COLUMBIA UNIVERSITY CENTER FOR COMPUTING ACTIVITIES
NEW YORK, NY 10025 (212) 854-8543 or -3508
NEWS RELEASE JULY 1, 1989 FOR USE UPON RECEIPT
THE FIRST INTERNATIONAL KERMIT CONFERENCE, MOSCOW, USSR, MAY 29-31, 1989
"Kermit" is a protocol developed at Columbia University that has become a
worldwide de-facto standard for transferring files between computers. A brief
history and description of the Kermit file transfer protocol is given in
Appendix A.
Columbia University has always made a point of sharing its Kermit software.
Several years ago, Kermit programs found their way into the Soviet Union via
the European Space Agency. In March 1986, Kermit was used to transfer data
gathered by the Soviet VEGA spacecraft to the European Space Operations Center
in West Germany in a cooperative East-West effort to study Halley's comet at
close range. Soon thereafter, Soviet computer specialists began to develop
Kermit programs for their own computers.
As the popularity of Kermit grew in the USSR, Soviet scientists made contact
with Columbia's Kermit Development and Distribution Center, managed by
Christine M. Gianone, for clarifications about the protocol and updates to
the software. In the discussions which followed, it became clear that there
was great interest in Kermit not only in the Soviet Union, but also in the
other Eastern Bloc countries, and the idea of an international conference
was born.
The First International Kermit Conference was sponsored by the International
Center for Scientific and Technical Information (ICSTI), Moscow, through the
efforts of Dr. Juri Gornostaev, Head of ICSTI's Computer Department, and
Professor A. Butrimenko, Director of ICSTI. ICSTI is an information resource
center for the USSR and ten other Easter Bloc countries. It was the first
international gathering devoted entirely to Kermit.
The conference was attended by more than 70 computer specialists from Eastern
Bloc countries including Hungary, East Germany, Poland, Bulgaria, Mongolia,
Czechoslovakia, Cuba, and many parts of the USSR. The conference was
conducted by:
Christine Gianone
Manager, Kermit Development and Distribution, Columbia University
Frank da Cruz
Manager, Network Planning, Columbia University
Gianone and da Cruz lectured in English, which was translated into Russian.
There were lively question-and-answer sessions after each lecture, followed by
practice in the computer laboratory. Hundreds of pages of lecture notes were
printed by ICSTI and distributed to the conference participants.
On May 29, Gianone and da Cruz discussed the history and philosophy of Kermit,
and gave a detailed introduction to the use of Kermit programs, including
advanced features such as the Kermit script programming language, which allows
users to create automated, unattended file transfer sessions. The attendees
proved very familiar with Kermit, some at the user level, others at the
implementation and theory level. The discussions which followed showed that
Kermit plays a key role in data communications in the Soviet Union and Eastern
Bloc countries.
On May 30, Gianone and da Cruz explained the Kermit prococol itself,
concentrating on the design decisions which have made it adaptable to nearly
every kind of computer, file system, or communication environment, and on
Kermit's extensibility mechanisms which allow new features to be added to the
protocol without introducing incompatibilities with older Kermit programs.
Detailed analysis of Kermit's performance options -- data compression, long
packets, and sliding windows -- gave the audience tools for evaluating and
predicting the performance of Kermit programs in various communication
environments.
On May 31, Gianone presented a Kermit protocol extension for international
character sets. This is the first major extension to the Kermit protocol
since 1985, and it is designed to address the increasing internationalization
of inter-computer communications. The mechanisms of this extension, which are
summarized in Appendix B, will allow text written in languages other than
English to be transferred between unlike computers and still retain its
original appearance. Gianone's presentation was followed by a series of
presentations from other conference attendees (See Appendix C for complete
program).
Gianone and da Cruz also visited other computing installations in Moscow,
including:
. VNIIPAS, the All-Union Institute of Automated Systems of the National
Centre for Automated Data Exchange. Programmers here have developed an
integrated software system that includes electronic mail, conferencing, data
search & retrieval, etc, with Kermit as its communication mechanism.
. INION, the Institute of Scientific Information for Social Science, where
Kermit is used extensively.
RESULTS OF THE CONFERENCE
The international character set extension was discussed. It emerged that at
least five different codes are in use for representing Cyrillic characters on
Soviet computers: KOI-7, KOI-8, DKOI, EBCDIC, and ECMA-113. Without the
extension to Kermit, each Soviet Kermit program would have to know its own
codes plus all the others in order to transfer Cyrillic text files. Multiply
this situation by all the world's languages, and all the different ways that
computers have of representing them, to appreciate the difficulties
encountered when trying to transfer text in any language between any two
computers.
The Kermit extension reduces this problem to manageable size by specifying a
small set of standard codes and the means by which one Kermit program can
identify them to the other. After some discussion, the basic ideas of the
international character set extension were endorsed by the conference.
Gianone presented ICSTI with a complete, up-to-date set of Kermit programs and
documentation on magnetic tape and diskette. By agreement, ICSTI in Moscow
will join other regional Kermit Distribution centers that have been set up by
Gianone on previous trips to Japan, France, and Switzerland, as well as those
in England, the Netherlands, Australia, and other parts of the world that
receive Kermit materials from Columbia by network or by post.
ICSTI's Kermit Distribution will serve the Soviet Union and other Eastern-Bloc
countries, and ICSTI will become a center for coordinating Kermit program
development in those countries. The link between Columbia University and
ICSTI considerably augments the worldwide pool of Kermit programming
expertise, and for the first time the Eastern-Block countries will have
convenient access to the vast body of current Kermit program source and
documentation. ICSTI has already published Russian-language translations of
Gianone's Kermit manuals for several popular computers, as well as for their
own home-grown versions.
ICSTI presented Columbia with a version of Kermit it had developed for IBM's
CICS MVS subsystem, which is already being studied by Western computer experts
for inclusion in the broader IBM mainframe Kermit family of programs which,
until now, did not support IBM's CICS subsystem.
Perhaps the greatest benefit of the conference was the face-to-face meeting of
people from from different worlds: East meets West. The problems of data
communications are remarkably similar in all parts of the world, and the
computer specialists from the widely scattered countries represented at the
conference shared a large body of common experience.
APPENDIX A: THE KERMIT FILE TRANSFER PROTOCOL
The world is populated with many millions of computers -- PCs, minicomputers,
mainframe computers, supercomputers -- used for academic, research, business,
medical, manufacturing, and many other purposes, and increasingly also in the
home.
The need for these computers to communicate with each other -- to share
information -- has become increasingly commonplace. Yet the means to do this
is not always readily apparent. The manufacturers may offer:
. Magnetic media, such as disks or tapes. But these are usually incompatible
between computers of different manufacture. Even when possible, data
transfer by this method can be impractical if the computers are separated
by long distances or national boundaries.
. Computer networks. But these are usually expensive, often
vendor-specific, and always difficult to install and manage. And usually
impractical if long distances or national boundaries are involved.
But there is one communication mechanism that almost all computers have in
common: the asynchronous serial communication port that is commonly used with
a modem. This is standard equipment on almost every computer.
By connecting their serial ports, two computers can transfer data. If the
computers are close by, then a simple cable provides the necessary connection.
If they are far apart, then modems may be used to make the connection through
the telephone system. Potentially any two computers in the world can transfer
data.
But data transfer is not necessarily information transfer. Before information
can be successfully transferred, several problems must be overcome:
1. Incompatibility of data format. Different computers may use different
codes to represent characters, and may use different formats for
representing lines of text.
2. Synchronization. If one computer sends data faster than the other
computer can process it, data will be lost.
3. Errors. If there is noise, static, electrical interference on the
connection, data can be corrupted during transfer.
To overcome these problems, we designed the Kermit File Transfer Protocol in
1981. (The Kermit protocol was named after Kermit the Frog, star of the
television series, THE MUPPET SHOW, and is used by permission of Henson
Associates, Inc., New York City.)
Kermit is a universal file transfer protocol, in that it can work on any kind
of computer at all. In order to have Kermit file transfer between two
computers, a Kermit program must be written for each computer. The computer
user establishes a physical connection between the two computers, activates
the Kermit program on each computer, and then the two Kermit programs transfer
the data in special messages called "packets".
The format of the packets is standard, so that any two Kermit programs can
communicate with each other. Therefore, the Kermit program for each computer
only has to know about its own local conventions for data representations and
the Kermit standard. It does not need to know anything about any other kind
of computer.
Kermit packets, and the rules for formatting and transmitting them, allow any
two computers to transfer information correctly and completely. The exchange
of packets provides the necessary synchronization. The packets include an
error-checking field, or "checksum", which allows the receiver to determine if
a packet has been damaged in transmission and if so to request retransmission.
And textual data within a packet is encoded and formatted in a standard way,
so that all Kermit programs know how to convert between Kermit's "transfer
syntax" and their own local data storage conventions.
The Kermit file transfer protocol was designed at Columbia University in 1981
to allow reliable data transfer among Columbia's computers: CP/M
microcomputers, the IBM PC, DECSYSTEM-20 timesharing systems, and IBM
System/370 series mainframes. The software, documentation, and the protocol
specification were freely shared with other universities, as well as with
corporations and government agencies, in the USA and other countries. Before
long versions of Kermit appeared for other systems too: the DEC PDP-11 and
VAX, UNIX, the Apple II and Macintosh, Commodore micromputers, CDC and Cray
supercomputers, etc. These new versions were contributed to Columbia
University in the same spirit in which Columbia shared its original versions.
Kermit has since become one of the most popular and widely implemented
software programs in existence, owing in large part to Columbia's philosophy
of sharing. Program source code is available so that users may make fixes or
improvements, adapt existing programs to new computers, or ensure that there
are no hidden viruses. The protocol specification is open and public. The
software is free, and it is not copy-protocted. In fact, copying and sharing
of Kermit programs is encouraged. Kermit programs, however, do bear a
copyright notice in order to protect the authors and users from entrepreneurs
who might try to turn Kermit into a commercial product.
After a few years of uncoordinated Kermit development, Columbia University
established a Center for Kermit Development and Distribution, managed by
Christine M. Gianone. Connected by computer network to universities and
research organizations throughout the world, this center coordinates the
development and distribution of Kermit programs. There now exist Kermit
programs for about 350 different computers and operating systems, from the
smallest PC to the largest supercomputer. These programs are obtained by
organizations in the corporate, government, and private sectors via wide-area
network connections to Columbia, or else by mail order. As of June 1989,
Columbia has sent Kermit programs by mail to about 12,000 sites, about half of
them corporate, in all states of the USA and more than 60 other countries.
Countless other sites have obtained Kermit over the networks. Since copying
and sharing is encouraged, any given shipment might serve hundreds or
thousands of people. For example, a single Kermit disk sent to a US
government agency is duplicated and furnished to about 10,000 medical research
scientists and other workers.
As the popularity of Kermit spread, Gianone and the staff of the Kermit
Development and Distribution Center have been invited to discuss Kermit at
computer gatherings in the US, Europe, the Far East, and now the Soviet Union.
Further information about Kermit may be obtained by writing to:
Kermit Distribution
Columbia University Center for Computing Activities
612 West 115th Street
New York, NY 10025
APPENDIX B, EXTENSION TO KERMIT PROTOCOL FOR INTERNATIONAL CHARACTER SETS
Kermit has always been able to transfer text files between any two computers,
no matter how differently they represent text internally. To do this, Kermit
transfers text in ASCII, with lines of text terminated by carriage return and
line feed (CRLF). This is Kermit's "transfer syntax" for text files:
_____________________________________________________________________________
KERMIT NORMAL TEXT TRANSFER SYNTAX (DEFAULT)
"SET FILE TYPE TEXT" "SET FILE TYPE TEXT"
SENDER RECEIVER
| ^
| |
Convert from local Convert from ASCII
character set to ASCII. to local character set.
Convert from local record format Convert from CRLF-terminated lines
to CRLF-terminated lines. to local record format.
| |
| file data |
+---------------------------------------------+
ASCII
_____________________________________________________________________________
Using this mechanism, two totally different kinds of computers can exchange
text files, for example a UNIX system with ASCII stream text files and an IBM
mainframe with EBCDIC record-oriented text files (ASCII and EBCDIC are two
totally different code systems for representing alphabetic, numeric,
punctuation, and other characters).
But ASCII only includes enough letters and symbols for English and the few
other languages that can be written using only the letters A-Z with no
diacritical marks (like Dutch, and German without umlauts).
As soon as you want to transfer files containing accented vowels, c-cedillas,
etc, not to mention files written in Greek, Arabic, Hebrew, Russian, Japanese,
or Chinese, you will find that Kermit can transfer them OK, but if the
receiving system is not the same as the originating one, the result will look
like garbage. This is because different computer vendors use different codes
for the same special characters. For example, an Apple Macintosh, an IBM PC,
and a DEC VAX use totally different codes for, say, A-umlaut, C-Cedilla,
E-grave, etc.
Since there are hundreds or thousads of computer vendors in the world and
nobody controls what codes they use to represent characters, it is clearly
impractical to require that every computer have knowledge of every other
computer's codes. Therefore, only a few standardized codes should be used
for file transfer. Then, each computer only has to know its own codes and
the standard ones, and does not need specific knowledge of any other kind of
computer.
The extension to Kermit has two levels, the first for files composed of any
single code other than ASCII, and the second for files made up of a mixture of
character codes.
LEVEL 1
Level 1 will allow the use of any single character code in the transfer
syntax. Such codes include ISO Latin Alphabet 1, which can be used to write
most Western European languages, or ISO Latin/Cyrillic, which can be used to
write Russian, Ukrainian, etc, or ISO Latin/Hebrew for Hebrew or Yiddish, or
JIS X 0208 for Japanese, or any other single character set. (ISO is the
International Organization for Standardization, JIS is the Japan Industrial
Standard).
_____________________________________________________________________________
EXTENSION LEVEL-1 TRANSFER SYNTAX - A SINGLE CHARACTER SET OTHER THAN ASCII
"SET FILE CHARACTER-SET XXX" "SET FILE CHARACTER-SET YYY"
"SET TRANSFER-SYNTAX LATIN1-ISO" "SET TRANSFER-SYNTAX LATIN1-ISO"
SENDER RECEIVER
| ^
| |
Convert from local character Convert from specified character
set to specified character set. set to local character set.
Convert from local record format Convert from CRLF-terminated lines
to CRLF-terminated lines to local record format.
| |
| file data |
+---------------------------------------------+
Selected Character Set
_____________________________________________________________________________
For example, a French-language document can be exchanged in this way between
an Apple Macintosh and an IBM PC, both of which have the necessary characters
for French, but use different codes to represent them. ISO Latin Alphabet 1
can be used as the transfer syntax.
Level 1 will accommodate most of today's computers, all over the world, which
use only a single character set.
LEVEL 2
Level 2 of the Kermit protocol extension will allow for transfer of files in a
mixture of alphabets, registered in the International Register of Coded
Character Sets, by embedding the alphabet designation and shift "escape
sequences" defined in ISO Standard 2022, and the alphabet designators listed
in the ISO International Register of Coded Character Sets, in the data stream:
_____________________________________________________________________________
EXTENSION LEVEL-2 ISO-2022 TRANSFER SYNTAX - MULTIPLE CHARACTER SETS
"SET TRANSFER-SYNTAX ISO-2022" "SET TRANSFER-SYNTAX ISO-2022"
SENDER RECEIVER
| ^
| |
Convert from local Convert from registered
character sets to standard character sets
registered standard sets. to local character sets.
Convert from local record format Convert from CRLF-terminated lines
to CRLF-terminated lines to local record format.
| |
| |
| File data, designators, and shifts |
+---------------------------------------------+
designated character sets
_____________________________________________________________________________
For example, a file containing Roman, Russian, and Greek characters can use
the ISO Latin/Cyrillic and ISO Latin/Greek alphabets. The initial alphabet is
designated by an escape sequence, and whenever there is a change of alphabet,
another escape sequence is used to identify the new alphabet:
<escape sequence for Latin/Cyrillic>characters in Latin/Cyrillic
<escape sequence for Latin/Greek>characters in Latin/Greek
<escape sequence for Latin/Cyrillic>characters in Latin/Cyrillic
The usefulness of Level 2 will become apparent in the future, as computer
manufacturers begin to bow to the needs of the international computer market
and start to employ mechanisms for storing text in multiple alphabets.
APPENDIX C, CONFERENCE AGENDA
May 29, 1989:
Christine Gianone and Frank da Cruz, Columbia University, New York,
"Introduction to Kermit"
May 30, 1989:
C. Gianone and F. da Cruz, "Kermit Protocol, Theory, and Operation"
May 31, 1989:
C. Gianone, "A Kermit Protocol Extension for International Character Sets"
A. Gruzdev, NPO Informatika, Ivanovo, USSR,
"Use of Kermit Protocol for Interconnection of Homogeneous PC and
Minicomputer Networks"
A. Smirnov and V. Rejma, EstNIINTI, Tallinn, Estonia, USSR,
"Adaptation of Kermit Programs for EC and SVM Mainframes and Iskra 1030 PCs"
A speaker from Moscow Ministry of Radio Industry:
"Use of Kermit in a PC Network for Secondary Schools".
Milan Motl and F. Zednik, Sigma Concern, Olomouc, Czechoslovakia,
"Use of Kermit Program Products in Heterogeneous Computer Networks"
I. Shmeikal, High Economic School, Prague, Czechoslovakia,
"Terminal Network and Experience in Implementation of Kermit Programs in the
DOS-4 ES Mainframe Environment"
Gennadiy I. Cherkes, Computer Research Institute, Minsk, USSR,
"Asynchronous Data Transmission on the EC-1840 PC"
Vladimir D. Novikov, VNIIPAS Institute of Automated Systems, Moscow, USSR,
"Experiments and Prolems in the Use of Kermit Programs in the National
Automatic Data Exchange Centre, USSR"
Konstantin Vinogradov, International Centre for Scientific and Technical
Information (ICSTI), Moscow, USSR, "Development and Implementation of Kermit
Programs for PCs, Minicomputers, and Mainframes"
Juri Gornostaev, ICSTI, Moscow, USSR,
"Proposals on the Arrangement of Distribution of Kermit Products to Member
Countries of ICSTI".
BIBLIOGRAPHY
[1] Gianone, Christine M., "The Kermit User Guide", Columbia University Center
for Computing Activities, New York City, 1989.
[2] Gianone, Christine M., and Frank da Cruz, "A Kermit Protocol Extension for
International Character Sets", technical paper, Columbia University Center
for Computing Activities, New York City, 1989.
[3] da Cruz, Frank, "Kermit, A File Transfer Protocol", Digital Press, Bedford,
MA, 1987.
[4] de Broeck, Paul, "Kermit Aids in Giotto Project", Kermit News, Volume 2,
Number 1, Columbia University Center for Computing Activities, New York
City, November 1987.
[5] Lowe, Sue J., "Kermit to Speak non-ASCII: Greek, Russian, et al.", Data
Communications Magazine, June 1989.
[6] "Paketi Prekladnikh Programm Teleobrabotki Dannikh na MikroEVM", ICSTI,
Moscow, 1987.
(end)