by fcambus on 4/25/23, 4:45 PM with 107 comments
by jerf on 4/25/23, 5:16 PM
ooooooo whooooooo OOOOOOOOO AAAAAAAAAAAAAAA !!!!!!!!!!1!! qcjqrjrcorRC!!!
WHAT THE HECK WAS THAT? Did I just do megabytes per second? Holy shit. Am I high or something? What is this hardware? What is going on here?
No... wait... is that the end? Am I done here? No! No, I want to transfer more! More! Megabytes per second! Gigabytes per second! I can see it now, I want it, I want more, please, let's transfer another file, come on man, I want to ride again, please ple
EXECUTION COMPLETED
$by NoZebra120vClip on 4/25/23, 11:36 PM
Once when I should've been in college still, I was using the Telix terminal program on a 286 with a 2400MNP5 modem. Now Telix had a really awesome scripting language that could do all sorts of terminal magic, but that's a story for another day.
My connection to the Internet was 8-bit clean, and at the time I was an avid Emacs user/evangelist. (Since then, I have seen the light of vim's face and have never turned away.)
I saw that Emacs included a mapping file for mskermit to generate ESC sequences for Alt+Keypress, and I figured out that if I mapped each scancode for Alt+Keypress to the appropriate 8-bit code, it would convert my Alt key into a Meta buckybit, perfect for Emacs usage!
So I painstakingly mapped out each scancode and transcribed it into the ms-kermit configuration, and before long I was Emacsing in all its 8-bit-meta-glory.
I figured it would be useful to other Emacs users, and so I sent the file upstream. Lo and behold, it was incorporated into the standard Emacs distribution, and it stayed that way for a long, long time.
Unfortunately, since I had never formally assigned copyright or done whatever legal bit needs to be done to GPL the code, TPTB did an audit of the Emacs distro and culled any code that did not have solid legal footing. At that point, ms-kermit was really obsolete anyway, so I suppose it's all for the best!
by bm3719 on 4/25/23, 5:31 PM
That worked fine but every now and then, a site would have some inline image that I'd want to see, so I'd view source to get the img URL and then download that to /tmp and transfer the file to my local machine via zmodem. Usually, it wouldn't be worth the effort. I'd also download mp3s off IRC via DCC and queue up a bunch of data for zmodem to transfer overnight or when I was in class. I really appreciated those bytes back then. Now, not so much.
by mikecoles on 4/25/23, 5:14 PM
https://web.archive.org/web/19980627010642/http://www.termin...
by glonq on 4/25/23, 5:30 PM
It felt familiar but strange and wrong to do such a thing. Worked great though.
by rsync on 4/25/23, 8:12 PM
I have actually used 'sz' and 'rz' in relatively modern times for quick and dirty file transfer and found it very convenient in a very narrow set of use-cases.
However ...
It's a serious violation of the cleanliness and available attack surface involved in a terminal interface and we should be on the lookout for, and reject, similar interfaces and applications.
In order for zmodem to work over the terminal, the terminal program itself needs to know something about the text flowing over the connection and then invoke special, extra routines based on monitoring that textual flow.
This opens up all manner of weird, extra attack surface.
The beauty of the text terminal is that I can, theoretically, cat any file I want to without fear of what it contains. I can open up (perhaps with 'strings' or 'hexedit') any email attachment without fear of the strings that it contains. I can do this because I am using a dumb terminal.
As soon as the terminal is smart - even a little bit - you've got vectors for weird strings doing things you don't want them to.
by diydsp on 4/25/23, 10:04 PM
Yep that's 1.5 million times the speed I used to get.
by jebr224 on 4/25/23, 5:52 PM
zmodem can also be used in embedded spaces to retrieve files if the only interface is a serial port.
by poettering on 4/25/23, 7:20 PM
by ractive on 4/25/23, 9:58 PM
by sedatk on 4/25/23, 5:31 PM
by jmclnx on 4/25/23, 5:10 PM
by spudlyo on 4/25/23, 6:34 PM
Looks like it's still quite possible, I wonder if our network monitoring tools would have noticed gigabytes of data flowing out of the network that way.
by distantsounds on 4/25/23, 7:40 PM
by seanmceligot on 4/25/23, 6:14 PM
You can accomplish the with a new scp session on the client server, but it's an extra step. I use this as a helper when for building the scp command.
function scppath() { echo $USER@$(hostname).$(dnsdomainname):$(realpath $1) ]
by Tepix on 4/25/23, 5:39 PM
Aren't brains amazing, storing all those ancient unused acronyms for decades?
by thomashabets2 on 4/25/23, 10:19 PM
Should work through multiple SSH hops, and not giving the hassle of using scp through those same sets of hops.
by dublin on 4/26/23, 12:29 AM
I used this frequently to move files back in the dialup modem days, after doing an ugly redirect to get the Zmodem binary on the far end, and because of its superior compression, Zmodem was faster than uucp/uucico (when you were lucky enough to have the uucp suite installed and configured, which unlike Kermit and Zmodem, required root privileges...)
by knorker on 4/25/23, 10:02 PM
by jedberg on 4/25/23, 5:31 PM
by pimlottc on 4/25/23, 6:25 PM
by zabzonk on 4/25/23, 6:45 PM
by marcodiego on 4/25/23, 5:21 PM
by ezekiel68 on 4/26/23, 5:53 AM
by tech-no-logical on 4/25/23, 9:28 PM
by thrownaway561 on 4/25/23, 5:47 PM
by johng on 4/25/23, 7:59 PM
by Jemm on 4/26/23, 1:30 PM
by YesThatTom2 on 4/25/23, 6:16 PM