<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>
Date   : Tue, 05 Apr 1983 0704-:00CST
From   : Doug <HUNEYCUTT@gunter-adam.arpa>
Subject: Retry...

Tried this msg a while back, but mailer died.

Those of you with TOPS-20 implementations of MODEM7 may need to be
aware of a new 'feature' of the OS.  TOPS-20 is now trying to impose
flow control on its TTY lines.  A guy at ISI told me that after about
60 chars coming in at 1200-up, the 20 would start throwing out CTL-S
characters trying to stop the sender.  These CTL-S's are interpreted
by MODEM7 (micro side) and MDM705, not as a NAK, but rather as ACK*
(not ACK).  This causes repeated retransmission attempts and finally
aborts the process.  The only way we could tell this was to hook up a
data line monitor.  The 20 side would ACK a block, but the CTL-S had
already initiated a retransmission.  Obviously, this only occurs when
you are trying to download from a micro to the 20.

  Solution:  In GETACK, after you compare for ACK, compare for NAK.  If
not NAK, loop through GETACK.  This will ONLY allow ACK and NAK to
be used for control, but it gets the job done.  Might cause some problems,
but we haven't experienced any.  

  Same guy at ISI told me that 131 or 132 characters at high load and
4800 baud would lose half the block on the 20 side.  In our experience,
this has not been the case.  The front end seems to handle our blocks
without any loss.

BE AWARE!!  If you simply add the two lines necessary for the fix
above, overlay segments for MDM705 will be affected.  We have made a 
separate overlay to correct the problem which removes some of the 
interpretation code in GETACK to maintain spacing.

If anyone sees problems with this approach, please let me know.

Doug
<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>