Date: Wed 23 Jul 86 18:13:17-PDT
From: Bob Larson <BLARSON@USC-ECLB.ARPA>
Subject: Protocol Manual Potential Changes

When using long packets and sliding windows together, the suggestion of
reducing the packet size (long packets section) cannot be done if a later
packet has been sent.  (Unless some further protocol extention is done.)

There are also a couple of problems in the attributes system/os field:
Prime/Primos is listed both as G and M4.  Tandy (J) should probably be
subdivided, TRSDOS and Disk Extended Color Basic versions of Tandy specific
Kermit exist.  (Tandy seems to be migrating away from proprietary os's, they
supply Xenix, MS-DOS and Os9.)  The assignment of these seems to be quite
random, UNIX and Os9 have one listing each, while cpm has 4.

The delimiter may be listed in both the type and format attribute packet if 
the type is A.  It should probablby only be in the format.

Encoding should have a compress 4.0 type (bits in the range 12-16 may also 
need to be listed.)

Shouldn't the "1-@ (ascii 49)" be "1 (ascii 49)"?

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

Date: Tue 16 Sep 86 17:14:33-EDT
From: "Christopher Lent" <OC.PEDHEM@CU20B.COLUMBIA.EDU>
Subject: IBM Kermit-MS 2.29A (test) and other things

This letter has a number of mostly cosmetic bugs in the documentation,
plus a few oddball errors in a number of programs. Oh, there's a bit
of stuff on the 2.29A (test) version for the IBM.
(I have't had the change to check 2.29A on a PC-Jr yet.)

Kermit error:
      6th Edition protocol manual
      Section 11.
      "...
      - Handle port input in one process and console input in another, paral-
      lel process.   The UNIX Kermit program listed in this manual uses this
      method."
----> Reference to C-Kermit example (which has been removed) probably should
      be removed.

Kermit error2:
     Kermit commands (the K packet)
---> Shouldn't K packet be included in section 8.2.7


Kermit repeat count question:
----> What does a repeat count of 0 mean? (I assume it means to ignore the
      following character.)  I could see this being useful
      when an interfering character is known to appear in the
      data stream of the sending machine (or intermediate line equipment).
      It could be used as a way to pass normal text characters which
      intermdiate equipment eats using a quoting mechanism for the other
      (usually ignored) character.

Kermit suggestions:
      'E' error packet for logout failure.
---->     For machines where the kermit cannot currently terminate the
      login session, I think sending an E (error) packet with some
      useful message would make sense.  Something on the order of,
        "Reconnect to terminate session on machine XYZ."
      This way the local kermit would not shut down, leaving the
      user to have to restart and reconnect to terminate the remote
      host session.

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