COMMENT ⊗   VALID 00007 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002					 NOTE
C00003 00003					 MAIL
C00006 00004					LINKS
C00008 00005					TELNET
C00009 00006				  PCnet and ARPAnet
C00012 00007				Other things
C00013 ENDMK
C⊗;
				 NOTE

This file is a collection of random notes for my use only.  Thus I feel
free to say what I want to.  If anybody reads this file and sees some
comment that they don't like, well, it's his/her/its fault for reading
it.
				 MAIL

I wish to make it known for the record that I am now, and will remain,
utterly opposed to hairy headers in DIALnet mail.  The headers generated
by the SAIL mailer for ARPAnet mail are about as much as I can stomach;
the mail headers generated by certain congenital losers (ie, MIT-DMS,
MIT-Multix, and CMU-10a/b/d) are totally revolting.

If at all possible, I would like to keep the mail headers down to the
type of short, consise headers that are used internally at SAIL and ITS
sites.  This may not be possible.

However, be that as it may, I will resist any attempt to repeat the
mistakes of the ARPAnet and assault DIALnet users with the 15 line
mail headers that you can get on the ARPAnet.  I think the garbage
produced (for the benefit of the CIA or something?) is not worth the
alleged gains.

This seems like a departure of mine from doing things the hairy way, but
in this case the hair is just plain dumb.

At the worse, a message should look like (and perhaps for a single
recipiant the To: field can be flushed as well):

Date: August 1, 1977  8:46 PM PDT
From: Mark Crispin at (415)-325-0263
To: Random Hacker at (321)-987-5432
Subject: Mail headers in DIALnet

This is what I consider a maximal mail header from one user to another
should look like.

------

And perhaps what could be done is:

Mark Crispin@(415)-325-0263 8/1/77 8:46PDT Re: Mail headers in DIALnet
This is what a message really should look like with all the other useless
crud removed.
				LINKS

Are network links really a win?  Maybe network sending at line-at-a-time
is more of a win, using winning line editor, etc.  Sending is also less
anti-social since the other user can do other things, but then again in a
conversation linking is a closer form of conversation.

Recently I found myself not automatically accepting links simply because
I was fed up with getting random links without asking.

Then again, it's the linker's 10 cents (or whatever the phone bill will be)
so maybe he has a right to get his link through.  The thing I wonder though,
is what if the link is refused?  Perhaps link should be a send request as
well or something.
				TELNET

How much are we willing to compromise on DIALnet for TELNETting purposes?
It isn't going to be economical to use DIALnet for TELNETting.  I feel that
optimization for FTP and mailing purposes is the way to win, and allow
TELNETting in the protocols (such as having a process assigned for a TELNET
server) is fine.

I doubt that we will be stuck to 1200 baud forever and maybe later on
TELNETting will be more economical.  Certainly I don't think the protocols
should be lobotomized for TELNET users (ie, make DIALnet merely a smart
DIAL program).
			  PCnet and ARPAnet

I would be happy to work with PCnet people, so long as I do not have to
work with either REM or Wilbur (Geoff claims that the latter is SRI's
REM, and even though I have not met Wilbur I have no reason not to believe
Geoff).  Above all I do not want to compete with PCnet.  DIALnet and PCnet
have two entirely different audiences and will cater to two different
communities of hackers.

While an interface between DIALnet and PCnet is inevitable, I am more
interested in an ARPAnet interface.  I am worried about Col. Russell though;
he might well disapprove of this (as well as disapprove of ME due to the
late lamented ITS limerick server!).  I know that BBN's commercial users
have some bit in their accounts that denies them ARPAnet access due to the
official non-commercial nature of the ARPAnet.  This would be a damn shame
if it happens to a DIALnet/ARPAnet interface, but we must (1) be prepared
and (2) try to prevent the kind of abuses that would cause such action.
This would involve co-operation with the PCnet people as well.

Question: would the commercial networks be interested in us?  (ie, TYMNET
and TELENET come to mind right away.  Of course, they are both cretins,
but there are some hacker types there that might be willing to hack up the
necessary support at their ends.

We will need a protocol assignment number for inter-net communication unless
we have one already?
			Other things

Full duplex only.  For medium sized systems such as SAIL or LOTS.

ASCII.