649/
  telnet://ftp.kermit-project.org:1649/

5.1. Features for System Administrators

The system administrator can supply IKS configuration parameters as
command-line options or in a configuration file, or both in combination.
Such parameters include:

 . Whether anonymous logins are allowed.

 . The file system or root directory to which anonymous users are
   restricted.

 . Specification of permissions and other attributes to be assigned to
   files uploaded by anonymous users.

 . Whether to make session entries in system logs.

 . Specific services to disable: reception of files, sending of files,
   sending of email, printing, changing of directories, getting
   directory listings, deleting files, etc (see next section).

 . Whether access to the interactive command prompt is allowed.

5.2. Features for Users

The IKS supports a wide range of services, including, but not limited
to, the following:

 . Authentication as a real user or anonymously.

 . Transmission of files to which read access is allowed.

 . Reception of files into directories or devices to which write access
   is allowed.

 . The ability to display a file on the client's screen.

 . Ability to list files.

 . Ability to change its working (default) directory.

 . Ability to delete files to which write or delete access is allowed.

 . Ability to rename and copy files

 . Ability to create and remove directories.

 . The ability to route received files to a specified printer, or to
   send them as email to a specified address list.

 . Client control of server parameter settings, within limits
   established by the server system administrator.

 . Transmission of variables from client to server or vice versa.

 . Remote and local script execution.

 . Remote and local procedure execution.

File transfer features include:

 . Kermit text-mode transfers incorporate not only record-format
   conversion, but also character-set translation;

 . Kermit can switch automatically between text and binary mode on a
   per-file basis when sending groups of files by matching each file's
   name with a pattern list.

 . A selection of file collision options, including "make backup copy
   of existing file and accept incoming file", "reject incoming file",
   "accept incoming file only if newer than existing file", etc.

 . Numerous methods for selecting the files to be transferred, including
   pattern matching, lists of filenames (or patterns), exception lists,
   date and/or size ranges, etc.

 . Filename conversion and file renaming.

 . Automatic directory creation if elected and enabled.

 . Standard mechanisms for directory traversal, allowing transmission of
   entire directory trees or other file hierarchies even between unlike
   file systems such as VMS, UNIX, and Windows.

 . Atomic file movement: optionally, the source file can be deleted
   (or renamed, or moved) when and only when it has been transferred
   successfully.

 . Kermit can retain file attributes including time stamps and
   permissions (at the user's or system administrator's discretion),
   even between unlike platforms;

 . Recovery of interrupted transfers from the point of failure.

 . File-transfer pipes and filters.

Script programming features include:

 . Macros with parameter substitution.

 . Built-in and user-defined variables and arrays, with global
   or local scope.

 . Built-in and user-defined functions.  Built-in functions include:
    - String functions
    - Arithmetic functions
    - Date / time functions
    - File functions

 . Input search for multiple simultaneous targets.

 . IF-ELSE, WHILE, FOR, SWITCH, GOTO, C-like block structure.

 . Every command returns a completion status that may be tested
   and used as a basis for subsequent actions.

5.3. User Interface

The Internet Kermit Service uses the Kermit command and script language,
as implemented in Columbia University's C-Kermit communication software
[CKB].  This program and its command language are portable to all known
varieties of UNIX, as well as to Windows 95/98/NT, OS/2, Digital
(Open)VMS, Stratus VOS, Data General AOS/VS, Plan 9, OS-9, QNX, the
Commodore Amiga, and other platforms.  The C-Kermit command language is
a superset of that of other Kermit software programs including MS-DOS
Kermit for DOS and Windows 3.x, IBM Mainframe Kermit for VM/CMS,
MVS/TSO, CICS, and MUSIC, PDP-11 Kermit for RT-11, RSTS/E, RSX-11, and
IAS, and dozens of other Kermit programs.

It is far beyond the scope of this document to enumerate, let alone
describe, the commands and services of C-Kermit; this is the subject of
a 600-page book [CKB], augmented by hundreds of pages of online
material.  A brief overview is included here.

Commands are based on English words.  There is no plan at present to
support other natural languages (Italian, Portuguese, Norwegian,
Russian, Hebrew, Japanese, Cherokee, etc) as alternative bases for
command words, since this would reduce the portability of scripts.
However, since the command language includes a macro capability, macros
may be defined to provide selected commands in different languages if
desired.

Certain commands can apply either locally or remotely, for example "CD"
(Change Directory).  The convention is to prefix the command with the
word REMOTE if it is to apply remotely.  Example: "cd foo" changes to
the "foo" directory on the computer where the command was given; "remote
cd foo" sends a Kermit packet to the Kermit server requesting it to
change its directory to "foo".  The commands in this category include:

  ASSIGN <variable> <value>      Assign a value to a variable.
  CD <directory>                 Change working directory.
  COPY <files> <destination>     Copy file(s)
  DELETE <files>                 Delete file(s)
  DIRECTORY [ <pattern> ]        List file(s)
  EXIT                           Exit
  HELP [ <topic-or-command> ]    Display help text
  MKDIR <directory>              Create a directory
  PRINT <files>                  Print file(s)
  PWD                            Print working directory
  RENAME <old> <new>             Rename file(s)
  RMDIR <directory>              Remove a directory
  SET <parameter> <value>        Change a parameter's value
  TYPE <file>                    Display the contents of a file

As a convenience, REMOTE commands also have short synonyms: RASSIGN,
RCD, RCOPY, RDELETE, and so forth.

The basic file transfer commands are:

  SEND [ modifiers ] <files>    Send file(s) (to server)
  GET [ modifiers ] <files>     Get file(s) (from server)

These commands take a file name, pattern, or list, plus various optional
modifiers, including transfer mode specifiers (text, binary), file
selectors (date, size, exception list), aliasing, name and path options,
disposition specifiers, and so on.

In addition to the commands listed above, the following commands are
sent by the client to the server:

  REMOTE QUERY                   Get value of variable or procedure
  BYE                            Log out and close the connection
  FINISH                         Request the server leave server mode

Like all Kermit client/server commands, these can be disabled if
desired.

Of course there are numerous other commands with purely local effect,
such as the many scripting commands.  These, plus all the commands
above, are fully documented in [CKB].  The repertoire grows over time,
but never in a way that invalidates existing scripts.

The system administrator can allow or forbid access to any of these
features, and to the command language as a whole.  In the latter case,
the IKS may be accessed only as a Kermit server, by giving commands to
the client.


6. REFERENCES

[TKO] Altman, Jeffrey E., and Frank da Cruz,
      Telnet Kermit Option,
      Internet Draft <draft-altman-telnet-kermit-server-00.txt>,
      August 1998.

[BCP] Bradner, Scott, RFC 2119, "Best Current Practice", March 1997.

[KER] da Cruz, Frank, "Kermit, A File Transfer Protocol", Digital Press/
      Butterworth Heinemann, Newton, MA (1987).  379 pages,
      ISBN 0-932376-88-6.

[CKB] da Cruz, Frank, and Christine M. Gianone, "Using C-Kermit", Second
      Edition, Digital Press / Butterworth-Heinemann, Woburn, MA (1997).
      622 pages, ISBN 1-55558-164-1.

[CMG] Gianone, Christine M., "Using MS-DOS Kermit", Second Edition,
      Digital Press / Butterworth-Heinemann, Woburn, MA (1992). 345
      pages, ISBN 1-55558-082-3.

[K95] Gianone, Christine M., and Frank da Cruz, "Kermit 95", Manning
      Publications, Greenwich CT, (1996). 88 pages, ISBN 1-884777-14-7.

[PRF] Huggins, James K., "Kermit Protocol - Formal Specification and
      Verification", in Boerger, E., "Specification and Validation
      Methods", Oxford University Press (1995).  ISBN 0-19-853854-5.

[FTP] Postel, J., and J. Reynolds, "File Transfer Protocol (FTP)",
      RFC959, October 1985.

[TEL] Postel, J., and J. Reynolds, "Telnet Protocol Specification",
      RFC854, May 1983, et seq.; "Telnet Option Specification",
      RFC855, May 1983, et seq.

[IAN] Internet Assigned Numbers Authority:
      http://www.iana.org/numbers.html
      http://www.iana.org/assignment/port-numbers


7. AUTHORS' ADDRESS

Frank da Cruz
fdc@columbia.edu

Jeffrey E. Altman
jaltman@columbia.edu

The Kermit Project
Columbia University
612 West 115th Street
New York NY 10025-7799
USA
http://www.columbia.edu/kermit/
http://www.kermit-project.org/

(end)


    Jeffrey Altman * Sr.Software Designer * Kermit-95 for Win32 and OS/2
                 The Kermit Project * Columbia University
              612 West 115th St #716 * New York, NY * 10025
  http://www.kermit-project.org/k95.html * kermit-support@kermit-project.org