home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Columbia Kermit
/
kermit.zip
/
archives
/
sperryunivac1100p.tar.gz
/
sperryunivac1100p.tar
/
univac.ann
next >
Wrap
Internet Message Format
|
1984-10-04
|
5KB
Date: Thu, 04 Oct 84 14:51:26 EDT
From: Edgar B. Butt <BUTT@UMD2.ARPA>
To: sy.fdc@cu20b
Subject: Kermit in Pascal for the Sperry 1100
Here is a Kermit implementation for the Sperry 1100 systems written
in Pascal. It has been run successfully here at the University of Maryland,
College Park, and at SUNY, Albany. Please add it to your selection
of Kermits. I would appreciate feedback from anyone who tries it.
The first page of code consists of comments explaining how to
use and generate Kermit1100.
Hop someone finds it useful,
Edgar Butt (Butt@umd2.arpa)
Computer Science Center
University of Maryland
College Park, Maryland 20742
(301) 454-2946
The source for Kermit1100 version 2.0 begins on the next line.
{Kermit1100 - see first executable line in main block for version
KERMIT1100 is yet another Kermit written to run on the Sperry (Univac)
1100 series of computers. It is written in Pascal to be compiled on
the NOSC Pascal Compiler, version 2.2 or later. This compiler is
available from the Computer Science Center of the University of
Maryland, College Park, for a nominal service charge.
Kermit aficianodos may notice that the structure of this version
differs from other versions in that packets are read and sequence
checked in the main program loop and are then dispatched to the
proper input or output state with a single case statement.
This structure has allowed the various state processes to be
relatively uncluttered. While doing this implementation I
discovered that NAK's are like tadpole tails. They seem like
a neat idea at first, but as the frog emerges, they serve no
useful purpose. Likewise, I have been unable to find a case
in which NAK's are necessary. Sending an ACK for the last
good packet received is just as good. If I'm wrong, I am sure
that some swamp dweller out there will let me know.
(Not to worry, I handle incoming NAK's even though they are not
necessary.)
By way of a quick synopsys of features, this version of Kermit has:
Simple server mode - processes S and R packets
8-bit quoting (Turned on by Q-option)
Repeat count prefixes
Error packet generation and processing
Kermit 1100 is called as a processor with the following control card:
@Q*F.KERMIT,OPTIONS 1100SPEC,REMOTESPEC
Q*F. is the file in which the processor resides.
1100SPEC is the 1100 file or element on which Kermit will operate.
REMOTESPEC is the file name sent to the remote Kermit(a fib of sorts)
OPTIONS:
B - big buffers. Kermit1100 normally tells the remote Kermit to send
packets that will fit in 84 characters. B-option causes it to
request the maximum size Kermit packets (which ain't as big as you
might wish) Make sure that your communications hardware and
software will let the long packets get through.
C - assume for sending or receiving that records are to be separated
by CR instead of CR-LF
L - log in the element KERMITLOG.MDSSS all file reads and writes and
all communication sends and receives. MDSSS is the month, day and
seconds/4 encoded base 32 (0,...,9,A,...,V). If a catalogued file
'KERMITLOG' is assignable, it is used. Otherwise a temporary file
is created.
Q - allow eight-bit quoting for sending or receiving. If the file
being sent or received has 8-bit data and if the remote kermit
is capable of 8-bit quoting, then all 8-bits of data can be
sent or received.
R - expect to receive data. Put the data in 1100SPEC if specified
or in the file or element name sent from the remote Kermit. No
transformation on the incoming name is done at present so it
had better be good.
S - send 1100SPEC to the remote Kermit. If REMOTESPEC is specified,
put it in the file header packet. Otherwise put 1100SPEC in the
packet.
T - test mode. Send (actually print on a terminal) packets as if
an S-option had been specified without reading ACK's.
W - If the S-option is used, wait 30 seconds before starting to send
Kermit1100 tries not to exit until an EOF is received in order to process
multiple requests from the remote Kermit.
Happy hopping,
Edgar Butt (BUTT@UMD2.ARPA)
Computer Science Center
University of Maryland
College Park, Maryland 20742
Phone (301) 454-2946
}