Tickling old memories

There was a time Kermit was an essential Swiss Army knife. Still is, I'm sure, in some cases.

What a relief it was to use when I could transfer files from two incompatible mainframes without punching cards or going through the bureaucracy to remove a tape from the machine room. There was a time also when two campus mainframes physically sitting 15 feet apart could only transfer files by using BITNET to cross half the country to a gateway machine and then CSNET get back.

Younger generations can take for granted what once was a thicket of competing protocols, character sets, line termination conventions (which are still around), etc.

Even ASCII wasn't a given. I routinely had to convert from 6 and 12 bit CDC display codes, GE/Honeywell's codes, BCD and EBCDIC, Univac Fieldata. At least I was beyond the Baudot era!

And Kermit implementations were made for all these machines that were otherwise incompatible, recognizing only the commonality of the recently-emerged RS232C protocol. And given mainframes of the time often had frontends that could intervene in malicious ways or with limited buffers for incoming data, handshaking wasn't trivial either.

Nor were the implementations easy: there wasn't a universal language like C to go around! That's why this one is explicitly called C-Kermit! There were different assembly implementations for 8080, DEC TOPS platforms, CDC Cybers, IBM 370, Apple II, Commodore 64 6502. Other implementations were made in PL/I subset languages, in ALGOL, in Lisp , and even in COBOL. Over 300 implementations existed at its peak!

Colombia University organized the effort to fit all these pieces together and stewarded it for 30 years retiring it only in 2011.

All honor to those who helped us move data around at a time when it wasn't easy to do so. Particular kudos to Frank da Cruz as the pioneer of practical interoperability. You will all not go forgotten.