Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!pasteur!ucbvax!AI.AI.MIT.EDU!JBVB
From: JBVB@AI.AI.MIT.EDU ("James B. VanBokkelen")
Newsgroups: comp.protocols.tcp-ip
Subject: Re: draft ulana sw & sec spec
Message-ID: <351396.880401.JBVB@AI.AI.MIT.EDU>
Date: 1 Apr 88 05:26:53 GMT
Sender: usenet@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 34

1. Why no mention of RFCs 963 and 964, detailing the ways in which
the descriptions of IP and TCP in MIL-STD 1777 and 1778 are wrong?
The 1983 dates on the MIL-STDs lead me to believe that they haven't
been revised to correct these problems.

2. Why no mention of RFC-950, "Internet Standard Subnetting Procedure"?

This brings up another issue that has been on my mind recently:  Is anyone
working on a "Requirements for Internet Hosts" RFC, in the vein of RFC1009?
I have recently encountered a number of very literal readings of the MIL-STDs
into government procurement specifications, such that a PC is required to:

a. Support 'Page Mode' in its FTP.

b. Generate ICMP Time Exceeded messages, even though it has only one interface,
and this represents discarding an otherwise perfectly acceptable packet
(albeit one which barely got to me).

c. Reply to ICMP Information Request messages, which appear to me to be 
intended for DDN use, and not for a broadcast LAN.  I can't guess exactly
why they might want this, but BOOTP seems to be much better suited.  There
is also the issue of whether a workstation should reply to this sort of
thing at all (particularly if the request is broadcast).

d. Support the IP option "Satnet stream ID", which I am informed is not
relevant to any of the standard upper layer protocols (TCP, UDP, etc.).
I can ignore it just fine, but I don't know if this is 'support'.

I would be interested in working with anyone who is working on a host-
oriented equivalent to RFC1009.  I might even have time to start one
myself.  Of course, if I'm entirely wrong in my criticism above, I'm
not the right person...

James VanBokkelen
FTP Software Inc.