←back to thread

Ward Christensen (of BBS and XMODEM fame) has died

(en.wikipedia.org)
322 points DamonHD | 2 comments | 13 Oct 24 15:39 UTC | HN request time: 0.403s | source
Show context
shrubble ◴[13 Oct 24 18:56 UTC] No.41830558[source]▶
A friend who works in embedded systems pointed out that XMODEM protocol communication is used everywhere in embedded; it may be that the protocol is more widely shipped now than it has been in the past!

Many Cisco, Adtran, Juniper etc switches and routers have it in their firmware also.

replies(2): >>41833347 #>>41841927 #
cmptrnerd6 ◴[14 Oct 24 01:29 UTC] No.41833347[source]▶
We just upgraded from Xmodem to Kermit last week in our embedded device.
replies(1): >>41833448 #
shrubble ◴[14 Oct 24 01:47 UTC] No.41833448[source]▶
Kermit is interesting also, though my understanding was that it was not quite as good as taking advantage of higher available data rates; am I wrong about that?
replies(2): >>41833571 #>>41843920 #
1. cmptrnerd6 ◴[14 Oct 24 02:12 UTC] No.41833571[source]▶
Entirely possible I don't understand something as I'm not sure what you mean by taking advantage of higher data rates. We're sending data over UART at the max rate that doesn't introduce too many bit errors. Mainly we needed a protocol to correct for the bit errors (by rerequesting data) and I've always used x/y/z modem for that purpose.

I picked Kermit this time because I didn't want to implement x/y/z modem again and had never used Kermit.

This [1] claims Kermit is faster in some instances depending on what features you have enabled.

[1] http://www.columbia.edu/kermit/perf.html

replies(1): >>41840950 #
2. shrubble ◴[14 Oct 24 19:24 UTC] No.41840950[source]▶
I was wrong, and you are correct! I had thought that XMODEM was more efficient as the bandwidth increases but I was wrong; due to the fixed length of time for the ACK response, the efficiency drops to under 80% at a data rate of 9600bps; Kermit does not have this issue.