Correction on my last e-mail - muscle memory took over and I typed 98N1D -
this should 98N1E with the E to enable XON/XOFF.  Classic case of target
fixation right there. Don't type D, don't type D.. Typed D.

John, I have had some issues with how enabling flow control is done. The
method I settled on was to enable it with stty after logging in through a
sourced .bashrc script. I abandoned trying to get it to work natively in
getty, even eventually swapped out getty for the more supported agetty set
in "old" mode.

While I may be connecting at 19200 the speed is dependent on the screen
unfortunately. That can get downright frustrating. Hackerb9 explains
several other fixes for keys and stuff way better than I ever could in that
git hub link I put up earlier. I recommend just about everything he talks
about except I can't condone the use of emacs or nano. Real men use VI.

One other possibility would be the serial adapter you are using. I have had
issues with the voltage levels being just off enough on cheap USB to serial
adapters that the M100 had troubles.

Brian


On Mon, Dec 11, 2023 at 4:45 PM John R. Hogerhuis <[email protected]> wrote:

> Interesting... I could never go that fast without h/w flow control...
> Linux wouldn't drop characters, but it would overrun the T's receive buffer
> because it didn't react immediately to xoff. I don't know why... my theory
> was maybe the driver doesn't see the xoff until it popped out of the
> stream. Maybe screen in the middle processes the xoff sooner. All I had in
> the loop was getty or equivalent.
>
> Between that and ANSI escapes from xterm and utf8 and whatnot full screen
> console stuff was always messed up without HTERM to filter and flow
> control.
>
> -- John.
>

Reply via email to