Network Working Group                                Tom O'Sullivan
Request for Comments:  340                           Raytheon Company
NIC          9933                                    Sudbury, Mass.
Categories:  Telnet
References:  RFC 328                                 15 May 1972


                        PROPOSED TELNET CHANGES

   The proposed change to the TELNET protocol calling for one standard
protocol and dropping the idea of minimum implementation seems
reasonable at this time.

   I suggest that both Data Types and Hide Your Input be kept for the
following reasons:

   Data Types:

The objection stating that switching out of ASCII results in an
irreversible change and loss of control can be met by requiring other
codes to provide to a return to ASCII.  Each other code may have its
own return code, however, it may not always be employed.  Other codes
are important for alphanumeric terminals that have special devices
attached.  Several potential cases can be cited:

   1.  Cal comp plotter attached to a teletype has logic permitting a
       program to turn the plotter on and off.  When operating I believe
       it uses an 8 bit code which could conflict with Telnet signals.

   2.  Numerically controlled machines, either controlled from a user
       terminal or code prepared by a HOST computer to be punched on the
       paper tape punch at a teletype way require the use of an arbitrary
       8 bit code.

   3.  Experiments controlled from alphanumeric terminal or sensor data
       collected through a cal-comp like connection may require the use
       of a full 8 bit code.

In these cases a transparent data type with a provision for a return
to ASCII mode seems desirable.











                                                                [Page 1]

Hide Your Input:

As more and more use of data base systems in the network is
considered, the need for and importance of using access keys,
passwords, etc. grows.  The fact that it is difficult to select the
length of input to be hidden is not a persuasive argument.  Potential
solutions seem to exist, e.g. the protocol could provide for accepting
length statements from the user program, data base system, operating
system, etc. and in default of this, use a default length representing
the server system expected optimum length.



       [ This RFC was put into machine readable form for entry ]
       [ into the online RFC archives by BBN Corp. under the   ]
       [ direction of Alex McKenzie.                   12/96   ]



































                                                                [Page 2]

mirror server hosted at Truenetwork, Russian Federation.