I complied Kermit for IBM's AIX in ~1989 as a study in "how compatible is AIX to other Unixes?" Kermit source code remains the high-water mark in my memory for how many different, incompatible, non-standard platforms one piece of software can support. More than even Unix itself, as it could also run on VMS and non-Unixes of the time. I have never in my life seen so many `#ifdef`s and such complicated "Am I running on this platform or that one? Does it have this feature or that feature? No, what about this other one? Is it HP-UX maybe? HP-UX 6? No, 7? No, 8? No, what about VMS? No, how about SunOS? No, how about Unix System III?" ...on and on, seemingly endlessly.
Kermit was an extraordinary monument to portability in an era when "portable" often meant "we have individually accommodated every incompatible system anyone has ever built or might possibly still be using, however improbably."
Which is interesting because most autoconf programs I've seen do the opposite. They have all these defines and then don't use them. Just seems like a waste of time. Cmake has been better but still sometimes cmake will pass and the code still doesn't build
I remember mostly ignoring Kermit back in my BBSing days; by the time "SuperKermit" came along, which could get performance competitive with ZMODEM, it alreayd seemed ancient.
Then, around 1991, a friend gave me access to his college Unix account so I could play with Gopher, Archie, and TinyFugue (the world of MUDs/MUCKs/etc.). The catch: to get to his account from my home machine, I had to call into a Florida state network that let you dial in and connect to machines on various college campuses. The network ran on some IBM minicomputer. And, it didn't connect to the Unix machine, oh no. It connected to a VAX, which was then on the same local network as the Unix machine.
So I had to dial in to the IBM machine, connect from it to a VAX, and then telnet from the VAX to the Unix box.
Incredibly, this all worked with only mildly annoying lag...but then I tried to transfer a file from the Unix machine to my personal machine. ZMODEM didn't work. XMODEM didn't work. I think I tried one or two other things that I've forgotten.
Then I remembered Kermit, and its reputation for being able to handle difficult connections. It got the file through on its first try.
Bill Catchings (the undergrad who helped Frank da Cruz with the original spec and coding) still blogs. It has been a while since he spoke specifically about kermit, but he has one or two interesting posts about the 40th anniversary of Kermit.
Good Lord. That takes me back. I ported Kermit to the Computervision CGOS environment. It was very popular in the user community. Internally, it was used to transfer files between systems over the terminal concentrator when you didn't want to walk down to the machine room, scavenge a mag tape, and put up with FUTIL.
I had never heard of CGOS, but this is still up on the Kermit archive! https://kermitproject.org/archive.html Are you Val? Looks like you had to add some hacks to deal with long file naming conventions on CGOS but short filenames on CP/M.
I honestly don't remember. It took me a moment to even get back to TPL, a variation on RATFOR. I don't remember anything about the BYU conversion. That must have happened after I left CV.
I am a gray beard and still use Kermit regularly for embedded development. It’s very handy for automating things over a serial port console. I wonder how much adoption it sees with the younger crowd.
I remember back in the day Kermit was pretty heavyweight. We used sz/rz for file transfer [z-modem: good to see the man pages are still up! https://linux.die.net/man/1/sz]. Is there a reason to prefer Kermit?
Kermit as a file transfer protocol, in the more modern implementations at least, supports 4K and 9K blocks, streaming mode (no need for ACKs), sliding windows, etc. As a protocol, it also has an OS-agnostic way to do file management; you can send commands like delete, rename, copy, etc. and the remote end does it in a general way. It's like SFTP, only a few decades earlier.
Kermit initially competed with xmodem which was very primitive (but a lot smaller/simpler).
By the time zmodem became popular, the biggest remaining selling point of Kermit was the built-in scripting. Not much use for one-off BBS-style transfers, but very handy for keeping a directory in sync etc.
Can you elaborate more on how you use it or do you recommend any codebase to learn more?
I remember evaluating Kermit and ymodem for my embedded work too, but didn't end up using them. Maybe I didn't understand Kermit fully. I wanted something to connect over serial (usb cdc and ble) via the browser and manage files and the state of the device.
I always end up writing something worse and half-baked in the end..
I used Telix under IBM PC-DOS for scripting. The SLT scripts were as versatile as Expect or Tcl in that day.
I used MS-Kermit as a terminal emulator on the same system. I was accessing academic Unix systems and probably also remote access at my first job. MS-Kermit had a keycode translation capability that I used extensively.
My 2400/MNP5 modem sometimes only had a 7-bit connection. Without being 8-bit clean, I couldn’t directly use the Meta key with GNU Emacs. But sometimes I did achieve an 8-N-1 clean connection in the same modem pool!
So I wrote up two large tables for MS-Kermit: one for the 7-bit connection that would send ESC to simulate Meta for Emacs. And one for 8-bit-clean operation that used ALT as Meta and sent the ASCII code >=128.
I eventually submitted them to the GNU Emacs project and one was distributed with Emacs 18. Eventually, they were forced to remove it during a copyright/license cleanup audit. But it was my claim to 15 minutes of fame.
Why wasn’t Kermit used by BBSes? Even at the time I remember having read about it in computer magazines, but I got my first modem in ‘82 or so and I don’t recall ever seeing it in the wild. My first actual contact with it was uploading firmware I wrote to an embedded controller. But the thing it was actually designed for? Not even once.
So… why? We didn’t collectively settle on YMODEM and ZMODEM until many years later. There were lots of early protocols to choose from back then, too. Punter, anyone? No Kermit, though.
Kermit had licensing issues in those days. And many bbs were focused on one specific system so many of the auto translate and nice compatibility features of Kermit weren’t that important.
And finally Kermit didn’t have great defaults especially compared to ymodem and zmodem. So getting good performance out of Kermit was much harder than the others.
All of that added up to much lower adoption outside of the mainframe world.
The very oldest (early 80s) Kermit implementations were limited to 96-byte packets. Kermit gained support for 4K and 9K jumbo frames, sliding windows, etc., and was generally faster than ZModem -- when there was a modern implementation. But a lot of BBSs (and also client-side communication software) did their own from-scratch Kermit implementation, which was tiny packets and no sliding windows, so it was slow. There's some discussion of it here: https://www.kermitproject.org/misconceptions.html#notslow
I had an Apple 2+ back in 1992. It was an old computer, even then.
The Apple 2+ only supported displaying capital letters.
I remember calling a few local BBSes using terminal software that let me press the right arrow (I think) and the character key, and it would display the character inverted (white background, black letter).
Hayes 1200 baud serial modem (I think) hooked up to a serial card.
Have you not been on the internet much lately? As the sibling comment pointed out, it’s Anubis, which shows up on a lot of websites of FOSS devs (and which has been discussed endlessly on HN).
Ages a ago I used it on Coherent OS via dial up to sign into a Sun Server at work that I used for USENET and general internet access. When I dialed in the server would hang up and call be back so I would avoid Long Distance Charges. ckermit handled that itself, fun times.
I purchased the book to help with the project, I should go and dig it out :)
It (or rather, they, as there are a number of Kermit books) is pretty good. I've got a few on my bookshelf. OK, I'll admit, more Kermit and UUCP books than I probably need, but oh well. The Kermit books are mostly available online now also: https://www.kermitproject.org/booksonline.html
I complied Kermit for IBM's AIX in ~1989 as a study in "how compatible is AIX to other Unixes?" Kermit source code remains the high-water mark in my memory for how many different, incompatible, non-standard platforms one piece of software can support. More than even Unix itself, as it could also run on VMS and non-Unixes of the time. I have never in my life seen so many `#ifdef`s and such complicated "Am I running on this platform or that one? Does it have this feature or that feature? No, what about this other one? Is it HP-UX maybe? HP-UX 6? No, 7? No, 8? No, what about VMS? No, how about SunOS? No, how about Unix System III?" ...on and on, seemingly endlessly.
Kermit was an extraordinary monument to portability in an era when "portable" often meant "we have individually accommodated every incompatible system anyone has ever built or might possibly still be using, however improbably."
Which is interesting because most autoconf programs I've seen do the opposite. They have all these defines and then don't use them. Just seems like a waste of time. Cmake has been better but still sometimes cmake will pass and the code still doesn't build
I remember mostly ignoring Kermit back in my BBSing days; by the time "SuperKermit" came along, which could get performance competitive with ZMODEM, it alreayd seemed ancient.
Then, around 1991, a friend gave me access to his college Unix account so I could play with Gopher, Archie, and TinyFugue (the world of MUDs/MUCKs/etc.). The catch: to get to his account from my home machine, I had to call into a Florida state network that let you dial in and connect to machines on various college campuses. The network ran on some IBM minicomputer. And, it didn't connect to the Unix machine, oh no. It connected to a VAX, which was then on the same local network as the Unix machine.
So I had to dial in to the IBM machine, connect from it to a VAX, and then telnet from the VAX to the Unix box.
Incredibly, this all worked with only mildly annoying lag...but then I tried to transfer a file from the Unix machine to my personal machine. ZMODEM didn't work. XMODEM didn't work. I think I tried one or two other things that I've forgotten.
Then I remembered Kermit, and its reputation for being able to handle difficult connections. It got the file through on its first try.
Bill Catchings (the undergrad who helped Frank da Cruz with the original spec and coding) still blogs. It has been a while since he spoke specifically about kermit, but he has one or two interesting posts about the 40th anniversary of Kermit.
https://www.billcatchings.com/2012/04/frog-from-my-past.html https://www.billcatchings.com/2013/11/kermit-oral-history-pa...
and here is a link to the oral history he discusses:
https://www.computerhistory.org/collections/catalog/10274641...
Good Lord. That takes me back. I ported Kermit to the Computervision CGOS environment. It was very popular in the user community. Internally, it was used to transfer files between systems over the terminal concentrator when you didn't want to walk down to the machine room, scavenge a mag tape, and put up with FUTIL.
I had never heard of CGOS, but this is still up on the Kermit archive! https://kermitproject.org/archive.html Are you Val? Looks like you had to add some hacks to deal with long file naming conventions on CGOS but short filenames on CP/M.
I honestly don't remember. It took me a moment to even get back to TPL, a variation on RATFOR. I don't remember anything about the BYU conversion. That must have happened after I left CV.
I am a gray beard and still use Kermit regularly for embedded development. It’s very handy for automating things over a serial port console. I wonder how much adoption it sees with the younger crowd.
I remember back in the day Kermit was pretty heavyweight. We used sz/rz for file transfer [z-modem: good to see the man pages are still up! https://linux.die.net/man/1/sz]. Is there a reason to prefer Kermit?
Kermit as a file transfer protocol, in the more modern implementations at least, supports 4K and 9K blocks, streaming mode (no need for ACKs), sliding windows, etc. As a protocol, it also has an OS-agnostic way to do file management; you can send commands like delete, rename, copy, etc. and the remote end does it in a general way. It's like SFTP, only a few decades earlier.
In modern times, I use it as a powerful ssh wrapper and BBS client. Its character-set translation is ideal for CP437 to UTF8. See https://www.openkermit.org/ckermit/ssh/ and https://www.openkermit.org/ckermit/bbs/
Kermit initially competed with xmodem which was very primitive (but a lot smaller/simpler).
By the time zmodem became popular, the biggest remaining selling point of Kermit was the built-in scripting. Not much use for one-off BBS-style transfers, but very handy for keeping a directory in sync etc.
Can you elaborate more on how you use it or do you recommend any codebase to learn more?
I remember evaluating Kermit and ymodem for my embedded work too, but didn't end up using them. Maybe I didn't understand Kermit fully. I wanted something to connect over serial (usb cdc and ble) via the browser and manage files and the state of the device.
I always end up writing something worse and half-baked in the end..
I used Telix under IBM PC-DOS for scripting. The SLT scripts were as versatile as Expect or Tcl in that day.
I used MS-Kermit as a terminal emulator on the same system. I was accessing academic Unix systems and probably also remote access at my first job. MS-Kermit had a keycode translation capability that I used extensively.
My 2400/MNP5 modem sometimes only had a 7-bit connection. Without being 8-bit clean, I couldn’t directly use the Meta key with GNU Emacs. But sometimes I did achieve an 8-N-1 clean connection in the same modem pool!
So I wrote up two large tables for MS-Kermit: one for the 7-bit connection that would send ESC to simulate Meta for Emacs. And one for 8-bit-clean operation that used ALT as Meta and sent the ASCII code >=128.
I eventually submitted them to the GNU Emacs project and one was distributed with Emacs 18. Eventually, they were forced to remove it during a copyright/license cleanup audit. But it was my claim to 15 minutes of fame.
Why wasn’t Kermit used by BBSes? Even at the time I remember having read about it in computer magazines, but I got my first modem in ‘82 or so and I don’t recall ever seeing it in the wild. My first actual contact with it was uploading firmware I wrote to an embedded controller. But the thing it was actually designed for? Not even once.
So… why? We didn’t collectively settle on YMODEM and ZMODEM until many years later. There were lots of early protocols to choose from back then, too. Punter, anyone? No Kermit, though.
Kermit had licensing issues in those days. And many bbs were focused on one specific system so many of the auto translate and nice compatibility features of Kermit weren’t that important.
And finally Kermit didn’t have great defaults especially compared to ymodem and zmodem. So getting good performance out of Kermit was much harder than the others.
All of that added up to much lower adoption outside of the mainframe world.
The very oldest (early 80s) Kermit implementations were limited to 96-byte packets. Kermit gained support for 4K and 9K jumbo frames, sliding windows, etc., and was generally faster than ZModem -- when there was a modern implementation. But a lot of BBSs (and also client-side communication software) did their own from-scratch Kermit implementation, which was tiny packets and no sliding windows, so it was slow. There's some discussion of it here: https://www.kermitproject.org/misconceptions.html#notslow
This is what I remember. Kermit hated latency, and dial up modems had that in spades. It was much slower than Z-Modem for pushing files around.
GOO. I remember Punter well.
I had an Apple 2+ back in 1992. It was an old computer, even then.
The Apple 2+ only supported displaying capital letters.
I remember calling a few local BBSes using terminal software that let me press the right arrow (I think) and the character key, and it would display the character inverted (white background, black letter).
Hayes 1200 baud serial modem (I think) hooked up to a serial card.
That's how I knew I was sending a capital letter.
Not sure if it was Kermit though..
"1981 was a different time for computing. It was expensive (both hardware and software)"
doesn't sound so different anymore...the VCs have managed to go back to the future...
"the name Kermit is used by permission of Henson Associates, Inc."
that certainly would never happen nowadays.
I used Kermit in many projects when I was younger, including synchronising data from remote branches of a Pool equipment company using dial-up modems.
Those were the days lol.
I'm amazed to see that Kermit is still alive and kicking. Thank you, team!
ok but what's the anime 'checking if you're not a bot' gatekeeper? asking for a friend...
Anubis.
https://hn.algolia.com/?q=anubis
Have you not been on the internet much lately? As the sibling comment pointed out, it’s Anubis, which shows up on a lot of websites of FOSS devs (and which has been discussed endlessly on HN).
Yeah. It's pretty unprofessional.
Nice and congratulations
Ages a ago I used it on Coherent OS via dial up to sign into a Sun Server at work that I used for USENET and general internet access. When I dialed in the server would hang up and call be back so I would avoid Long Distance Charges. ckermit handled that itself, fun times.
I purchased the book to help with the project, I should go and dig it out :)
It (or rather, they, as there are a number of Kermit books) is pretty good. I've got a few on my bookshelf. OK, I'll admit, more Kermit and UUCP books than I probably need, but oh well. The Kermit books are mostly available online now also: https://www.kermitproject.org/booksonline.html