27 Feb
2013
27 Feb
'13
9:27 a.m.
On Wed, Feb 27, 2013 at 5:02 AM, Peter Coghlan <cctech at beyondthepale.ie> wrote:
Hardware flow control in the DEC world (and many other places) is commonly
RTS/CTS (Request To Send, Clear To Send).
DTR (Data Terminal Ready) is, as you say, for indicating the device (terminal)
is operational and on-line. I won't say nobody has ever done something
"creative" with that signal, but IME, when you power up a terminal
or when you open a serial port on a computer, DTR asserts after all
the initialization stuff is done and the device is good to go.
The VT100
doesn't need padd characters IIRC (VT05 and VT52 definitely
do), but it uses a signalling scheme to tell the host that it's
internal buffer is full. This signalling scheme can be either in the
hardware signalling (DTR/CTS) or it can be XON/XOFF (DC1/DC3) control
characters in the data stream.
According to my VT102 manual, DTR is only turned off when the terminal is
offline or performing a line disconnect. It doesn't seem to turn off DTR or
any other signal in response to the input buffer becoming full. Under "Input Buffer Overflow Prevention", it
goes on to say that there are
three methods of input buffer overflow prevention and lists XON/XOFF, fill
characters and low speed operation.
Yep.
A table shows the requirement for fill characters for
various functions at
various baud rates. It suggests that all functions require at least one fill
character (null) at 9600 baud, even when not in smooth scroll mode.
Back in the day when I used real DEC terminals 8-10 hrs a day,
every day, we ran them at 9600 baud, jump scroll, no fill chars,
with no problems, but we were feeding them from a slow VAX
(11/730 or 11/750) or a medium-grade PDP-11 (11/24 or 11/34,
usually). There's no way we were saturating the line with that
equipment, in that environment.
Where we _did_ have problems is 9600 baud connections for file
transfer between machines. Notably, we had to ensure flow-control
(hardware or software) was on and working to successfully Kermit
files between our MicroVAX and 11/750 (we had Emulex
serial cards on our 11/750 with DMA capability on transmission
but I don't think there was even a 3-char silo on input - one
received character, one interrupt to handle. Humans didn't
type fast enough to cause problems, but Kermit could spew
plenty fast enough to make either end choke).
-ethan