Network Working Group
Request for Comments: 393                         Joel M. Winett
NIC 11585                                         Lincoln Laboratory
Categories: TELNET                                LL-67
References: RFC 109, 139, 158,318, and 328        3 October 1972


                 Comments on TELNET Protocol Changes



   Through this RFC, I am registering my objection to two of the
three suggestions for changing the TELNET protocol as described in RFC
328 and am adding my suggestion for the interpretation of the TELNET
Reverse Break Control Code.


1.  Hide-your-input

   This code was originally put in the TELNET protocol to give the
   virtual terminal the ability to simulate a real terminal which has the
   print suppress capability. If the terminals being used at some
   installations do not have the ability to disable the printing
   mechanism, the TELNET being used can either ignore this code or
   attempt to simulate the function using other means (e. g., blacking
   out a number of character positions and returning to the first
   character position).  Every attempt should be made to allow a network
   user of a time-sharing system to have the same facilities as a local
   user of the time-sharing system. The specification of TELNET protocol
   should not limit the function of users if a function cannot be
   supported by all users.

   The "Hide-your-input" and "Echo" TELNET control codes provide for the
   support of two functions available in some time-sharing systems.  The
   "Hide-your-input" function is really a special case of the "Echo" mode
   of operation where the server tells the user that the server will echo
   but the server does not. A separate code is used for this func- tion
   since some servers may support this function but may not support the
   full "Echo" mode of operation.


  ] This material has not been reviewed for public release and is [
  ] intended only for use with the ARPA network. It should not be [
  ] quoted or cited in any publication not related to the ARPA    [
  ] network.                                                      [






                                                               [Page 1]

   The "Hide-your-input" and "Echo" modes of operation are disabled with
   the "No-echo" control. ASCII control codes could have been chosen for
   these functions but it was decided that the NVT ASCII control codes
   should only be specified for commonly used functions.

   To indicate the number of characters for which the printing should be
   suppressed, the "Hide-your-input" TELNET control could be rede- fined
   to include a byte following the "Hide-your-input" control to indicate
   the number of characters for which the printing should be concealed.
   The "No-echo" control would still be sent so that systems with the
   print suppress feature would not have to count characters.


2.  Data Types

   The protocol should allow a server to support users with character
   codes other than ASCII, e. g., EBCDIC. The definition of an alter-
   nate character code should include the definition of the TELNET
   control codes. An EBCDIC code has been proposed in RFC # 109 and has
   been implemented on the Lincoln Laboratory 360/67. If it is desired to
   allow one to return to the network standard ASCII code, the non-ASCII
   code should contain a code to indicate return to ASCII.


3.  Reverse Break

   The code for Break is defined as a 129th ASCII data code. It is
   usually transmitted from a user's network virtual terminal to a server
   when a corresponding key (break key or attention key) is typed on the
   TELNET terminal and is interpreted by serving systems as that special
   key. Since a common function of this key is to interrupt a running
   process the server must be alerted to the fact that this code has been
   transmitted no matter when it is sent.  Thus, the TELNET SYNC (TELNET
   data mark together with a network interrupt on the TELNET send socket)
   must also be trans- mitted to cause the serving process to examine the
   received charac- ters. The ASCII control code EOT (Octal 4) could have
   been chosen for the break function but his code is not interpreted by
   all systems.  Thus, it was decided that an NVT TELNET control code
   should be used for this purpose.

   The use of the Break Code from server to user TELNET has not
   previously been defined and, thus, could be used to solve the
   following problems which occur with line at a time and half duplex
   systems. Line at a time systems do not output characters to the
   terminal a character at a time but, instead, wait until a line is ready
   for output. If a CR-LF sequence (TELNET protocol for end of line)
   is received it is interpreted as an end of line and the characters
   received are output. If characters are received which do not end



                                                               [Page 2]

   with CR-LF the user TELNET does not know whether or not other
   characters will follow which are part of the current line. Thus, the
   characters received thus far must be output, without a CR-LF (new
   line). If an end of message code were transmitted, the user TELNET
   would know whether or not other characters would be received for
   output. The user TELNET would then print characters either when
   the TELNET Break control is received or when the CR-LF newline
   sequence is received.

   If the user TELNET is being run from a half duplex terminal, the
   terminal cannot receive input and type output at the same time.  Thus,
   if output is received while the terminal is being used for input the
   TELNET program must either buffer the received characters or abort the
   input mode of operation to write out the received charac- ters. If
   characters received are written out as they are received, the terminal
   operation would be very similar to a full duplex terminal.  This mode
   of operation requires that the terminal have a reverse break
   capability to allow the input mode to be aborted by program control.

   In some systems it is only desirable to abort the input mode of
   operation when a complete line is ready for output. If a string of
   characters received does not end with an end of line code, the
   characters received will not be output until after the input line is
   entered, i. e., the mode of operation changed from input to output.
   If an end of message code were transmitted, the user TELNET could
   abort the input mode of operation even though the end of line code was
   not received.

   In systems which do not support the reverse break feature or if the
   terminal does not have this feature it is not possible to abort an
   input mode of operation in order to output received characters. In
   this case, the systems can operate in either of two modes, a) un-
   locked keyboard, or b) locked keyboard mode.

   In an unlocked keyboard system, received characters are not output
   until the user completes an input line. An input line is completed
   when the end of line code is entered. This might be a CR, a LF, or
   a NL code. After received characters, if any, are output, the input
   modes is re-entered. To receive output the user must enter an input
   line (possibly a null line). If the user is waiting for output, he must
   repeatedly enter a line until the output has been received and typed.
   Since an input line must be entered just to receive output, it is
   desirable to define an input line which does not result in anything
   being sent to the serving system. If a null line (a line consisting of
   just the end of line code) is chosen for this purpose, some other input
   line must be defined to cause a null line to be transmitted.

   In a locked keyboard system, the input mode is not immediately



                                                               [Page 3]

   re-entered after an input line is entered. It is re-entered only after
   a defined prompt is received. The prompt can be defined to be the
   reception of any character or can be defined to be a specific charac-
   ter code. If a specific code is chosen the serving site must send this
   code whenever the terminal should be put into input mode. If an end of
   message code were transmitted this code could be inter- preted to be
   the input prompt code.

    In summary, three situations have been described where an end of
message code would be desirable.

    a) To indicate when a line which does not end with CR-LF should
       be output for line at a time systems

    b) To indicate that the input mode in half duplex operation should
       be aborted so that received characters can be output

    c) As a prompt character to cause the input mode to be entered
       for locked keyboard half duplex systems

    The ASCII TELNET control code for Break (Reverse Break) could be
interpreted as an end of message code when sent from server to user.



         [ This RFC was put into machine readable form for entry ]
         [ into the online RFC archives by BBN Corp. under the   ]
         [ direction of Alex McKenzie.                      1/97 ]























                                                               [Page 4]