Kermit Software Source Code Archive - Web Access

Frank da Cruz
Page created:   September 2011
Most recent update:   Wed Sep 11 15:19:51 2024

[ Go straight to software archive table ]    [ NEW: DECSYSTEM-20 TOPS-20 Kermit 5.3(277), August 2024 ]

This is a newer version of the Kermit software archive page at Columbia University. 99% of the links still go to Columbia's archive, but anything new since 2011 is also reflected here on this page, with links going to kermitproject.org; these new additions show up with a  pale green background , CLICK HERE for a summary of the new additions.

Index to table entries:
[TOP] [acorn] [amdahl] [alphamicro] [apollo] [apple] [atari] [basicfour] [belllabs] [bootstrap] [burroughs] [ccdos] [cdc] [cie] [ckermit] [commodore] [computervision] [convergent] [cpm80] [cpm86] [cray] [datageneral] [dec] [dtss] [EC] [ekermit] [emacs] [gec] [gkermit] [gould] [harris] [honeywell] [hp] [ibm] [icl] [intel] [kermit95] [lilith] [lispmachine] [luxor] [microware] [minix] [modcomp] [motorola] [msdos] [mskermit] [mswindows] [ncr] [nicolet] [norskdata] [os-9] [perkinelmer] [perq] [pick] [prime] [radioshack] [rml] [sinclair] [sperry] [stratus] [sun] [tandem] [ti] [tripos] [ucsd] [umicrocomputers] [uniflex] [unisys] [unix] [victor] [END MATTER]

Welcome to what might be the biggest collection on earth of software source code for implementations of the same program on different computers and operating systems in different programming languages. This page provides, for the first time, convenient Web (HTTP) access to all the programs in the Kermit software archive. Each program can be downloaded as either a “tarball” (gzip'd tar archive for Unix) or a Zip archive (for Windows, VMS, and other platforms that have Unzip), and you can also look at or download individual files by clicking on "file list". In many cases manuals are also available, sometimes as PDF, others as plain text.

The once-common '.doc' and '.hlp' extensions that were used for plain-text-file have been converted (foo.doc → foodoc.txt; foo.hlp → foohlp.txt) to avoid unwanted invocations of Microsoft Word or Help programs; others have been left as they were. Here are some common ones:

Many more extensions can be found, for example for source code (.c, .h, .pas, .pl1, .for, .lsp, etc), but all files (with very few exceptions, such as PDF and PostScript) are plain text, unless they are in the archives or bin directories. More about the Kermit archive after the big table.

Here are the programs in the Kermit software archive. In addition to these, there are hundreds, perhaps thousands, of Kermit protocol implementations in commercial and shareware software packages such as Reflexions, Crosstalk, Procomm, Hyperterminal. Kermit protocol is also embedded in numerous devices, including routers, switches, factory-floor devices, medical devices and implants, cash registers, and perhaps most famously, HP calculators.

More about the Kermit software archive

I have not made any attempt to include offsite links to information about the many platforms where Kermit programs run, because any links will inevitably go stale and I might not be around to update them. If you want to find out more about some platform, search for it in Google or Wikipedia, or whatever has replaced them by the time you read this.

For the history of each Kermit program, there are several onsite resources:

1. The Info-Kermit Digest, a public mailing list run by the Kermit Project from 1983 to 1995 over ARPANET, BITNET, CSNET, CCNET, Usenet, LISTSERV, etc, and eventually the Internet.
2. The Kermit Newsgroups, comp.protocols.kermit.misc and comp.protocols.kermit.announce, which replaced the high-maintenance labor-intensive mailing list in 1994-95, and was active until 2011, maybe beyond. Newsgroup archives are available (as of this writing) in Google Groups here and here. However, there is not much filtering in the Google archive; the real messages are mixed with a fair amount of spam. As of 16 August 2011, the newsgroup archives are also available in the Kermit software archive (table just below).
3. The program's source files and accompanying documentation, available in the big table above.

The .misc newsgroup is also available in separate files, about 1MB each, at the Columbia Kermit archive.

Brief history of the Kermit Archive

Also there is no directory structure on these tapes; therefore files had to be grouped by filename. All files for a particular Kermit program had to be named with a unique 2- or 3-character prefix.

Furthermore, some of the most popular platforms in the early 80s, such as TOPS-10 and RT-11, restricted filenames to 6.3 format — six or fewer characters, a period, and then an "extension" of one to three characters. Thus, even for platforms that had more liberal naming rules, we tried to keep names within the 6.3 format so the files could be read, stored, and re-distributed from platforms that tolerated only short names.

Initially all the Kermit programs fit on one tape. But soon with the great flourishing of Kermit programs in the mid-1980s, the distribution grew to four tapes. By this time we also had our primary development and distribution machine, CU20B (Columbia University DECSYSTEM-20 B) on the ARPANET. Thus the Kermit files were split up into four directories:

It was a neverending struggle to keep each directory within the capacity of a tape reel. Every time a new version of some Kermit program was released, a tape overflowed. Then we would have to look for ways to save space. This was typically done by concatenating all the files for a particular Kermit program into one big (plain-text) file, with markers showing the divisions and filenames. This eliminated large numbers of interfile gaps, each of which took up a disproportionate amount of tape.

It must be said that also we removed some Kermit programs from circulation when they were "superseded" by a "better" version for the same platform. We did this simply because we could not fit a lot of alternative versions for the same platform on one tape.

By the mid-1990s, with the growth of the Internet, tape distribution had diminished to a trickle so we did not have to be so careful about the size of each directory. Meanwhile, the ever-increasing volume of Kermit Digest issues, newsletters, newsgroup archives, version lists, and so on demanded a "tape" of their own, E, while C-Kermit grew to such proportions that it received its own directory too, F (now ckermit). Also, since Kermit binaries and Zip and Tar archives could be distributed over the new networks (Internet, for example, unlike, say BITNET), a directory (archives) was created for them too.

For the assembly of this new Internet archive, all of the superseded programs have been made available again, and all of the text archives have been unpacked into their original constituent files with their original names (but made lowercase, if they were uppercase).

Another noteworthy event is that on August 16, 1988, we moved all the Kermit files from CU20B to a Unix machine (it was either a VAX-11/750 running 4.2BSD, or a VAX 8700 running Ultrix), just before CU20B was turned off for the last time. Since neither Kermit nor FTP nor ANSI-format tapes had a mechanism to preserve file dates, a lot of the older files in the present archive have that date. In putting this archive together, some of the earlier file dates were restored, but not consistently.

BOO Files

KERMIT.EXE
CEX00<81~2801:2OooHT60~3402[=ah0~20@00~~~~~~~~~~~~~=n0l]7;4OhRe?=P\m0?@0
\NSA]L_XU[G;j9C5bnQa7\_XZ[G;j3TFbnRLXL]A7Xc1S]WX9J<OFL_XSa7;j0igbnR>4\_X
FQ;;j1H2bnPS0\_X@0;;1XcHS\2Jb`gh3`O3j=2IbnR;8L_XVQ7;j5ZRbhTVg@W61\h10:3J
9Z;K9X3l17D6VP46n0o3P?`1M@JJ007h3l>0o0Ee1YYa1?P?`h3l1WD6VT85n0o3]0VjM03=
8NX^7OP?RbKM2OW3RAK[2ERSk@V99_49D8c8Xnl9aPIc~2a`Ke2O85a`KS2@X2a`KU2@X2
<l261W<0QRI]050b`8H6C2m@3^Q`05R61T`_F:9c08PVK@295VT0F:=S08TVI`1@S<RSI@2h

The first line gives the name of the original binary file, and the remaining lines are its encoded contents. Tildes (~) indicate places where compression was done. The writing of BOO-file encoders and decoders became quite the cottage industry. You can read the specification here, and you can find an impressive collection of BOO-file makers and decoders in the table below. In any case, with this mechanism we were able to mix text and "binary" files in the same directory and put them on tape. Although BOO files never caught on to any great extent in the wider world, they did their job for Kermit. In retrospect, a big improvement would have been a mechanism for gathering multiple files into a single BOO file.

The Kermit Protocol

The protocol manual is quite serviceable; it covers file attributes as well as both long packets and sliding windows, so it can be the basis of an efficient and fairly advanced Kermit implementation. It does not, however, cover character-set conversion, Unicode, "resend" (recovery of interrupted binary-mode transfers), streaming, control-character unprefixing, cross-platform directory-tree transfer, or various client-server protocol extensions. The files in this section are mostly plain text, but there are some typeset files too:

Kermit Books and Manuals Online

The KERMIT Acronym

KL10 Error-free Reciprocal Micro-Interface Transfer

(and some similar variations). KL10 was the processor of our DECSYSTEM-20 where the first Kermit program was written. Of course the protocol was designed to work on any computer but Bill couldn't find anything appropriate that started with K. Anyway, this legend was put into one or two pieces of source souce, which escaped from Columbia, and then found their way into other pieces of source code that you will find in this archive, and still provokes a great deal mirth. For our part, we removed it from our own code as soon as had a letter of permission from Henson Associates to keep using the name Kermit.

Patterns and Trends

As to languages, at first we had a great proliferation of them, including many different assembly languages (PDP-10, IBM mainframe, 8080, 8086, mc68000...) But by, say, the mid-90s, the only non-interpretive language that mattered any more was C and its offshoots. Which is kind of a shame because many of the older languages were more consistent, expressive, powerful, stable, and robust. And better documented and less aggravating.

About 36 languages in all; more counting variant dialects. Among the missing: APL, SNOBOL, SAIL, Simula, Algol-68, Ada... And the Kermit language itself!

Documents, Utilities, Scripts, Tables, and Miscellaneous

Unix versions covered by C-Kermit and/or G-Kermit