[HECnet] Re: PRO/DECnet?

On Oct 1, 2023, at 11:14 AM, Johnny Billquist <bqt(a)softjar.se> wrote: ARGH! I've finally found a good argument for really kicking that whole ugly mess of pacing serial ports to "fake" the real experience. Turns out it's not as good as it thinks. For normal terminal use, that might be fine, but I'm actually doing something else here... So, as I mentioned, I have this XHomer instance, which I've actually have hooked up to HECnet in the past. Well, as I tried yeasterday, I failed to get it working. The circuit was constantly going down and coming up again. I had a nagging suspicion about the pacing, since it had worked in the past, but I wasn't entirely sure if that was with a simh old enough to not have had that pacing. And looking at communication, I didn't see any huge time issues, so I went and put in loads of debugging in simh, in XHomer, and also on my fake null modem cable between the two instances. After a lot of messing around I was finally sure that the data as such was not corrupted anywhere between the two sides, in either direction. So all data was getting across correctly... However, there was a little bit of funniness I noticed on the timing sometimes. Well, it turns out that the pacing actually is the problem. The DDCMP handling in DECnet have tight enough timing requirements on the serial that the pacing in simh cause timeouts in the DECnet communication. I wasted almost two days on that "feature" that I have hated from day 1.