Columbia Kermit/kermit.zip/old/misc/prime/prime800.bwr

Internet Message Format | 2020-01-01 | 7KB

Date: Sat, 20 Jan 90 20:21:12 EST
From: Frank da Cruz <[e-mail]>
To: Steve Jenkins <[e-mail]>
Subject: New Prime Kermit

Some of the source lines are quite long, up to 144 characters, which makes
BITNET network distribution problematic.  Can you ask the contributor, John
Horne, if there are some rules I can apply to shorten long lines?  For
example, in "call" statements, can the arguments be broken onto multiple
lines?  Is there a generally valid line continuation mechanism?  Thanks!
- Frank

------------------------------

Date: Wed, 31 Jan 90 14:27:35 EST
From: Frank da Cruz <[e-mail]>
To: John Horne <[e-mail]>
Cc: Steve Jenkins <[e-mail]>
Subject: Prime Kermit 8.00

John,

As you may know by now, the new Prime Kermit has been installed and announced
on Info-Kermit.  But I just got a call from a guy named Matthew Sutter
somewhere in the American midwest, who says that he has added a lot of
features to version 7.57, including dialout capability.  So I just sent him a
tape containing 8.00 so that he can merge his changes in.  I don't know if you
were planning to do any more work on Prime Kermit, but if you are, you might
want to wait until I get the new version back from Matthew and install it.
I also asked Matthew if he could do something about getting the source code
lines to fit within the BITNET mail limit of 80 characters.

- Frank

P.S. Steve, I don't know if I can mail directly to John Horne, so you might
need to forward this message.  Thanks!

------------------------------

Date: Fri, 02 Feb 90 11:17:31
From: [e-mail]
Subject: Prime Kermit version 8.00
To: fdc <@NSFnet-Relay.AC.UK:[e-mail]>

>From : John Horne, Principal Computing Officer,

       Systems and Operations Support,
       Computing Service,
       General Teaching Block,
       Polytechnic South West,
       Drakes Circus,
       Plymouth. U.K.
       PL4 8AA

Phone : 0752 - 233921         Fax : 0752 - 233922

Email : C20211 @ UK.AC.POLY-SOUTH-WEST.PRIME-A    (UK.AC.PSW.PA)

    Frank,

    I received some mail from Lancaster University (U.K.) regarding
    version 8.00 of Kermit for Prime Computers.

    First of all it is possible that the version you have may have a
    corrupted line of source code. If Lancaster sent you the source via
    tape then it should be okay. Unfortunately when I sent the source code
    up to Lancaster I was unaware that the Primes couldn't receive this line
    of code (using FTP) because of the octal characters it contains! I have
    corrected the code (but Lancs haven't put it on-line yet, 2nd Feb '90).

    In your version the line of code is : (Line 33 of XFER_MODE.PLP)

         CALL ERKL$$ (K$WRIT, 'ab', 'cd', CODE);

    Where a,b,c, and d are octal characters entered in directly using
    an Editor. There values are a = :000, b = :210, c = :000, and d = :377.

    Secondly, the CONVERT command in this version of Kermit is incorrect,
    i.e. it doesn't work. It has been corrected and, if you receive this mail
    okay then, I could send you the corrected version directly if that is
    acceptable. Obviously I would send it to Lancaster as well.

    Now finally your original request regarding long line lengths. (I didn't
    realise that this was going to be sent to the States, so I didn't worry
    too much about line lengths!)

    In PLP (Prime's version of PL/1) lines are of free format. There is no
    line continuation character since lines are only delimited by a
    semicolon (';'). Arguments to subroutine calls may be spread over several
    lines. Obviously variable names may not be split across lines.

    E.g.
           CALL SUBR1 (ABC, DEF);            /* This is okay. */
           CALL SUBR1 (ABC,
                       DEF);                 /* So is this. */
           CALL SUBR1 (AB
                       C,D
                       EF);                  /* This is NOT okay. */

    Comments are begun by the character sequence '/*' and are ended by the
    sequence '*/'. They may be spread over several lines, or even embedded
    in source lines.

    E.g.     /* This is a valid comment - no semicolon needed to end it. */
             /* This is also a valid comment
                occupying two lines. */
             CALL SUBR1 (A, B, /* This is an embedded comment
                         using two lines */ C, D);

    Unfortunately literal character strings cannot be spread over several
    lines, since the compiler will include all the spaces as part of the
    string.

    E.g.          CALL SUBR1 ('ABCDEF');     /* The string here is ABCDEF */
                  CALL SUBR1 ('ABC           /* But in this case the string */
                               DEF');        /* is ABC followed by a lot of
                                                spaces then DEF */

    To get around this problem it will be neccessary to assign the character
    string to a variable initially. This can then be split over several
    lines using the concatenation operator ('||').

    E.g.          X := 'ABC' ||          /* This assignment statement */
                       'DEF';            /* uses two lines. */
                  CALL SUBR1 (X);

    I shall go through the code myself and shorten the lines to 80 characters
    or less. Since this will make things easier in the future. Again this will
    be sent to Lancaster at some future date. (It won't take long to do the
    work, but I don't want to plague Lancaster with copies of Kermit!)

    Hope this mail gets through,
         Mail me (via Lancaster if neccessary) if you have any more problems,

                                John Horne.

------------------------------

Date: THU, JUN 28 1990 15:43:15
From: [e-mail]
Subject: Prime Kermit Version 8.00 bug
Keywords: Prime Kermit

We have been using Prime Kermit Version 8.00 since March 1990 with mostly
good results.  However, we have been having occasional problems with
corrupted files particularly when using sliding windows.  I asked one
of my student assistants to look into the problem.  After some investigation,
we have made a correction to the DUPLX$ call in XFER_MODE.PLP.

We changed the line which reads:

   addr (code) -> bit16_based = duplx$ ('A000'b4);  /* Set to half duplex. */

to

   addr (code) -> bit16_based = duplx$ ('E000'b4);  /* Set to half duplex. */

The problem we were seeing was a series of extraneous linefeed characters
scattered throughout the packets being received at the PC.  These extra LF
characters were corrupting some of the received packets.  The DUPLX$ call
with the 'A000' argument causes Primos, even in half-duplex mode, to echo
all received CR characters back to the PC as LF characters.  Setting the
DUPLX$ argument to 'E000' suppresses the echo of the LF chars.  This fix
significantly improves both the reliability and performance of Kermit 8.00.

I looked back at version 7.57 and find the same bug present there!
I find it difficult to believe that ANYONE could be using Kermit reliably
with this bug present.  But perhaps this is old news since I am a new
participant on Info-Kermit.  I hope this information is helpful.


Ted Flory  [phone]  [e-mail]
Manager of the Computing Center
Bridgewater College
Bridgewater, Virginia 22812-1599
USA