>I figured somebody with a lot more knowledge then I have would have
>answered by now too.  The answer to "how is flow control handled
>on the HP3000" is that it depends on the term type used (set either
>in sysgen or in the HELLO command).
>
>For there standard terminals (using term type 10) the HP3000 uses a
>very non-standard protocol called ENQ-ACK.  I forget the details but
>someone else can enlighten you if more information is need or I can
>dig up some old documentation.
>
>NMD
>
>
 
I'm not sure what flow control or other information you need regarding
the Classic HP 3000's, so pardon me if I ramble.
 
I'm sure Ross et al at Telamon can provide more details, but if I
remember, the the standard MPE V/E TT10 file uses Enq-Ack as a pacing
method rather than as true flow control.  The HP would send a buffer
(typically 80 bytes, but in the old days it was configurable) followed
by an ENQ character.  The terminal would absorb the characters and
respond with an ACK character.  The HP would then send the next block,
etc.
 
But, there was also a time-out, which if I remember correctly was also
configurable and was something like 10 seconds.  If the HP 3000 didn't
receive an ACK within 10 seconds it would proceed as if it had and would
sent the next buffer of output data.  The early terminals (HP 2600) could
run at very high data rates - as high as 2400, but they had trouble keeping
up with the data at such high transmission rates and ENQ-ACK was the way
they kept the system from overrunning the terminal.
 
The older HP 3000's didn't always respond to X-OFF - either the first time
an X-OFF was sent or sometimes the first dozen or so X-OFFs.  Some HP
devices (like the old serial printers) had a "robust X-OFF" feature which
meant that if after the first X-OFF they were still receiving data, they
would keep sending X-OFFs until the HP 3000 stopped sending data.  Some
datacomm devices REALLY appreciated robust X-OFF (muxs, modems) and caused
all sorts of problems.
 
ESC was another problem character.  It was paired with ":" and ";" so that
the user could easily turn input echo on or off.  Unfortunately, the
processing of the ESC in the input data stream took some time and caused
those with smaller (slower) systems no end of grief at data rates of
9600 or 19200.  The Micro GX, XE, and Series 37 suffered this problem.
Programs like PSCREEN, (and many others) which read data from
terminal memory would have their FREADs fail with data overruns if there
were ESC sequences in them.  The ESC sequences took so long to check to
make sure they were not ESC ;  or ESC :  that a buffer overrun would
occur (if anyone has this problem and needs the workaround, let me know
and I'll dig up the solution).
 
Also, the flow control options vary with the release of MPE V/E.  On 2P and
older versions you could change the size of the Enq-Ack buffer from 80
bytes to as much as 255 bytes.  "Progressing" to 3P took this "feature"
away.
 
Anyway, there were a lot of quirks and some really nice features as
well.  Hope this helps.
 
--------------------------------------------------------------
John Korb                            email: [log in to unmask]
Innovative Software Solutions, Inc.
 
The thoughts, comments, and opinions expressed herein are mine
and do not reflect those of my employer(s), or anyone else.